Performance optimization and stability guarantee method for multi-calculation-layer real-time control system

Abstract

The invention discloses a performance optimization and stability guarantee method for a multi-calculation-layer real-time control system, and relates to the field of edge calculation and real-time control. The invention provides a new utility function, namely a time / state / utility function, which is used for quantizing the relationship between the control performance and the system state. On the basis, a cross-time-scale task scheduling mechanism is designed, an edge server coordinates a scheduling strategy in a large time scale, and a local controller schedules and controls tasks in real time in a small time scale. The scheduling mechanism divides tasks into stability tasks and performance optimization tasks. And predicting the task value by using a time / state / utility function, and dynamically adjusting the performance optimization task priority to optimize the control performance. By designing a hard real-time stability task model, a stability guarantee problem is converted into a schedulability problem of a stability task. On-line control performance optimization and stability guarantee of the multi-computing-layer edge control system are achieved from the aspects of end-side collaboration and real-time task scheduling.

Description

Technical Field

[0001] The present invention relates to the field of edge computing and real-time control, and in particular to a control performance optimization and stability assurance method for a multi-computing layer real-time control system. Background Art

[0002] As edge computing and wireless technologies mature, industrial automation is undergoing major changes. Traditional industrial automation relies on local controllers running on microcontrollers or programmable logic controllers, which are usually embedded in devices and wired to sensors and actuators. Due to the limitations of embedded device computing power, local controllers alone cannot achieve monitoring, coordination, and human-computer interaction functions for a large number of physical controlled objects (such as distillation towers, turbines, machine tools, industrial robots, temperature control equipment, etc.). To overcome the above limitations, industrial automation systems are developing into multi-computing layer control systems consisting of edge layers and terminal layers. The terminal layer consists of a large number of sensor terminals, local controllers, and execution terminals, which are used to control multiple subsystems, while the edge servers located in the edge layer usually have higher computing power and can play a role in monitoring and human-computer interaction. As edge control systems continue to integrate more and more control loops and data-intensive computing tasks, they face the following key challenges.

[0003] When multiple control tasks share a limited computing platform, meeting the hard real-time requirements of all control tasks greatly limits the platform's ability to accommodate a large number of control tasks. This limitation is further exacerbated by the discrepancy between real-time scheduling performance and control performance. This discrepancy stems from the intricate relationship between actual control performance, the execution of computing tasks, and the physical device state. At the same time, changes in the physical state of the controlled object (such as temperature, speed, position, current, etc.) and the state of the information subsystem describing the information processing process (such as end-to-end delay, etc.) have different effects on device control. However, analytically modeling the control performance that combines the information and physical states of the above control systems is a key challenge, even for scalar systems. Strong assumptions are required. In addition, there is an inherent conflict between stability and control performance at run time. Stability is usually guaranteed by offline design and analysis. In contrast, dynamic control performance changes at run time, reflecting the transient process of device control. Therefore, dynamic control performance should be improved through online scheduling under stability assurance. However, how to coordinate the conflicting factors in scheduling and control at multiple time scales remains a key problem.

[0004] In recent years, the control and scheduling joint design schemes for real-time control systems can be mainly divided into three categories. The first category is offline scheduling to ensure static control performance. The second category is runtime scheduling without considering the physical state. However, the execution value of the control task will change with the change of the physical state. In addition, it cannot adapt well to physical disturbances at runtime because the scheduling strategy does not consider the current physical state as a decision variable. The third category is runtime scheduling to ensure online control performance. However, all studies ignore the specific relationship between delay, physical state, and runtime and static control performance.

[0005] Therefore, the present invention is committed to providing a performance optimization and stability guarantee method for a multi-computing layer real-time control system. First, the present invention introduces a new utility function of the control task, namely the time / state / utility function, which is used to reflect the relationship between the real-time control performance and the information physical state. Subsequently, the present invention proposes a multi-computing layer cross-time scale control task scheduling mechanism based on the time / state / utility function. Among them, the edge server coordinates the scheduling strategy at a large time scale to improve the control performance, and the local controller performs online real-time scheduling of control tasks at a small time scale. The scheduling mechanism divides the control task into a stability task to ensure stability and a performance optimization task to optimize the control performance; the scheduling scheme predicts the control value based on the time / state / utility function, dynamically adjusts the priority of the performance optimization task, and realizes the online optimization of the control performance; at the same time, the system stability guarantee problem is converted into the schedulability problem of the stability task, and a hard real-time stability task model is designed to obtain stability guarantee. The present invention realizes the online control performance optimization and stability guarantee of the multi-computing layer edge control system from the aspects of end-edge collaboration and real-time computing task scheduling. Summary of the invention

[0006] In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is to achieve control performance optimization and stability guarantee of a multi-computing level edge real-time control system.

[0007] To achieve the above object, the present invention provides a method for performance optimization and stability assurance of a multi-computing layer real-time control system, comprising the following steps:

[0008] Step 1: Establish a data-driven time / state / utility function;

[0009] Step 2: Analyze the stability conditions of the control loop;

[0010] Step 3: The edge server predicts the priority of the control task on a large time scale;

[0011] Step 4: The local controller performs priority-based real-time scheduling of control tasks on a small time scale.

[0012] Furthermore, the time / state / utility function characterizes the control value and constructs a coupling model between computational delay, physical state and control performance.

[0013] Furthermore, the step 1 comprises the following steps:

[0014] Step 1.1, define the input and output variables of the time / state / utility function;

[0015] Furthermore, the input variables of the machine learning model include information of the real-time control system and physical state information, including parameters of the physical controlled object, state of the physical controlled object, time delay, etc.; the output variable is the performance of the control system.

[0016] Step 1.2: Collect the input and output data of the time / state / utility function and divide it into training set and test set;

[0017] Step 1.3: Select a machine learning model, perform model training, and test model accuracy.

[0018] Furthermore, the machine learning model includes a polynomial regression model, a support vector machine, a least squares lifting method, and a neural network.

[0019] Furthermore, the model accuracy testing method includes calculating the normalized error of the model, calculating the absolute error of the model, etc.

[0020] Furthermore, the stability of the control loop in step 2 is characterized by the schedulability of the real-time control task triggered by a stable period.

[0021] Furthermore, the step 2 comprises the following steps:

[0022] Step 2.1, find the stable period, that is, the maximum control period that ensures the stability of the control loop;

[0023] Step 2.2: Ensure the schedulability of real-time control tasks triggered with a stable period.

[0024] Furthermore, the step 3 comprises the following steps:

[0025] Step 3.1, divide the stability tasks and performance optimization tasks;

[0026] Step 3.2: predict the scheduling priority of each performance optimization task;

[0027] Step 3.3: Clarify the priority allocation of stability tasks and performance optimization tasks.

[0028] Furthermore, in step 3.2, the edge server predicts the scheduling priority of each performance optimization task according to the time / state / utility function and the physical state.

[0029] Furthermore, in step 3.3, the priority of all stability tasks is higher than the priority of performance optimization tasks. This approach can ensure the schedulability of stability tasks, that is, the stability of the control system.

[0030] Furthermore, the edge server collects the state of the physical controlled object and the execution state of the real-time control task from the local device on a large time scale, and predicts the value of the control task based on the real-time information physical state; according to the prediction result, the priority of the control task is dynamically adjusted to coordinate the global resource allocation.

[0031] Furthermore, the local controller completes real-time scheduling of control tasks online based on the priorities provided by the edge server.

[0032] The present invention belongs to the field of edge computing and real-time control, and mainly relates to a real-time scheduling mechanism for control tasks for multi-computing layer real-time edge control systems, and in particular to a real-time scheduling mechanism for control tasks based on control value with the goal of achieving control performance optimization and stability assurance.

[0033] Existing scheduling schemes adopt offline static or online scheduling strategies that do not consider the physical state, which cannot adapt to various interferences and other phenomena that the physical controlled objects are subjected to during operation, thereby causing a significant decrease in the control performance of industrial equipment. The present invention designs an edge control architecture, in which the edge server adjusts the priority of local control tasks online at each large time scale according to the state of the physical controlled object and the execution state of the real-time control task, thereby responding to the state changes of the physical device in a timely manner. The multi-computing level edge real-time control system architecture combines the large time scale coordination capability of the edge server and the fine-grained real-time scheduling capability of the local controller to dynamically adapt to interference in the operation of the physical system. (1) Coordination role of the edge server: The edge server collects the operating status of the physical system from the local device within a large time scale, and predicts the value of the control task based on the real-time information physical state. According to the prediction results, the priority of each control task is dynamically adjusted to coordinate the global resource allocation of the task. (2) Fine-grained scheduling of the local controller: The local controller completes the real-time scheduling of the control task online based on the priority guidance provided by the edge server. The present invention (1) improves the real-time performance and dynamic adaptability of the system: by adjusting the task priority online, the scheduling scheme can dynamically adapt to the state changes of the physical equipment, thereby significantly improving the system's responsiveness to physical interference; (2) optimizes resource utilization efficiency: when key physical states change (such as external interference), resources are allocated preferentially to avoid blind resource allocation, reduce performance degradation caused by resource waste, and improve overall system efficiency; (3) simplifies scheduling design under complex scenarios: the dynamic scheduling of edge servers reduces the preset requirements for various extreme situations in the offline design process, reduces the design complexity, and enhances the versatility and scalability of the system; (4) improves the stability and reliability of the system: by linking the physical state and scheduling decisions, the risk of disconnection between the scheduling scheme and the device state is reduced, ensuring that the system can still operate stably under complex interference conditions.

[0034] Existing online scheduling strategies ignore the complex coupling relationship between construction delay, physical state, and runtime and static control performance. However, there are limitations in using analytical expressions to model this complex coupling relationship. For example, current technology only supports specialized models such as vector systems or low-dimensional linear systems. The present invention adopts a data-driven approach to obtain time / state / utility functions to characterize control value, which associates computational delay, physical state, and runtime and static control performance. The present invention accurately describes the coupling relationship between computational delay, physical state, and control performance by constructing a data-driven control value prediction model, and designs an online scheduling strategy based on this. The principles are mainly the following three points: (1) Comprehensive characterization of control value: Introducing time / state / utility functions, and associating the following key factors based on a data-driven approach: ① computational delay ② physical state ③ control performance. (2) Coupling relationship modeling and optimization: The above factors are converted into quantifiable indicators through the control value function to construct their coupling relationship. In the scheduling process, the coupling model is used to realize the dynamic allocation of task priorities. (3) Data-driven prediction capability: Using real-time information physical state to dynamically predict control value, ensure that the scheduling scheme can be adjusted in time according to runtime changes. The present invention (1) accurately characterizes complex coupling relationships: using a data-driven approach, a coupling model between computational delay, physical state and control performance is constructed to achieve an accurate description of the behavior of complex dynamic systems. A unified time / state / utility function framework is provided to associate multiple performance indicators with the same optimization goal, significantly reducing the complexity of the system scheduling strategy; (2) improves the real-time and flexibility of the system: dynamically updates the utility function to ensure that the scheduling strategy can quickly adapt to changes in physical state and task delay, significantly improving the response speed and flexibility of the system; (3) reduces design and scheduling costs: the data-driven approach reduces the reliance on complex mathematical derivations in traditional model construction, reduces design costs, and improves the practicality of scheduling algorithms in complex systems; (4) improves control performance: real-time computational delay and control performance are directly introduced into scheduling decisions, avoiding the disconnection between the scheduling process and the system control objectives. The optimization of control performance can be achieved in different operating scenarios, including the balance of transient and steady-state performance.

[0035] Most existing real-time scheduling schemes emphasize that all task instances must meet the schedulability requirements, resulting in extremely pessimistic resource utilization. The present invention proposes a task scheduling method based on control value prediction, which distinguishes between stability tasks and performance optimization tasks. At the same time, different categories of tasks have different schedulability requirements and scheduling methods. The present invention breaks through the limitations of traditional scheduling methods by introducing a classification mechanism for control tasks and a differentiated scheduling strategy based on control value prediction, and effectively improves resource utilization. (1) Task classification and differentiated scheduling: ① Stability tasks: Key tasks to ensure system stability must meet strict hard real-time schedulability requirements; ② Performance optimization tasks: Used to improve the control performance of the system, use the remaining computing resources for scheduling, and the priority can be dynamically adjusted; (2) Dynamic priority adjustment: Based on the control value prediction model, dynamically evaluate the scheduling value of the performance optimization task, and adjust its priority according to the real-time situation of the system resources. Ensure high priority scheduling of stability tasks, while executing performance optimization tasks under the remaining resources; (3) Transformation of stability problems: Transform the system stability guarantee problem into the schedulability problem of stability tasks, and use the theoretical hard real-time task model to provide strict stability guarantees. The present invention (1) improves resource utilization: by dividing tasks into stability tasks and performance optimization tasks, the present invention allows the performance optimization task to moderately sacrifice its scheduling requirements when resources are tight, while ensuring system stability, thereby avoiding overly conservative resource allocation; (2) enhances the scheduling flexibility of the system: based on the control value prediction method, dynamic adjustment of different task priorities is achieved, so that the scheduling strategy can flexibly adapt to changes in the operating environment and task requirements; (3) ensures system stability: the strict schedulability of the stability task is guaranteed to ensure system stability; (4) is suitable for complex real-time system scenarios: the scheduling method of the present invention is particularly suitable for complex real-time systems with a large number of tasks and large differences in task characteristics, and shows excellent scalability and applicability in a multi-task environment.

[0036] Compared with the prior art, the present invention has the following obvious substantial features and significant advantages:

[0037] 1. Technical advantages of the present invention

[0038] The real-time scheduling strategy of control tasks based on control value prediction proposed in this paper provides an innovative solution for multi-computing-level edge real-time control systems, overcoming many limitations of traditional industrial automation systems in terms of dynamic control performance optimization and stability assurance. Specific advantages include:

[0039] 1.1 Strong dynamic adaptability: By coordinating the control task priorities on a large time scale through the edge server and combining it with the online scheduling of the local controller, it can respond to changes in the state of physical devices in real time and effectively deal with interference and dynamic demand changes during operation.

[0040] 1.2 Resource utilization optimization: The control tasks are divided into stability tasks and performance optimization tasks in a fine-grained manner, thereby achieving hierarchical and efficient utilization of computing resources, overcoming the resource waste problem caused by hard real-time constraints in traditional scheduling schemes.

[0041] 1.3 Theoretical stability guarantee: By converting the stability guarantee problem into the schedulability problem of stability tasks and designing a hard real-time stability task model, a strong guarantee of system stability is provided in theory.

[0042] 1.4 Data-driven control performance optimization: Use data-driven methods to build a control value prediction model, and dynamically adjust the priority of performance optimization tasks online, thereby significantly improving real-time control performance.

[0043] 2. Performance indicators of the present invention

[0044] According to current technical evaluation and experimental results, the present invention shows significant advantages in the following key performance indicators:

[0045] 2.1 Real-time guarantee of stability tasks: In complex industrial environments, since stability tasks need to meet hard real-time constraints, the completion rate of stability tasks always remains 100%, and the stability of the system is successfully guaranteed accordingly.

[0046] 2.2 Dynamic control performance: After optimization based on control value prediction, the performance indicators of the control system, such as the mean absolute error, are improved by more than 90% compared with the scheduling results based on the rate monotonic scheduling strategy.

[0047] 3. The present invention in production implementation

[0048] The technical solution of the present invention has good industrial adaptability and can be easily combined with the hardware and software architecture of existing industrial automation systems.

[0049] 3.1 Hardware Implementation:

[0050] 3.1.1 The hardware platforms required for edge servers (such as industrial-grade servers and edge computing nodes) have been widely used in industrial scenarios and there is no need to make major modifications to existing equipment.

[0051] 3.1.2 The local controller is compatible with a variety of embedded computing platforms, and upgrade deployment can be completed by simply updating the software algorithm.

[0052] 3.2 Software Integration:

[0053] 3.2.1 The scheduling algorithm of the present invention supports standard industrial wired communication protocols (such as EtherCAT, Profinet, etc.), which facilitates rapid integration into existing control systems.

[0054] 3.2.2 The data-driven control value prediction model can be trained and deployed using the existing industrial big data platform, significantly reducing the cost of model development.

[0055] 4. Industrial application prospects of the present invention

[0056] 4.1 Industrial Automation Scenario:

[0057] The present invention has broad application prospects in scenarios such as intelligent manufacturing, industrial robots, and process control. In particular, it can significantly improve the flexibility and intelligence of the system in complex industrial environments that require dynamic optimization of control performance.

[0058] 4.2 Smart Grid and Energy Management:

[0059] The real-time scheduling characteristics of multi-computing-level edge control systems are suitable for distributed energy management in smart grids, optimizing resource scheduling and ensuring system stability.

[0060] 4.3 Transportation and Autonomous Driving:

[0061] In transportation systems with high requirements for multi-task concurrency and dynamic scheduling, the technical solution of the present invention can improve real-time performance and stability, and provide technical support for the fields of smart transportation and Internet of Vehicles.

[0062] The concept, specific structure and technical effects of the present invention will be further described below in conjunction with the accompanying drawings to fully understand the purpose, characteristics and effects of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a dual computing layer architecture diagram of an edge control system of a preferred embodiment of the present invention;

[0064] Figure 2 It is a machine learning model of time / state / utility function of a preferred embodiment of the present invention;

[0065] Figure 3 It is a classification and release mode of stability and performance optimization tasks of a preferred embodiment of the present invention;

[0066] Figure 4 It is a control task scheduling mechanism based on control value prediction in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0067] The following describes several preferred embodiments of the present invention with reference to the drawings in the specification, so that the technical content is clearer and easier to understand. The present invention can be embodied in many different forms of embodiments, and the protection scope of the present invention is not limited to the embodiments mentioned in the text.

[0068] In the drawings, components with the same structure are indicated by the same numerical reference numerals, and components with similar structures or functions are indicated by similar numerical reference numerals. The size and thickness of each component shown in the drawings are arbitrarily shown, and the present invention does not limit the size and thickness of each component. In order to make the illustration clearer, the thickness of the components is appropriately exaggerated in some places in the drawings.

[0069] The present invention provides a real-time task scheduling strategy based on control value prediction for a multi-computing layer real-time edge control system, which can achieve online optimization of control performance and stability assurance.

[0070] like Figure 1 As shown, the technical solution adopted by the present invention to solve this technical problem is: a real-time scheduling mechanism for control tasks based on time / state / utility functions with the goal of achieving control performance optimization and stability guarantee, comprising the following steps:

[0071] Step 1: Establish a data-driven time / state / utility function;

[0072] Considering the control system model shown in formula (1):

[0073]

[0074] Where i∈{1,2,...,M} is the control loop index, k is the control period T i The index of x i (k) is the physical state vector, u i (k) is the control command vector, θ i is a configurable model parameter vector, f i (·) and g i (·) are the equipment dynamics model and control law respectively. The physical state of the equipment x i (k) is collected by the sensor and is recorded at a period T i Sent to the local controller. The local controller sends i Internal basis x i (k) and the scheduling priority determined by the edge server, and generates and schedules the execution command u of the control device for the executor i (k). M control tasks share the local processor, each control task τ i With implicit deadline D i =T i .

[0075] Offline training of machine learning models to obtain time / state / utility functions Then obtain V i Relationship with cyber-physical state:

[0076]

[0077] 1.1) Define the input and output variables of the time / state / utility function;

[0078] The input variable of the time / state / utility function is s i , and x i , where the static characteristic s i It remains unchanged during the entire operation of the real-time control system, including the control period T i , model parameters θ i and control strategy g i (·), etc.; the information-physical state includes end-to-end delay and the physical state x i Etc. Output variable V i Represents the prediction of control performance (such as mean absolute error, maximum absolute error, and adjustment time).

[0079] 1.2) Collect the input and output data of the time / state / utility function and divide it into training set and test set;

[0080] By adjusting the input value (end-to-end delay) within the feasible domain within the physical state Physical state x i ) to simulate the control system. The format of the collected data is where h i For prediction in the time domain, its setting depends on the time constant of the control system. After completing the data collection, the training data set and the test data set are divided according to a certain ratio.

[0081] 1.3) Select a suitable machine learning model, conduct model training and model accuracy test.

[0082] Use machine learning models such as polynomial regression model, support vector machine (SVM), least squares boosting (LSBoost), or neural network (NN) to train the model on the training data set, such as Figure 2 The trained model is tested for accuracy on the test data set, such as calculating the normalized error or absolute error of the model.

[0083] Step 2: Analyze the stability conditions of all control loops;

[0084] 2.1) Find the stable period, that is, the maximum control period that ensures the stability of the control loop;

[0085] The stability of the control loop can be expressed as the maximum control period that ensures the stability of the control loop i, that is, the stability period SPi To characterize. Stable period SP i is the control period T i multiples of, indicating that when faced with at most n consecutive i When the deadline is missed, the system can still be guaranteed to be stable.

[0086] Calculate the stable period SP i :The stability condition of the linear time-invariant system studied by the weak hard task model is used to obtain the control task τ i The maximum number of consecutive deadlines missed while ensuring stability is n i , stable period SP i Calculated by formula (3):

[0087] SP i =(n i +1)T i (3)

[0088] Where n i Satisfying the joint spectrum radius ρ(∑ i )<1, For the control task τ i The closed-loop characteristic matrix of loop i when the deadline constraint is met, A o For the control task τ i The closed-loop feature matrix of loop i when the deadline is missed.

[0089] 2.2) Ensure the schedulability of real-time control tasks triggered with a stable period.

[0090] The schedulability of stability tasks can be tested by the analysis algorithm corresponding to the scheduling strategy. For example, if the deadline monotonic algorithm is used, the response time analysis can be used to verify whether the worst response time exceeds the deadline; if the earliest deadline first algorithm is used, the processor demand analysis can be used to verify whether the total execution time requirement of the task in any period exceeds the processor available time.

[0091] Step 3: The edge server predicts the priority of the control task on a large time scale;

[0092] 3.1) Divide the stability tasks and performance optimization tasks;

[0093] Task τ i Each stable period of SP i The first task in is designated as the stability task, and the jth stability task during the entire run is represented as Stability tasks The period is SP i , the constraint deadline is T i τi Except for stability tasks The remaining tasks are defined as performance optimization tasks, expressed as The task release modes of the two are as follows Figure 3 shown.

[0094] 3.2) Predict the scheduling priority of each performance optimization task;

[0095] In each runtime coordination cycle, the edge server predicts the scheduling priority (control value) of each performance optimization task based on the time / state / utility function and the runtime information physical state runtime.

[0096] 3.2.1) The edge server performs a coordination operation in each coordination period T c The starting time of the cycle [k'T c ,(k'+1)T c ] within the control value density δ i (k'):

[0097]

[0098] in is the time / state / utility function, C i is the control task τ i The worst-case execution time of , k' is the time index of the large time scale priority prediction. Δ is the end-to-end delay interval used to control the value density prediction. i and Δ are coupled, in order to decouple scheduling and value density prediction to obtain δ i , the present invention applies the same end-to-end delay interval Δ to all control loops to achieve fairness. Thus, δ i Represents the physical state x at a given runtime i (k' ) Next execution Sensitivity to computational delays.

[0099] 3.2.2) According to the value density δ i The order of (k') sets the priority set of M control loops P(k') = [p1(k'), p2(k'), ..., p M (k')].

[0100] 3.3) Clarify the priority of stability tasks and performance optimization tasks, such as Figure 4 shown.

[0101] 3.3.1) Assign the highest priority range [1,M] to M stability tasks

[0102] Assign the M highest priorities [1,M] to M tasks τ i The stability task The allocation is not limited to any priority allocation strategy, such as applying deadline rate monotonicity or earliest deadline first.

[0103] 3.3.2) Assign the priority range [M+1,2M] to M performance optimization tasks

[0104] For performance optimization tasks Map the priority set P(k') of the performance optimization task obtained in step 3.2.2) into the range [M+1,2M].

[0105] 3.3.3) The priorities of stability tasks and performance optimization tasks are transmitted to the local controller via the edge wired network.

[0106] Step 4: The local controller performs priority-based real-time scheduling of control tasks at a small time scale.

[0107] The local controller executes a priority-based control task scheduling strategy, and the priority is transmitted to the local controller by the edge server at the beginning of each large time scale.

[0108] The local controller uses the P i (k') in the next coordination period T c Priority remains the same as the previous scheduling priority P until an update is received. i (k'-1) is the same.

[0109] The preferred specific embodiments of the present invention are described in detail above. It should be understood that ordinary technicians in the field can make many modifications and changes based on the concept of the present invention without creative work. Therefore, all technical solutions that can be obtained by technicians in the technical field based on the concept of the present invention through logical analysis, reasoning or limited experiments on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A performance optimization and stability assurance method for a multi-computing layer real-time control system, characterized in that: The following steps are involved: Step 1: Establish a data-driven time / state / utility function; Step 2: Analyze the stability conditions of the control loop; Step 3: The edge server predicts the priority of the control task on a large time scale; Step 4: Based on the priority obtained in step 3, the local controller performs real-time scheduling of control tasks on a small time scale.

2. The performance optimization and stability guarantee method of a multi-computing layer real-time control system according to claim 1, characterized in that: The time / state / utility function characterizes the control value and constructs a coupling model between computational delay, physical state and control performance.

3. The performance optimization and stability guarantee method of a multi-computing layer real-time control system as claimed in claim 1, characterized in that: The step 1 comprises the following steps: Step 1.1, define the input and output variables of the time / state / utility function; Step 1.2: Collect the input and output data of the time / state / utility function and divide it into training set and test set; Step 1.3: Select a machine learning model, perform model training, and test model accuracy.

4. The performance optimization and stability assurance method of a multi-computing layer real-time control system as described in claim 3 is characterized in that, in step 1.1, the input variables of the machine learning model include information and physical states of the real-time control system, wherein the physical states include parameters of the physical controlled object and the states of the physical controlled object, and the information states include time delays; and the output variables are the performance of the control system.

5. The performance optimization and stability guarantee method of a multi-computing layer real-time control system as claimed in claim 1, characterized in that: The step 2, the stability of the control loop, is characterized by the schedulability of the real-time control task triggered by a stable period.

6. The performance optimization and stability guarantee method of a multi-computing layer real-time control system according to claim 1, characterized in that: The step 2 comprises the following steps: Step 2.1, find the stable period, that is, the maximum control period that ensures the stability of the control loop; Step 2.2: Analyze the schedulability of real-time control tasks triggered by a stable period.

7. The performance optimization and stability guarantee method of a multi-computing layer real-time control system according to claim 1, characterized in that: The step 3 comprises the following steps: Step 3.1, divide the stability tasks and performance optimization tasks; Step 3.2: predict the scheduling priority of each performance optimization task; Step 3.3: Clarify the priority allocation of stability tasks and performance optimization tasks.

8. The performance optimization and stability guarantee method of a multi-computing layer real-time control system according to claim 7, wherein step 3.2 is characterized in that: The edge server predicts the scheduling priority of each performance optimization task based on the time / state / utility function and the physical state.

9. The performance optimization and stability guarantee method of a multi-computing layer real-time control system according to claim 7, wherein step 3.3 is characterized in that: All stability tasks take priority over performance optimization tasks.

10. The performance optimization and stability guarantee method of a multi-computing layer real-time control system according to claim 1, characterized in that: The edge server collects the state of the physical controlled object and the execution state of the real-time control task from the local device on a large time scale, and predicts the value of the control task according to the time / state / utility function based on the real-time information physical state; according to the prediction result, the priority of the control task is dynamically adjusted to coordinate the real-time scheduling of the computing tasks on the local controller.

Images