Intelligent networked automobile quality inspection task scheduling method and system

Abstract

The invention relates to the technical field of intelligent network connection automobiles, and discloses an intelligent network connection automobile quality inspection task scheduling method and system, and the method comprises the steps: receiving and processing a quality inspection task request; constructing a quality inspection model based on the quality inspection task request; analyzing the quality inspection model by adopting a near-end optimization algorithm to obtain a state space and a decision space of the task; and performing task scheduling on the plurality of edge quality inspection servers, the state space and the decision-making space to obtain a scheduling decision-making result. According to the intelligent networked automobile quality inspection task scheduling method and system, an edge computing architecture is introduced, quality inspection task scheduling is optimized through quality inspection task modeling and a near-end strategy, and a low-delay edge collaborative quality inspection mechanism is designed, so that the generalization ability of a quality inspection model can be improved, the execution efficiency of a quality inspection task can be remarkably improved, and the quality inspection task scheduling efficiency is improved. Therefore, the whole quality inspection process is more efficient and flexible.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent networked vehicles, and in particular to a method and system for scheduling quality inspection tasks for intelligent networked vehicles. Background Art

[0002] As the future trend of the automotive industry, the quality inspection link of intelligent connected vehicles is not only the cornerstone to ensure product quality and safety, but also the key driving force to promote the transformation of the industry to intelligence and automation. However, faced with the complex challenges in the field of intelligent connected vehicle manufacturing, such as high technical barriers, great data collection challenges, diversified application scenarios, and complex and changeable deployment environments, the process of intelligent upgrading of enterprises has encountered considerable obstacles. In this context, it is particularly important and urgent to realize the intelligent upgrade of quality inspection scenarios. It can not only effectively improve the production efficiency of intelligent connected vehicles and significantly reduce labor costs, but also is an important milestone in promoting the entire automotive industry to move towards comprehensive intelligence, informatization, and automation. Among them, efficient and accurate product surface anomaly detection technology, as a core component of intelligent quality inspection, is of self-evident importance.

[0003] Product surface anomaly detection is designed to accurately identify and quickly locate various defects on the surface of vehicle parts and even the entire vehicle, including but not limited to fine scratches, stubborn stains, surface depressions, and uneven materials. These defects not only affect the appearance quality of the vehicle, but may also pose potential safety hazards and threaten the safety performance and service life of the vehicle. Therefore, the development of efficient and accurate surface anomaly detection technology is crucial to improving the overall quality level of intelligent connected vehicles.

[0004] However, in large-scale automobile production environments, existing quality inspection systems often face massive concurrent tasks, which seriously restricts the quality inspection efficiency of the production line due to long processing delays. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention proposes a method and system for scheduling quality inspection tasks for intelligent connected vehicles to solve the above-mentioned technical problems.

[0006] First, a method for scheduling quality inspection tasks for intelligent connected vehicles is provided, including:

[0007] Receive and process quality inspection task requests;

[0008] Build a quality inspection model based on quality inspection task requests;

[0009] The quality inspection model is analyzed using a proximal optimization algorithm to obtain the state space and decision space of the task;

[0010] Multiple edge quality inspection servers are scheduled with the state space and decision space to obtain scheduling decision results.

[0011] Furthermore, receiving and processing quality inspection task requests, including:

[0012] Receive quality inspection task data from the image acquisition terminal on the intelligent connected vehicle production line;

[0013] Parse quality inspection task data and extract task type, resource requirements, and latency requirements information.

[0014] Furthermore, based on the quality inspection task request, a quality inspection model is constructed, including:

[0015] Define the resource requirements, computing resource constraints, total delay constraints, and task time thresholds for quality inspection tasks;

[0016] The quality inspection task is modeled as a mixed integer nonlinear programming problem with the goal of minimizing the average delay.

[0017] Furthermore, the process of using the proximal optimization algorithm for the quality inspection task model includes:

[0018] Design and train actor and critic networks in proximal policy optimization algorithms;

[0019] The proximal policy optimization algorithm is used to optimize the quality inspection task scheduling strategy to maximize the cumulative reward and reduce the task delay.

[0020] Further, the steps of obtaining the state space and decision space include:

[0021] Through the training process of the proximal strategy optimization algorithm, the optimal scheduling strategy for quality inspection tasks under different conditions is learned;

[0022] Based on the learned optimal scheduling strategy, the state space and decision space of the quality inspection task are clarified.

[0023] Further, the steps of coordinating multiple edge quality inspection servers include:

[0024] Realize information sharing and communication between multiple edge quality inspection servers;

[0025] Based on the state space and decision space, collaboratively make scheduling decisions for quality inspection tasks.

[0026] Secondly, it provides a smart connected vehicle quality inspection task scheduling system, including:

[0027] Task request receiving module, used to obtain quality inspection task requests;

[0028] A quality inspection model building module is used to build a quality inspection task model according to the quality inspection task request;

[0029] The optimization algorithm module is used to apply the proximal optimization algorithm to the quality inspection task model to obtain the state space and decision space;

[0030] The collaborative scheduling module is used to coordinate multiple edge quality inspection servers with the state space and decision space to obtain scheduling decision results.

[0031] Furthermore, the task request receiving module also includes:

[0032] A data receiving unit, used for receiving quality inspection task data from an image acquisition terminal;

[0033] The information parsing unit is used to parse the quality inspection task data and extract key information.

[0034] Furthermore, the quality inspection model building module also includes:

[0035] The constraint definition unit is used to define the resource requirements, computing resource constraints, total delay constraints and task time thresholds of quality inspection tasks;

[0036] The modeling unit is used to model the quality inspection task as a mixed integer nonlinear programming problem.

[0037] Furthermore, the collaborative scheduling module also includes:

[0038] An information sharing and communication unit, used to realize information sharing and communication between multiple edge quality inspection servers;

[0039] The scheduling decision unit is used to collaboratively make scheduling decisions for quality inspection tasks based on the state space and decision space.

[0040] The invention adopting the above technical solution has the following advantages:

[0041] The present invention introduces an edge computing architecture, and designs a low-latency edge collaborative quality inspection mechanism through quality inspection task modeling and proximal strategy optimization of quality inspection task scheduling. This mechanism can not only improve the generalization ability of the quality inspection model, but also significantly improve the execution efficiency of quality inspection tasks, thereby making the entire quality inspection process more efficient and flexible. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the specific implementation of the present invention, the following will briefly introduce the drawings required for use in the specific implementation. In all the drawings, each element or part is not necessarily drawn according to the actual scale.

[0043] Figure 1 This is a flow chart of the method for scheduling quality inspection tasks for intelligent connected vehicles of the present invention;

[0044] Figure 2 This is a collaborative flow chart of low-latency quality inspection tasks in the intelligent connected vehicle quality inspection task scheduling method of the present invention. DETAILED DESCRIPTION

[0045] The following embodiments of the technical solution of the present invention are described in detail in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only used as examples, and cannot be used to limit the protection scope of the present invention.

[0046] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should be the common meanings understood by technicians in the field to which the present invention belongs. The terms "first", "second", etc. in the specification and claims of the embodiments of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so as to implement the embodiments of the present disclosure described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. Unless otherwise specified, the term "multiple" means two or more. In the embodiments of the present disclosure, the character " / " indicates that the objects before and after are in an "or" relationship. For example, A / B means: A or B. The term "and / or" is a description of the association relationship of objects, indicating that there can be three relationships. For example, A and / or B means: A or B, or, A and B. The term "corresponding" can refer to an association relationship or a binding relationship, and A and B correspondingly refer to an association relationship or a binding relationship between A and B.

[0047] like Figure 1-2 As shown, the intelligent networked vehicle quality inspection task scheduling method of the present invention includes:

[0048] Step S01, receiving and processing quality inspection task requests;

[0049] Step S02: construct a quality inspection model based on the quality inspection task request;

[0050] Step S03: Analyze the quality inspection model using a proximal optimization algorithm to obtain the state space and decision space of the task;

[0051] Step S04: perform task scheduling on multiple edge quality inspection servers, state spaces, and decision spaces to obtain scheduling decision results.

[0052] Specifically, the present invention introduces an edge computing architecture, and designs a low-latency edge collaborative quality inspection mechanism through quality inspection task modeling and proximal strategy optimization of quality inspection task scheduling. This mechanism can not only improve the generalization ability of the quality inspection model, but also significantly improve the execution efficiency of quality inspection tasks, thereby making the entire quality inspection process more efficient and flexible.

[0053] In some embodiments, receiving and processing a quality inspection task request includes:

[0054] Receive quality inspection task data from the image acquisition terminal on the intelligent connected vehicle production line. The data includes product images, task types (such as size defects), resource requirements (such as CPU, memory), and latency requirements;

[0055] Parse quality inspection task data and extract task type, resource requirements, and latency requirements information.

[0056] In some embodiments, based on the quality inspection task request, a quality inspection model is constructed, including:

[0057] Define the resource requirements, computing resource constraints (such as the computing power of the edge server), total latency constraints (the time from task request to result feedback), and task time thresholds for quality inspection tasks;

[0058] The quality inspection task is modeled as a mixed integer nonlinear programming problem with the goal of minimizing the average delay.

[0059] Specifically, the quality inspection task modeling method:

[0060] Step 1: After collecting data, the production line image acquisition terminal set transmits it to the edge quality inspection server set to collaboratively perform quality inspection tasks.

[0061] Step 2: The image data of the quality inspection task arrives randomly at the edge quality inspection server. The resource requirements of each quality inspection task are known, and the latency optimization of quality inspection task scheduling is performed.

[0062] Step 3: The optimization objective function is to minimize the average delay.

[0063] Step 4: Limit the time product image data is offloaded to the edge.

[0064] Step 5: Determine whether the total calculation time of the quality inspection task is less than the preset time threshold.

[0065] Quality inspection task modeling: Assume that V = {V1, V2, ..., V n} represents the set of image quality inspection terminals for automobile production lines; E = {E1, E2, ..., E m} represents a set of edge servers that collaboratively perform data quality inspection tasks. In time slot t∈T, the image data of the automotive industry production visual quality inspection task arrives randomly, and the resource requirement of each quality inspection task is known, which is α={α1,α2,α3,......,α n}, where α n is the resource required for the nth quality inspection task. The delay-optimized quality inspection task scheduling problem can be expressed as:

[0066]

[0067]

[0068] Among them, A is the resource constraint, C is the computing resource constraint, and T is the total delay constraint.

[0069] The problem objective is to minimize the average delay;

[0070] It indicates that the maximum time for product image data to be unloaded to the edge at a given time is θ;

[0071] Indicates that the total calculation time of the quality inspection task should not exceed the time threshold τ of task α α ;

[0072] Represents the available bandwidth. The above optimization objective is a mixed integer nonlinear programming, which is an NP-hard problem.

[0073] In some embodiments, the process of applying the proximal optimization algorithm to the quality inspection task model includes:

[0074] Design and train actor and critic networks in proximal policy optimization algorithms;

[0075] The proximal policy optimization algorithm is used to optimize the quality inspection task scheduling strategy to maximize the cumulative reward and reduce the task delay, and to determine the state space and decision space of the quality inspection task.

[0076] In some embodiments, the step of obtaining the state space and the decision space includes:

[0077] Through the training process of the proximal strategy optimization algorithm, the optimal scheduling strategy for quality inspection tasks under different conditions is learned;

[0078] According to the learned strategy, the state space and decision space of the quality inspection task are determined.

[0079] Specifically, quality inspection task scheduling: The above problem cannot be solved directly, but it can be abstracted into a Markov decision process, and the Proximal Policy Optimization (PPO) algorithm is used to optimize the objective function. PPO consists of an actor network Actor and a critic network Critic, which jointly determine the action and state. Considering that the decision for each task is independent in the Markov decision process, low latency and high accuracy for each task will be achieved through cumulative rewards.

[0080] Assume that the quality inspection task has N time slots in total. represents the set of states and actions of the nth quality inspection system in the time slot of T, taking g n The probability is as follows.

[0081]

[0082]

[0083] The expected rewards and gradients brought by the quality inspection tasks in N time slots are:

[0084]

[0085] The actor network uses the gradient to update the policy to maximize the reward. The final objective function can be defined as:

[0086]

[0087] in, represents the state of the nth system in the tth time slot, Represents the action of the nth system in the tth time slot. Indicates in status Next, select Action The conditional probability of is affected by the policy parameter θ. Represents the state transition probability, indicating that the action is executed After that, from the state Transfer to state The probability of R s Represents the reward function, which indicates the reward obtained in the system. θ (g n ) represents the probability of the state and action sequence of the nth quality inspection system. represents the gradient of the reward function, which is used to optimize the policy to maximize the expected reward, express The gradient of the policy parameter θ is calculated by differentiating the probabilities of all actions and states in the system. It is the core of the policy gradient theorem, which represents the ratio of the gradient of probability to the probability itself. θ' (θ) is the objective function, which means updating the existing strategy θ by minimizing a certain loss under the new strategy parameters θ'. Represents the ratio of the action selection probabilities under the current strategy θ and the old strategy θ'. is the advantage function, which means that in a specific state Next, take action The advantage of the average strategy measures whether the action is better than a randomly selected action. [1-ε,1+ε] is the constraint range of the strategy update, which is The constraint is used to limit the change range between the new and old strategies, prevent the strategy from being updated too quickly and causing unstable training, and limit the fluctuation between the θ strategy and the θ' strategy. The clip function represents the ratio It is limited to a certain range to prevent excessively drastic changes during the policy update process.

[0088] In some embodiments, the step of coordinating multiple edge quality inspection servers includes:

[0089] Realize information sharing and communication between multiple edge quality inspection servers;

[0090] Based on the state space and decision space, collaboratively make scheduling decisions for quality inspection tasks.

[0091] Specifically, the quality inspection model designed above is deployed in the edge quality inspection server. Multiple edge quality inspection servers of the automobile industry production line adopt the above method to coordinate the scheduling of quality inspection tasks, thereby improving the factory's execution efficiency for massive concurrent quality inspection tasks.

[0092] The steps of the low-latency task collaborative detection process are as follows:

[0093] Step 1: After the quality inspection task request arrives, the resources required for the quality inspection task are modeled and optimized.

[0094] Step 2: In the edge collaborative quality inspection model, the resource demand model obtains the state space and decision space through proximal strategy optimization after mixed integer nonlinear programming.

[0095] Step 3: Multiple edge quality inspection servers coordinate scheduling decisions with the proximal strategy optimization results to improve the task execution rate.

[0096] In some other embodiments, a smart connected vehicle quality inspection task scheduling system is provided, including:

[0097] Task request receiving module, used to obtain quality inspection task requests;

[0098] A quality inspection model building module is used to build a quality inspection task model according to the quality inspection task request;

[0099] The optimization algorithm module is used to apply the proximal optimization algorithm to the quality inspection task model to obtain the state space and decision space;

[0100] The collaborative scheduling module is used to coordinate multiple edge quality inspection servers with the state space and decision space to obtain scheduling decision results.

[0101] In some embodiments, the task request receiving module further includes:

[0102] A data receiving unit, used for receiving quality inspection task data from an image acquisition terminal;

[0103] The information parsing unit is used to parse the quality inspection task data and extract key information.

[0104] In some embodiments, the quality inspection model building module further includes:

[0105] The constraint definition unit is used to define the resource requirements, computing resource constraints, total delay constraints and task time thresholds of quality inspection tasks;

[0106] The modeling unit is used to model the quality inspection task as a mixed integer nonlinear programming problem.

[0107] In some embodiments, the collaborative scheduling module further includes:

[0108] An information sharing and communication unit, used to realize information sharing and communication between multiple edge quality inspection servers;

[0109] The scheduling decision unit is used to collaboratively make scheduling decisions for quality inspection tasks based on the state space and decision space.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A method for scheduling quality inspection tasks for intelligent connected vehicles, characterized in that: include: Receive and process quality inspection task requests; Build a quality inspection model based on quality inspection task requests; The quality inspection model is analyzed using a proximal optimization algorithm to obtain the state space and decision space of the task; Multiple edge quality inspection servers are scheduled with the state space and decision space to obtain scheduling decision results.

2. The method for scheduling quality inspection tasks of intelligent connected vehicles according to claim 1, characterized in that: Receive and process quality inspection task requests, including: Receive quality inspection task data from the image acquisition terminal on the intelligent connected vehicle production line; Parse quality inspection task data and extract task type, resource requirements, and latency requirements information.

3. The method for scheduling quality inspection tasks of intelligent connected vehicles according to claim 1, characterized in that: Based on the quality inspection task request, build a quality inspection model, including: Define the resource requirements, computing resource constraints, total delay constraints, and task time thresholds for quality inspection tasks; The quality inspection task is modeled as a mixed integer nonlinear programming problem with the goal of minimizing the average delay.

4. The method for scheduling quality inspection tasks of intelligent connected vehicles according to claim 1, characterized in that: The process of using the proximal optimization algorithm for the quality inspection task model includes: Design and train actor and critic networks in proximal policy optimization algorithms; The proximal policy optimization algorithm is used to optimize the quality inspection task scheduling strategy to maximize the cumulative reward and reduce the task delay.

5. The method for scheduling quality inspection tasks of intelligent connected vehicles according to claim 1, characterized in that: The steps to obtain the state space and decision space include: Through the training process of the proximal strategy optimization algorithm, the optimal scheduling strategy for quality inspection tasks under different conditions is learned; Based on the learned optimal scheduling strategy, the state space and decision space of the quality inspection task are clarified.

6. The method for scheduling quality inspection tasks of intelligent connected vehicles according to claim 1, characterized in that: The steps of coordinating multiple edge quality inspection servers include: Realize information sharing and communication between multiple edge quality inspection servers; Based on the state space and decision space, collaboratively make scheduling decisions for quality inspection tasks.

7. Intelligent connected vehicle quality inspection task scheduling system, characterized by: include: Task request receiving module, used to obtain quality inspection task requests; A quality inspection model building module is used to build a quality inspection task model according to the quality inspection task request; The optimization algorithm module is used to apply the proximal optimization algorithm to the quality inspection task model to obtain the state space and decision space; The collaborative scheduling module is used to coordinate multiple edge quality inspection servers with the state space and decision space to obtain scheduling decision results.

8. The intelligent connected vehicle quality inspection task scheduling system according to claim 7, characterized in that: The task request receiving module also includes: A data receiving unit, used for receiving quality inspection task data from an image acquisition terminal; The information parsing unit is used to parse the quality inspection task data and extract key information.

9. The intelligent connected vehicle quality inspection task scheduling system according to claim 7, characterized in that: The quality inspection model building module also includes: The constraint definition unit is used to define the resource requirements, computing resource constraints, total delay constraints and task time thresholds of quality inspection tasks; The modeling unit is used to model the quality inspection task as a mixed integer nonlinear programming problem.

10. The intelligent connected vehicle quality inspection task scheduling system according to claim 7, characterized in that: The collaborative scheduling module also includes: An information sharing and communication unit, used to realize information sharing and communication between multiple edge quality inspection servers; The scheduling decision unit is used to collaboratively make scheduling decisions for quality inspection tasks based on the state space and decision space.

Images