Production element collaborative awareness service combination optimization method under industrial internet

Abstract

The invention relates to a production element collaborative awareness service combination optimization method under an industrial internet, and the method comprises the steps: obtaining a production process related to a user demand, and decomposing the production process into a plurality of subtasks; based on the historical cooperation relationship and the current cooperation relationship, calculating internal and external cooperation coefficients of the task and the current cooperation relationship, and obtaining time consumption and cost consumption of the sub-task; time consumption and cost consumption are aggregated, and an optimal service combination scheme is obtained through iterative analysis by means of a teaching optimization model. According to the method, the historical cooperation relationship and the current cooperation relationship are introduced, the internal cooperation coefficient and the external cooperation coefficient of the production element service are combined, and the cooperation effectiveness of the service is comprehensively evaluated. Meanwhile, the cooperation effectiveness index is combined with a multi-objective optimization model based on a teaching optimization algorithm (TLBO) so as to screen an optimal service combination.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial Internet, and particularly to an optimization method for collaborative perception service combination of production factors under industrial Internet. Background Art

[0002] With the development of a new generation of information technologies represented by cloud computing, big data, and artificial intelligence, the real economy and the virtual economy penetrate and integrate with each other. The development concepts, production tools, and production methods of various industries are gradually changing, bringing about another leap in productivity. The industrial Internet emerges as the times require, integrating information technology, communication technology, and operation technology, aiming to connect people, data, and machines with an open and global communication network platform, sharing various production factor resources in the whole industrial production process, making production digital, networked, automated, and intelligent, so as to improve the efficiency and reduce the cost of industrial production. The parts, personnel, equipment, vehicles, etc. required in the production process are the industrial production factors and resources concerned in the business process.

[0003] At the same time, a large number of industrial Internet platforms have emerged, such as GE Predix, ABB Capability, Siemens MindSphere, PTC ThingWorx, etc., enabling various production factor resources required for production to be packaged into services and published on the platform for sharing, collaboration, and combination. In such a situation, manufacturing enterprises or users may not possess some of the factors and resources required for industrial production, but can complete more personalized and complex production tasks through the way of service invocation. This enables manufacturing enterprises or users to use various industrial manufacturing services as conveniently as using water, electricity, and gas. In such a situation, different production factor resources can be regarded as services, which are called production factor services. Similarly, the combination of services is also called production factor service combination.

[0004] In the industrial Internet environment, a complex production task may consist of multiple subtasks, and completing a single subtask may require multiple production factor resource services. At the same time, production factor resource services have spatio-temporal attributes. Traditional production factor management mainly focuses on the static management of production factors or production resources, and pays less attention to the dynamic physical environment when scheduling production factors and resources, such as the geographical location of resources, the timeliness of completing production tasks, etc. Therefore, compared with the traditional service combination in the industrial Internet environment, the service combination in actual industrial applications is more complex and challenging.

[0005] There is no research work on the optimization of the combination of production factor services yet. At the same time, most of the existing research does not consider the collaboration between production factor services, only considers service cost and service time, and then uses multi-objective optimization algorithms to obtain the optimal solution of the production factor service combination. Undoubtedly, the collaboration between production factor services will affect the overall service quality. The higher the degree of collaboration between production factor services, the higher the fluency of the business process, and the corresponding communication costs, etc. will also be reduced, thus effectively avoiding delays and other situations.

[0006] With the development of the industrial Internet and industrial Internet platforms, how to optimize the combination of production factor services has become an urgent problem to be solved. At the same time, how to quantify the degree of collaboration between production factor services and incorporate it into the calculation of the service quality of the final production factor service combination is also a question worthy of consideration. Studying the impact of the collaboration degree on the production process and service quality is of great significance for optimizing the combination of production factor services. Summary of the Invention

[0007] In order to solve the problems existing in the above-mentioned prior art, the purpose of the present invention is to provide an optimization method for collaborative perception service combination of production factors under the industrial Internet to improve the performance of the service combination.

[0008] To achieve the above object, the present invention provides the following solutions:

[0009] An optimization method for collaborative perception service combination of production factors under the industrial Internet, comprising:

[0010] Obtain the production process involved in the user requirements, and decompose the production process into multiple subtasks;

[0011] Based on the historical collaboration relationship and the current collaboration relationship, calculate the internal and external collaboration coefficients of the task and the current collaboration relationship respectively, and obtain the time consumption and cost consumption of the subtask;

[0012] Aggregate the time consumption and the cost consumption, and use the teaching optimization model to iteratively analyze the optimal service combination plan.

[0013] Optionally, calculating the internal and external collaboration coefficients of the task includes:

[0014] Use the average collaboration frequency between internal services of the task to obtain the internal collaboration coefficient:

[0015]

[0016] Wherein, IC(t k ) is the internal collaboration degree of the task, freq(s i , s j ) is s i and sj In task t k The historical call frequency, Is the combination number of selecting two services from |S k | services, s i And s j Are two different tasks t k And t j The candidate services of, s p Is the candidate service set One of the candidate services in, s q Is the candidate service set One of the candidate services in;

[0017] The average collaboration frequency of services between tasks, obtain the external collaboration coefficient:

[0018]

[0019] Among them, EC(t k ) is the external collaboration degree of the task, t j And t k Are tasks, Is the number of tasks directly following task t k , |S k | Is the number of services required to complete task t k , |S j | Is the number of services required to complete task t j , freq(s i , s j ) is the historical call count of s i And s j , Is the candidate service set corresponding to service s i , Is the candidate service set corresponding to service s j .

[0020] Optionally, calculating the current collaboration relationship includes:

[0021] Based on the number of services from the same service provider in the current service combination, determine the current collaboration coefficient:

[0022]

[0023] Among them, P i Is the service provider, PC(P i ) is the current collaboration parameter corresponding to service provider P i , d max And α are adjustment parameters, 1 - d max Represents the lowest discount allowed, Num(Pi ) is the number of services selected from service provider P in the current service composition. i in the current service composition.

[0024] Optionally, the cost consumption for obtaining the subtask includes:

[0025]

[0026] where Cost(t k ) is the cost required to complete task t after considering the collaboration effectiveness of the current service, cost(s k ) is the cost consumption of the initial service s without considering the current collaboration, S i ) is the number of candidate services required to complete task t i , k is the provider of service s k , is the provider of service s i , is the service provider corresponding to the current collaboration parameter.

[0027] Optionally, aggregating the time consumption and the cost consumption includes:

[0028] Performing a QoS aggregation operation on the time consumption and the cost consumption according to the execution structure in the subtask to obtain the comprehensive QoS value of the composite service:

[0029]

[0030] where F(sc) is the comprehensive QoS value of the composite service, sc is a service composition scheme, and are respectively the maximum time and the minimum time required in all service composition schemes, and are respectively the highest cost and the lowest cost required in all service composition schemes, and γ is the weight parameter in F(sc) for determining the weights of different QoS attributes.

[0031] Optionally, the execution structure includes: sequential time and cost, parallel time and cost, selective time and cost, and loop time and cost.

[0032] Optionally, iteratively analyzing the optimal service composition scheme using a teaching optimization algorithm includes:

[0033] According to the aggregation result, establishing an objective function for service composition optimization, and iteratively analyzing the optimal service composition scheme using a teaching optimization model.

[0034] Optionally, establishing the objective function for optimizing the service composition includes:

[0035] F = min(Time(sc), Cost(sc))

[0036]

[0037] where Cost(sc) is the cost consumption of the service composition based on collaborative perception, Time(sc) is the time consumption of the service composition, t0 is the maximum time limit of the task, c0 is the maximum cost limit of the task, and F is the objective function used to quantify the quality of the service composition scheme.

[0038] Optionally, using the teaching optimization model to iteratively analyze the optimal service composition scheme includes:

[0039] Step 1: Generate an initial service composition scheme for each subtask and calculate the initial objective function value;

[0040] Step 2: Take the service composition scheme as a student, and use the teacher module of the teaching optimization model to iteratively update each student and calculate the objective function value of the updated service composition scheme;

[0041] The expression for using the teacher module to iteratively update each student is:

[0042]

[0043] difference = r i *(X teacer - TF i * Mean)

[0044] where and respectively represent the pre-learning value and the post-learning value of the i-th student under the guidance of the teacher, r i is the learning step of the i-th student, TF i is the learning factor, Mean is the average of all students, and X teacer is the individual with the optimal fitness value in the current population;

[0045] Step 3: If the objective function value of the updated service composition scheme is better, accept the update; otherwise, retain the original value;

[0046] Step 4: Randomly select students for communication based on the student module of the teaching optimization model and perform updates, and calculate the objective function value of each updated student;

[0047] The expression for randomly selecting students for communication and performing updates using the student module is:

[0048]

[0049] Among them, and are respectively the pre - learning value and the post - learning value of the \(i\) - th student under the teacher's guidance, and \(r\) i is the learning step of the \(i\) - th student, \(X\) i and \(X\) j represent the \(i\) - th and \(j\) - th students respectively, and \(F(X\) i ) and \(F(X\) j ) represent the fitness values of the \(i\) - th and \(j\) - th students respectively;

[0050] Step 5: If the objective function value of the updated student is better, then accept the update; otherwise, retain the original value.

[0051] Step 6: According to the set number of iterations or convergence conditions, repeat Steps 2 - 5. In each iteration, record the current optimal service combination plan and its objective function value. When the number of iterations is reached or the convergence conditions are met, output the optimal service combination plan.

[0052] The beneficial effects of the present invention are as follows:

[0053] By introducing historical collaboration relationships and current collaboration relationships, the present invention combines the internal collaboration coefficient and the external collaboration coefficient of production factor services to comprehensively evaluate the collaboration effectiveness of services. At the same time, it combines the collaboration effectiveness index with a multi - objective optimization model based on the Teaching - Learning - Based Optimization (TLBO) algorithm to screen the optimal service combination.

[0054] The present invention has a significant improvement in terms of time and cost optimization compared with traditional meta - heuristic algorithms (such as genetic algorithms, particle swarm optimization, etc.), and the comprehensive QoS index performs more excellently in the optimized service combination plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0056] Figure 1 is a flowchart of a method for optimizing the production factor collaborative perception service combination under an industrial Internet according to an embodiment of the present invention;

[0057] Figure 2 is a sample diagram of a ship production process according to an embodiment of the present invention;

[0058] Figure 3Comparison graph of the evolution curves of different methods of the embodiments of the present invention. Detailed implementation manners

[0059] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0060] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0061] As Figure 1 shown, this embodiment discloses an optimization method for collaborative perception service combination of production factors under industrial Internet, including: obtaining the production process involved in user requirements, and decomposing the production process into multiple subtasks; based on historical collaboration relationships and current collaboration relationships, calculating the internal and external collaboration coefficients of tasks and the current collaboration relationship respectively, and obtaining the time consumption and cost consumption of subtasks; aggregating the time consumption and cost consumption, and iteratively analyzing the optimal service combination plan by using a teaching optimization model.

[0062] Specifically, this embodiment discloses an optimization method for collaborative perception service combination of production factors under industrial Internet, including the following steps:

[0063] (1) Task decomposition: Decompose the production process involved in user requirements into multiple subtasks, and each subtask requires one or more types of production factor services; (2) QoS attribute definition: Define the QoS attributes of production factor services, including time consumption and cost consumption; (3) Time consumption calculation: Considering historical collaboration relationships, calculate the internal collaboration coefficient and external collaboration coefficient of tasks, and calculate the time consumption of tasks; (4) Cost consumption calculation: Considering the current collaboration relationship, calculate the current collaboration coefficient, and calculate the cost consumption of tasks; (5) QoS aggregation: Considering the QoS aggregation module, aggregate the time consumption and cost consumption of tasks to obtain the comprehensive QoS value of tasks. (6) Service combination optimization: Use the teaching optimization algorithm (TLBO) to derive the optimal service combination plan. Use the teaching optimization algorithm (TLBO) to achieve the global optimization of the service combination plan.

[0064] Further, in step (1), the tasks released by enterprises or users on the industrial Internet platform are decomposed. The production process can be decomposed into a triple P, including a task set, a set of execution relationships between tasks, and a set of production factor services that tasks can use.

[0065] Further, in step (2), the QoS attributes of production factor services are defined, including time consumption and cost consumption. The present invention considers three types of services, namely human resources, material resources, and machine resources. Time and cost are regarded as QoS attributes in the present invention. It should be noted that different types of services have different QoS attributes. The QoS attributes of human resources only consider the usage cost; the QoS attributes of material resources consider the usage cost, delivery cost, and delivery time; the QoS attributes of machine resources include usage cost, delivery cost, delivery time, and execution time.

[0066] Further, calculating the internal and external collaboration coefficients of tasks includes:

[0067] Obtaining the internal collaboration coefficient by using the average collaboration frequency between internal services of the task:

[0068]

[0069] where IC(t k ) is the degree of internal collaboration of the task, freq(s i , s j ) is the historical call frequency of s i and s j in the task t k , is the combination number of selecting two services from |S k | services, s i and s j are candidate services of two different tasks t k and t j respectively, s p is a candidate service in the candidate service set , s q is a candidate service in the candidate service set ;

[0070] Obtaining the external collaboration coefficient by using the average collaboration frequency of services between tasks:

[0071]

[0072] where EC(t k ) is the degree of external collaboration of the task, t j and t k are tasks, is the number of tasks directly following the task t k , |S k | is the number of services required to complete the task t k , |S j | is the number of services required to complete the task t j , freq(si , s j ) is s i and s j 's historical call count, is the candidate service set corresponding to service s i , is the candidate service set corresponding to service s j .

[0073] Specifically, the historical collaboration relationship includes intra-task collaboration and inter-task collaboration. The intra-collaboration coefficient is obtained by calculating the average collaboration frequency between intra-task services, and the inter-collaboration coefficient is obtained by calculating the average collaboration frequency between services of different tasks.

[0074] The calculation formula for the degree of intra-task collaboration includes:

[0075]

[0076] where t k is a subtask, freq(s i , s j ) is the historical call frequency of s i and s j in task t k . Only when s i and s j are called simultaneously, s i and s j are considered to have historical collaboration. is the number of combinations of choosing two services from |S k | services.

[0077] The calculation formula for the degree of inter-task collaboration includes:

[0078]

[0079] where task t j is executed after task t k is completed, that is, the start of task t j depends on the completion of task t k . is the number of tasks directly following task t k . s i and s j are candidate services for two different tasks t k and t j respectively. |S k | is the number of services required to complete task t k , and |S j | is the number of services required to complete task t j . freq(si , s j ) is s i and s j 's historical call count. For service s i the corresponding candidate service set, For service s j the corresponding candidate service set.

[0080] Furthermore, calculating the current collaboration relationship includes:

[0081] Based on the number of services from the same service provider in the current service composition, determine the current collaboration coefficient:

[0082]

[0083] where P i is the service provider, PC(P i ) is the current collaboration parameter corresponding to service provider P i , d max and α are adjustment parameters, 1 - d max represents the allowed minimum discount, Num(P i ) is the number of services selected from service provider P i in the current service composition.

[0084] Specifically, the current collaboration relationship determines the current collaboration coefficient by calculating the number of services from the same service provider in the current service composition.

[0085] The specific calculation formula for the collaboration coefficient includes:

[0086]

[0087] where P i is the service provider, PC(P i ) is the current collaboration parameter corresponding to service provider P i , whose range is PC(P i ) ∈ (d max - 0.5, d max , and d max and α are adjustment parameters, with ranges d max ∈ [0.5, 1], α ∈ (0, 1], so 1 - d max represents the allowed minimum discount. Num(P i ) is the number of services selected from service provider P i in the current service composition.

[0088] Furthermore, obtaining the cost consumption of the subtask includes:

[0089]

[0090] Among them, Cost(t k ) is the cost required to complete task t after considering the collaboration effectiveness of the current service, cost(s k ) is the cost consumption of the initial service s without considering the current collaboration, S i ) is the number of candidate services required to complete task t, i is the provider of service s, k is the current collaboration parameter corresponding to the service provider. k Specifically, the current collaboration optimizes the allocation and scheduling of resources, further reducing the service cost and thus the overall cost consumption of the task. The specific calculation formula for cost consumption includes: is the provider of service s, i is the current collaboration parameter corresponding to the service provider. is the service provider corresponding to the current collaboration parameter.

[0091] Specifically, the current collaboration optimizes the allocation and scheduling of resources, further reducing the service cost and thus the overall cost consumption of the task. The specific calculation formula for cost consumption includes:

[0092]

[0093] Among them, Cost(t k ) is the cost required to complete task t after considering the collaboration effectiveness of the current service, cost(s k ) is the cost consumption of the initial service s without considering the current collaboration, S i ) is the number of candidate services required to complete task t, i is the provider of service s, k is the current collaboration parameter corresponding to the service provider. k Therefore, considering the current collaboration between services can further reduce the cost consumption in the production process. is the provider of service s, i is the current collaboration parameter corresponding to the service provider. is the service provider corresponding to the current collaboration parameter. Therefore, considering the current collaboration between services can further reduce the cost consumption in the production process.

[0094] Further, the aggregation of time consumption and cost consumption includes:

[0095] Perform QoS aggregation operations on time consumption and cost consumption according to the execution structure in the subtask to obtain the comprehensive QoS value of the composite service:

[0096]

[0097] Among them, F(sc) is the comprehensive QoS value of the composite service, sc is a service composition scheme, and are respectively the maximum time and minimum time required in all service composition schemes, and are the highest cost and the lowest cost required in all service combination scenarios, respectively, and γ is the weight parameter in F(sc), which is used to determine the weights of different QoS attributes.

[0098] Furthermore, the execution structures include: sequential time and cost, parallel time and cost, selective time and cost, and loop time and cost. Among them, sequential means that tasks are executed sequentially in a predetermined order, and the next task starts only after the previous task is completed; parallel means that multiple tasks are executed simultaneously and independently of each other without waiting for other tasks to complete; conditional means that which task to execute is determined according to specific conditions; loop means that a task is repeatedly executed when the condition is met until the condition is no longer satisfied.

[0099] Specifically, the QoS aggregation module is also used to perform QoS aggregation operations according to the execution structure (sequential, parallel, conditional, or loop) of tasks to obtain the comprehensive QoS value of the composite service. Among them, sequential means that tasks are executed sequentially in a predetermined order, and the next task starts only after the previous task is completed; parallel means that multiple tasks are executed simultaneously and independently of each other without waiting for other tasks to complete; conditional means that which task to execute is determined according to specific conditions; loop means that a task is repeatedly executed when the condition is met until the condition is no longer satisfied.

[0100] As shown in Table 1, the specific formulas or rules of the aggregation methods include:

[0101] Table 1

[0102]

[0103] where t i is the i-th task in the production process, Cost(t i ) is the cost consumption generated by the service required to invoke t i , and Time(t i ) is the time consumption generated by the service required to invoke t i . N s refers to the number of tasks in the sequential structure, N p is the number of branches in the parallel structure, N c is the number of branches in the selective structure, p i is the probability of triggering the branch and there is loop(t i ) is the number of loops in the loop structure.

[0104] Furthermore, using the teaching optimization algorithm to iteratively analyze the optimal service combination scenario includes:

[0105] According to the aggregation result, establish the objective function for service combination optimization, and use the teaching optimization model to iteratively analyze the optimal service combination scenario.

[0106] Furthermore, the objective function for service composition optimization includes:

[0107] F = min(Time(sc), Cost(sc))

[0108]

[0109] where Cost(sc) is the cost consumption of the service composition based on collaborative awareness, Time(sc) is the time consumption of the service composition, t0 is the maximum time limit of the task, and c0 is the maximum cost limit of the task.

[0110] Specifically, the teaching learning based optimization (TLBO) in the optimization algorithm module is used to improve the comprehensive QoS of the service composition as much as possible by considering collaborative effectiveness. Among them, the attributes of the comprehensive QoS include time consumption and cost consumption. The service composition optimization model includes:

[0111] F = min(Time(sc), Cost(sc))

[0112]

[0113] where Cost(sc) is the cost consumption of the service composition based on collaborative awareness, Time(sc) is the time consumption of the service composition, t0 is the maximum time limit of the task, c0 is the maximum cost limit of the task, and F is the objective function, which is used to quantify the quality of the service composition scheme.

[0114] Furthermore, using the teaching optimization model to iteratively analyze the optimal service composition scheme includes:

[0115] Step 1: Generate an initial service composition scheme for each subtask and calculate the initial objective function value;

[0116] Step 2: Take the service composition scheme as a student, and iteratively update each student based on the teacher module of the teaching optimization model, and calculate the objective function value of the updated service composition scheme;

[0117] The expression for iteratively updating each student using the teacher module is:

[0118]

[0119] difference = r i *(X teacer -TF i *Mean)

[0120] where the i-th student before and after learning under the guidance of the teacher is respectively represented as and r i is the learning step size of the i-th student, with a range of [0, 1], TF i is the learning factor, TF i = round[1 + rand(0, 1)], and its value is between [1, 2]. Mean is the average of all students, X teacer is the individual with the optimal fitness value in the current population.

[0121] Step 3, if the objective function value of the updated service composition scheme is better, then accept the update; otherwise, retain the original value.

[0122] Step 4, randomly select students for communication based on the student module of the teaching optimization model, and perform updates. Calculate the objective function value of each updated student.

[0123] The expression for randomly selecting students for communication based on the student module and performing updates is:

[0124]

[0125] where the i-th student before and after learning under the guidance of the teacher is represented as and r i is the learning step size of the i-th student, X i and X j represent the i-th and j-th students respectively, and F(X i ) and F(X j ) represent the fitness values of the i-th and j-th students respectively.

[0126] Step 5, if the objective function value of the updated student is better, then accept the update; otherwise, retain the original value.

[0127] Step 6, repeat Steps 2 - 5 according to the set number of iterations or convergence conditions. In each iteration, record the current optimal service composition scheme and its objective function value. When the number of iterations is reached or the convergence conditions are met, output the optimal service composition scheme.

[0128] Since there are relatively few studies on the service composition of production factors in the industrial Internet environment at present, existing studies mostly draw on the ideas in the traditional service computing field, mainly focusing on selecting a service for each task or optimizing service composition through a combination optimization model and heuristic algorithm. However, these methods ignore the collaboration relationship between services, which may lead to poor overall performance of service composition. Therefore, this embodiment proposes a new method for optimizing the collaborative perception service composition of production factors under the industrial Internet, named CaSCO. First, this embodiment calculates the collaboration frequency between services by analyzing historical collaboration data, including service collaboration within a task (internal collaboration) and service collaboration between tasks (external collaboration). Subsequently, based on the number of service providers in the current service composition, the effectiveness index of the current collaboration is quantified. Second, the collaborative perception service composition optimization model is combined with the teaching learning-based optimization algorithm (TLBO), that is, the CaSCO method, to optimize the comprehensive QoS of the service composition. Finally, the optimized service composition scheme is delivered to the user to achieve double optimization of time and cost.

[0129] The overall framework of the method for optimizing the collaborative perception service composition of production factors under the industrial Internet proposed in this embodiment is as Figure 2 shown, and mainly includes the following four core parts:

[0130] (1) Task decomposition: Decompose the user's production process into multiple tasks, and each task requires one or more types of production factor services; (2) QoS attribute definition: Define the QoS attributes of production factor services, including time consumption and cost consumption; (3) Time consumption calculation: Considering the historical collaboration relationship, calculate the internal collaboration coefficient and external collaboration coefficient of the task, and calculate the time consumption of the task; (4) Cost consumption calculation: Considering the current collaboration relationship, calculate the current collaboration coefficient, and calculate the cost consumption of the task; (5) QoS aggregation: Considering the QoS aggregation module, aggregate the time consumption and cost consumption of the task to obtain the comprehensive QoS value of the task. (6) Service composition optimization: Use the teaching learning-based optimization algorithm (TLBO) to achieve global optimization of the service composition scheme.

[0131] The research motivation, model construction and experimental verification of this embodiment will be described in detail below.

[0132] To describe the service composition problem of production factors in the industrial manufacturing link in the industrial Internet environment, the present invention takes a coarse-grained ship production process in a real scenario as an example to elaborate on its production process. The ship production process is as Figure 2As shown in the figure, the ship production process includes 5 different tasks: steel processing, section production, equipment processing, berth enclosure, and ship testing. Each task requires at least one type of production factor resource, namely material resource or machine resource, and each type of production factor resource may require one or more. Task T1 requires two types of production factor resources and each type of production factor resource requires two. Task T3 requires two machine resources, while task T5 only requires one machine resource. The precedence relationship between tasks can be sequential, parallel, conditional, or cyclic. There are only two execution relationships between subtasks in this ship production process: sequential and parallel. In this process, T1, T2, T3, and T4 all need to be completed by composite services, that is, the collaboration of two or more manufacturing services, while T5 only requires one manufacturing service to complete. Based on each task (single service) or composite service, a complete service composition chain can be formed.

[0133] First, decompose the tasks published by enterprises or users on the industrial Internet platform. The production process can be decomposed into a triple P, including a set of tasks, a set of execution relationships between tasks, and a set of production factor services that tasks can use.

[0134] Then, define the QoS attributes of production factor services, including time consumption and cost consumption. The present invention considers three types of services, namely human resources, material resources, and machine resources. Time and cost are regarded as QoS attributes in the present invention. It should be noted that different types of services have different QoS attributes. The QoS attributes of human resources only consider the usage cost; the QoS attributes of material resources consider the usage cost, delivery cost, and delivery time; the QoS attributes of machine resources include usage cost, delivery cost, delivery time, and execution time.

[0135] Consider the historical collaboration relationship of tasks for calculating the time consumption of tasks. The historical collaboration relationship includes internal collaboration within tasks and external collaboration between tasks. The internal collaboration coefficient is obtained by calculating the historical average collaboration frequency between internal services of tasks, and the external collaboration coefficient is obtained by calculating the historical average collaboration frequency between services of tasks.

[0136] The calculation formula for the degree of internal collaboration of tasks includes:

[0137]

[0138] Among them, t k is a subtask, and freq(s i , s j ) is the historical call frequency of s i and s j in task t k . Only s i and s jWhen called simultaneously, s is called i and s j have historical collaboration. is the combination number of selecting two services from |S k | services.

[0139] The calculation formula for the external collaboration degree of tasks includes:

[0140]

[0141] Among them, task t j is executed after task t k is completed, that is, the start of task t j depends on the completion of task t k is completed. is the number of tasks directly following task t k , s i and s j are the candidate services of two different tasks t k and t j respectively, |S k | is the number of services required to complete task t k , |S j | is the number of services required to complete task t j , freq(s i , s j ) is the historical call count of s i and s j . is the set of candidate services corresponding to service s i , is the set of candidate services corresponding to service s j .

[0142] Next, consider the current collaboration relationship of tasks, which is used to calculate the cost consumption of tasks. The current collaboration relationship determines the current collaboration coefficient by calculating the number of services from the same service provider in the current service combination. The specific calculation formula for the collaboration coefficient includes:

[0143]

[0144] Among them, P i is the service provider, PC(P i ) is the current collaboration parameter corresponding to service provider P i , and its range is PC(P i ) ∈ (d max - 0.5, d max , and d max and α are adjustment parameters, and their ranges are d max∈[0.5, 1], α ∈ (0, 1], so 1 - d max represents the lowest allowed discount. Num(P i ) is the number of services selected from service provider P i in the current service portfolio.

[0145] Furthermore, the current collaboration optimizes the allocation and scheduling of resources, further reducing the service cost and thus the overall cost consumption of the tasks. The specific calculation formula for cost consumption includes:

[0146]

[0147] where Cost(t k ) is the cost required to complete task t k after considering the collaboration effectiveness of the current service, cost(s i ) is the cost consumption of the initial service s i without considering the current collaboration, S k is the number of candidate services required to complete task t k , is the provider of service s i . Therefore, considering the current collaboration between services can further reduce the cost consumption in the production process.

[0148] Then, the QoS aggregation module performs QoS aggregation operations according to the execution structure (sequential, parallel, conditional, or loop) of the task to obtain the comprehensive QoS value of the composite service.

[0149] Finally, the teaching - learning - based optimization algorithm (TLBO) in the optimization algorithm module is used to maximize the comprehensive QoS of the service portfolio as much as possible by considering the collaboration effectiveness. Among them, the attributes of the comprehensive QoS include time consumption and cost consumption. The service portfolio optimization model includes:

[0150] F = min(Time(sc), Cost(sc))

[0151]

[0152] where Cost(sc) is the cost consumption of the service portfolio based on collaboration awareness, Time(sc) is the time consumption of the service portfolio, t0 is the maximum time limit of the task, and c0 is the maximum cost limit of the task.

[0153] The teaching - learning - based optimization algorithm (TLBO) includes a teacher phase and a student phase. Through the iterative optimization of the teacher phase and the student phase, the optimal service portfolio scheme is derived.

[0154] The update formula for the teacher phase includes:

[0155]

[0156] difference = r i *(X teacher - Tf i *Mean)

[0157] Among them, the i-th student before and after learning under the guidance of the teacher is respectively represented as and r i is the learning step size of the i-th student, with a range of [0, 1], and TF i is the learning factor, and TF i = round[1 + rand(0, 1)], and its value is between [1, 2], and Mean is the average of all students.

[0158] The update formula in the student stage includes:

[0159]

[0160] Among them, the i-th student before and after learning under the guidance of the teacher is respectively represented as and r i is the learning step size of the i-th student, X i and X j respectively represent the i-th and j-th students, and F(X i ) and F(X j ) respectively represent the fitness values of the i-th and j-th students.

[0161] In this embodiment, for the recommendation method provided by the present invention, experimental verification is carried out. Next, the data set, experimental parameter settings, evaluation metrics, comparison methods, and experimental results of this embodiment are described in detail.

[0162] Data set and experimental settings:

[0163] Figure 2 The specific production process of the ship is given. This production process can be decomposed into five subtasks, and there are two execution modes, sequential and parallel, between the tasks, that is, only sequential and parallel cases are considered when performing QoS aggregation. In addition, based on this data set, the present invention also simulated and generated 400 historical production factor service combination data records. The specific information of this data set is shown in Table 2.

[0164] Table 2 Parameter settings of the data set

[0165]

[0166] As can be seen from Table 3, the dataset generates the QoS data of candidate services and the historical service composition data. In the generated dataset, an abstract service has 10 candidate services, and each candidate service requires a certain time consumption and cost consumption. The range of time consumption is between 5 and 20, and the range of cost consumption is between 40 and 50. Each service is provided by the corresponding provider. In the present invention, a total of 6 service providers are involved in providing services. Therefore, there are 10 - 1000 composite service candidates to complete a task. In the experiment, α = 0.1, d max = 0.5, γ = 0.5.

[0167] Evaluation metrics:

[0168] To evaluate the performance of the proposed method, the present invention selects the following metrics as the evaluation criteria: time consumption (time), cost consumption (cost), and overall QoS. Among them, time consumption (time) refers to the total time in the production process, cost consumption (cost) is the total cost in the production process, and overall QoS represents the comprehensive quality index. The time consumption and cost consumption of the production task are calculated by the historical collaboration module and the current collaboration module respectively, and then the time and cost of each subtask are aggregated by the QoS aggregation module to obtain the overall time consumption and cost consumption of the service composition scheme. For the overall QoS (i.e., fitness value) of a service composition scheme, since the present invention needs to optimize both the time and cost metrics simultaneously, a dimensionless standardization method is adopted to transform the multi-objective optimization problem into a single-objective optimization problem for easy optimization and evaluation. The calculation formula for the overall QoS of the service composition is as follows.

[0169]

[0170] Among them, sc is a service composition scheme, and are the maximum time and minimum time required in all service composition schemes respectively, and are the highest cost and lowest cost required in all service composition schemes respectively. Through dimensionless standardization, the service composition scheme can be further screened to obtain a service composition scheme with the highest service quality.

[0171] Comparison with baseline methods:

[0172] To verify the algorithm performance of CaSCO, a series of experiments are carried out. The method proposed in the present invention will be compared with the following methods: GA, PSO, TLBO, GA - CaSCO, PSO - CaSCO.

[0173] (1) GA: The Genetic Algorithm has powerful global search capabilities and can effectively avoid local optimal solutions. It is suitable for complex constraints and multi-objective optimization and can flexibly handle different requirements. Through population evolution and adaptive adjustment, the Genetic Algorithm can dynamically adapt to changes and provide high-quality service composition solutions.

[0174] (2) PSO: The Particle Swarm Optimization algorithm has strong global search capabilities and fast convergence. By simulating the group behavior of particles and local information sharing, it can effectively explore the solution space. PSO has strong adaptability to constraint conditions and is suitable for dealing with dynamically changing service composition problems.

[0175] (3) TLBO: The Teaching-Learning-Based Optimization algorithm is simple to implement and does not require gradient information of the problem. By simulating the interactive learning process of teachers and students, TLBO can effectively explore the solution space and find the global optimal solution. It has good processing ability for constraint conditions and can also quickly adapt in a dynamic environment.

[0176] (4) GA-CaSCO: The Genetic Algorithm-Collaboration aware Service Composition Optimization method takes into account the collaboration relationship between services in the service composition optimization method, can adapt to dynamic environmental changes, meet multi-objective requirements, and handle complex constraint conditions.

[0177] (5) PSO-CaSCO: Similar to GA-CaSCO, the Particle Swarm Optimization-Collaboration aware Service Composition Optimization method takes into account the collaboration relationship between services in the service composition optimization method, can effectively handle multi-objective optimization problems, and adapt to dynamic environmental changes.

[0178] Table 3 Performance Comparison with the Baseline

[0179]

[0180] Experimental Results and Analysis:

[0181] The performance evaluation is demonstrated by comparison with the baseline, and the evolution process of the collaborative sensing service composition optimization process is studied using different metaheuristic algorithms. The performance comparison with the baseline under the total time, cost, and overall QoS metrics is shown in Table 3. It can be seen that among GA, PSO, and TLBO, TLBO is slightly better than GA and PSO. Without considering collaboration, GA is the worst for service composition optimization. Similarly, among GA-CaSCO, PSO-CaSCO, and TLBO-CaSCO, TLBO-CaSCO is slightly better than GA-CaSCO and PSO-CaSCO. And GA-CaSCO is the worst in service composition optimization considering collaboration. GA-CaSCO, PSO-CaSCO, and TLBO-CaSCO are respectively better than GA, PSO, and TLBO. In summary, regardless of which metaheuristic algorithm is used to solve the proposed collaborative sensing service composition optimization model, considering the collaboration between services can effectively improve the comprehensive QoS.

[0182] To further study the effectiveness of the method, the evolution curves of the comprehensive QoS under different methods are studied to solve the collaborative sensing service composition optimization problem. Figure 3 The change of the comprehensive QoS value with the increase of the number of iterations is shown. From Figure 3 it can be seen that TLBO-CaSCO obtains the optimal solution at the 38th iteration, PSO-CaSCO obtains the optimal solution at the 86th iteration, and GA-CaSCO obtains the optimal solution at the 45th iteration. Generally speaking, TLBO-CaSCO obtains the optimal solution with the least number of iterations.

[0183] The collaborative sensing service composition optimization model of production factors under the industrial Internet proposed by the present invention is solved by using an advanced teaching optimization algorithm. In the proposed method, both historical collaboration relationships (including internal collaboration and external collaboration) and current collaboration relationships are considered. Historical collaboration optimizes the time of the target service composition, and current collaboration optimizes the cost of the target service composition. Therefore, the comprehensive QoS can be improved by considering the collaboration relationship. The experimental results on the simulation dataset show that regardless of which metaheuristic algorithm is used to solve the proposed collaborative sensing service composition optimization model, considering the collaboration between services can effectively improve the comprehensive QoS.

[0184] The embodiments described above are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. An optimization method for collaborative perception service composition of production factors under industrial Internet, characterized in that, Including: Obtain the production process involved in user requirements, and decompose the production process into multiple subtasks; Based on the historical collaboration relationship and the current collaboration relationship, calculate the internal and external collaboration coefficients of the tasks and the current collaboration relationship respectively, and obtain the time consumption and cost consumption of the subtasks; Aggregate the time consumption and the cost consumption, and use the teaching optimization model to iteratively analyze the optimal service combination plan.

2. The method for optimizing the collaborative perception service combination of production factors under the industrial Internet according to claim 1, wherein Calculating the internal and external collaboration coefficients of the task includes: Using the average collaboration frequency between internal services of the task to obtain the internal collaboration coefficient: Among them, IC(t k ) is the degree of internal collaboration of the task, freq(s i , s j ) is the historical call frequency of s i and s j in the task t k . is the number of combinations of selecting two services from |S k | services, s i and s j are candidate services for two different tasks t k and t j respectively, s p is a candidate service in the candidate service set , and s q is a candidate service in the candidate service set . The average collaboration frequency of services between tasks to obtain the external collaboration coefficient: Among them, EC(t k ) is the degree of external collaboration of the task, t j and t k are tasks, is the number of tasks directly following task t k . |S k | is the number of services required to complete task t k . |S j | is the number of services required to complete task t j . freq(s i , s j ) is the historical call count of s i and s j . is the set of candidate services corresponding to service s i . is the set of candidate services corresponding to service s j .

3. The method for optimizing the collaborative perception service combination of production factors under the industrial Internet according to claim 1, wherein, Calculating the current collaboration relationship includes: Based on the number of services from the same service provider in the current service combination, determine the current collaboration coefficient: Among them, P i is the service provider, and PC(P i ) is the current cooperation parameter corresponding to the service provider P i , d max and α are adjustment parameters, 1 - d max represents the lowest allowable discount, and Num(P i ) is the number of services selected from the service provider P i in the current service portfolio.

4. The method for optimizing the collaborative perception service combination of production factors under the industrial Internet according to claim 1, wherein Obtaining the cost consumption of the subtask includes: Among them, Cost(t k ) is the cost required to complete task t k after considering the collaboration effectiveness of the current service, cost(s i ) is the cost consumption of the initial service s i without considering the current collaboration, S k is the number of candidate services required to complete task t k , is the provider of service s i , is the current collaboration parameter corresponding to the service provider .

5. The collaborative perception service combination optimization method for production factors under the industrial Internet according to claim 1, characterized in that Aggregating the time consumption and the cost consumption includes: Perform QoS aggregation operations on the time consumption and the cost consumption according to the execution structure in the subtask to obtain the comprehensive QoS value of the combined service: Among them, F(sc) is the comprehensive QoS value of the composite service, and sc is a service composition scheme. and are the maximum time and minimum time required in all service composition schemes respectively. and are the highest cost and lowest cost required in all service composition schemes respectively. γ is the weight parameter in F(sc), which is used to determine the weights of different QoS attributes.

6. The method for optimizing the collaborative perception service combination of production factors under the industrial Internet according to claim 5, wherein The execution structure includes: sequential time and cost, parallel time and cost, selective time and cost, and cyclic time and cost.

7. The method for optimizing the collaborative perception service combination of production factors under the industrial Internet according to claim 1, wherein Iteratively analyzing the optimal service combination plan using the teaching optimization algorithm includes: According to the aggregation result, establish the objective function for optimizing the service combination, and use the teaching optimization model to iteratively analyze the optimal service combination plan.

8. The method for optimizing the collaborative perception service combination of production factors under the industrial Internet according to claim 6, characterized in that Establishing the objective function for optimizing the service combination includes: F = min(Time(sc), Cost(sc)) Where Cost(sc) is the cost consumption of the service combination based on collaboration awareness, Time(sc) is the time consumption of the service combination, t0 is the maximum time limit of the task, c0 is the maximum cost limit of the task, and F is the objective function used to quantify the quality of the service combination plan.

9. The method for optimizing the collaborative perception service combination of production factors under the industrial Internet according to claim 8, wherein Iteratively analyzing the optimal service combination plan using the teaching optimization model includes: Step 1, generate an initial service combination plan for each subtask and calculate the initial objective function value; Step 2, take the service combination plan as a student, and iteratively update each student based on the teacher module of the teaching optimization model, and calculate the objective function value of the updated service combination plan; The expression for iteratively updating each student using the teacher module is: difference=r i *(X teacer -TF i *Mean) Among them, and respectively represent the pre-learning value and the post-learning value of the i-th student under the guidance of the teacher, and r i is the learning step of the i-th student, TF i is the learning factor, Mean is the average of all students, and X teacer is the individual with the optimal fitness value in the current population; Step 3, if the objective function value of the updated service combination plan is better, accept the update, otherwise, keep the original value; Step 4, randomly select students to communicate based on the student module of the teaching optimization model and perform updates, and calculate the objective function value of each updated student; The expression for randomly selecting students to communicate and perform updates using the student module is: Among them, and are the pre-learning value and the post-learning value of the i-th student under the guidance of the teacher, respectively, and r i is the learning step of the i-th student, X i and X j represent the i-th and j-th students respectively, and F(X i ) and F(X j ) represent the fitness values of the i-th and j-th students respectively; Step 5, if the objective function value of the updated student is better, accept the update, otherwise, keep the original value; Step 6, repeat steps 2 - 5 according to the set number of iterations or convergence conditions. In each iteration, record the current optimal service combination plan and its objective function value. When the number of iterations is reached or the convergence condition is met, output the optimal service combination plan.

Images