An assembly arrangement optimization method and system

Abstract

The application discloses a kind of component orchestration optimization method and system, method includes: obtaining the service request of target object;Based on service request, establish service combination model;Wherein, the definition of service combination model includes task set, service component set, service component action set and service component attribute value set;Through service combination model, based on service component combination reliability optimization, obtain target service component combination;Wherein, service component attribute value in service component attribute value set is associated with service component combination reliability;Service component combination reliability includes service component discovery reliability, service component binding reliability and service component execution reliability.The present application takes enhancing service component combination reliability as the primary optimization goal, can effectively solve the problem of poor optimization ability in large-scale service combination scene, and then realize the reliable and flexible combination of industrial software components in cloud edge collaboration, efficiently and high quality meet user demand, can be widely applied in component orchestration technical field.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of component orchestration, and in particular to a component orchestration optimization method and system. BACKGROUND

[0002] Automation equipment uses modular machine architecture to improve system flexibility and reduce design work. The current technology used by these equipment vendors and system integrators in developing control software for machines in various industrial automation applications is still based on centralized control. Improving the reusability of existing programs and software efficiency has become an important field of research in industrial automation, and service-oriented architecture (SOA) is a software design pattern that provides application functions (as services) to other larger applications based on different small software modules.

[0003] IEC 61499 is specifically designed for the development of distributed automation systems. It supports componentized services, promoting features such as modularity, reusability, flexibility, scalability, interoperability, portability, and reconfigurability. According to the IEC 61499 standard, services can be encapsulated as function blocks (FBs), thereby bridging the gap between plant and information technology (IT) distributed systems. The implementation of SOA is based on a set of open vendor-independent standards, such as Web services. By deploying these services in cloud centers and edge nodes, one can also benefit from the many advantages of cloud edge computing, making the system more intelligent, reconfigurable, flexible, and reusable.

[0004] When service components are deployed to cloud edge nodes, there are problems of reliability, cost, and response time related to quality of service (QoS) constraints in the process of service composition. To solve the above problems, there are currently various service orchestration methods based on meta-heuristic algorithms, but there are problems of low load balancing and high cost, and in large-scale service composition scenarios, existing service composition algorithms usually also have the problem of poor optimization ability. SUMMARY

[0005] Therefore, the embodiments of the present application provide a component orchestration optimization method and system to solve at least one problem existing in the prior art.

[0006] In one aspect, the embodiments of the present application provide a component orchestration optimization method, comprising:

[0007] obtaining a service request of a target object;

[0008] establishing a service composition model based on the service request;

[0009] The definition of the service composition model includes a task set, a service component set, a service component action set, and a service component attribute value set.

[0010] By using a service composition model, the target service component composition is obtained through optimization based on the reliability of service component composition.

[0011] Among them, the service component attribute values ​​in the service component attribute value set are associated with the service component composition reliability; the service component composition reliability includes service component discovery reliability, service component binding reliability, and service component execution reliability.

[0012] Optionally, based on service requests, a mathematical model for optimizing service component orchestration is established, including:

[0013] Based on service requests, construct a service composition model based on Markov decision processes;

[0014] The expression defining the service composition model is:

[0015] T = [t1, t2, t3, ..., t i ,…,t n ]

[0016]

[0017]

[0018]

[0019] In the formula, T represents the task set, t i This represents the i-th task of T; Represents a set of service components. Represents task t i The corresponding m-th service component; This represents the action set of the service component. Represents task t i The action corresponding to the m-th service component; Represents the set of attribute values ​​for a service component. Represents task t i The service quality attribute value of the corresponding m-th service component.

[0020] Optionally, the target service component composition can be obtained by optimizing the service component composition based on the reliability of the service component composition using a service composition model, including:

[0021] Using a service composition model, under constraints, the maximum value of the service component composition reliability is found by employing a deep reinforcement learning algorithm, with the service component composition reliability as the optimization objective.

[0022] The service component combination that corresponds to the maximum reliability of the service component combination is the target service component combination.

[0023] Optionally, the method further comprises a step of presetting a constraint condition, which comprises the following steps:

[0024] a first constraint condition is that a cost of the service component combination does not exceed a preset cost;

[0025] wherein an expression of the first constraint condition is:

[0026]

[0027] wherein c i represents a cost of a single service component, n represents a number of sub-tasks of the service component combination, and C represents the preset cost;

[0028] a second constraint condition is that a response time of the service component combination does not exceed a preset response time;

[0029] wherein an expression of the second constraint condition is:

[0030]

[0031] wherein t i represents a response time of a single service component, n represents a number of sub-tasks of the service component combination, and T represents the preset response time.

[0032] Optionally, the method further comprises a step of calculating a reliability of a single service component according to a reliability of the service component combination, which comprises the following steps:

[0033] based on the service combination model, calculating a service component discovery reliability, a service component binding reliability and a service component execution reliability;

[0034] according to the calculated service component discovery reliability, the service component binding reliability and the service component execution reliability, calculating the reliability of the single service component.

[0035] Optionally, an expression of the service component discovery reliability is:

[0036]

[0037]

[0038] wherein R i,dis represents a service component discovery reliability of an i-th service component in the service component combination, λ represents a parameter of an exponential distribution to which service requests arrive, μ represents a parameter of an exponential distribution to which a time of processing each search request arrives, t out represents a preset request timeout threshold, represents a probability that a request waiting time exceeds t out .

[0039] Optionally, the expression of the service component binding reliability is:

[0040]

[0041] wherein R i,bind represents the service component binding reliability of the i-th service component in the service component combination, represents the probability of a binding error caused by an error in the service description document, and λ represents a first failure intensity parameter, represents the probability of a binding failure caused by a protocol problem negotiation, and μ represents a second failure intensity parameter, and n represents the number of sub-tasks of the service component combination.

[0042] Optionally, the expression of the service component execution reliability is:

[0043] R ij,exc = exp(-λ i,j T i,j )

[0044] wherein R ij,exc represents the service component execution reliability between the i-th and j-th service components in the service component combination, λ i,j represents a parameter of a Poisson process after a link failure, T i,j represents the total communication time of exchanged information between the service components i and j.

[0045] Optionally, the method further comprises a step of training the deep reinforcement learning algorithm, and the step comprises the following steps:

[0046] initializing target parameters; wherein the target parameters comprise a size of an experience pool, a number of gradient descent samples, a learning rate, a decay coefficient, a maximum number of iterations, first network parameters, second network parameters, and a greediness coefficient;

[0047] initializing a task set, and based on the task set, combining a service component set, a service component action set, and a service component attribute value set, performing a loop iteration training until a maximum number of iterations is reached, and completing the training of the deep reinforcement learning algorithm; wherein the task set is initialized each time, and then a loop processing is performed in the task set after each iteration initialization; and every preset number of iterations, the second network parameters are copied and updated by the first network parameters.

[0048] In another aspect, an embodiment of the present application provides a component orchestration optimization system, comprising:

[0049] a first module configured to obtain a service request of a target object;

[0050] a second module configured to establish a service combination model based on the service request;

[0051] The definition of the service combination model comprises a task set, a service component set, a service component action set and a service component attribute value set.

[0052] The third module optimizes based on the service component combination reliability through the service combination model, and obtains a target service component combination.

[0053] The service component attribute value in the service component attribute value set is associated with the service component combination reliability. The service component combination reliability comprises service component discovery reliability, service component binding reliability and service component execution reliability.

[0054] In another aspect, an embodiment of the present application provides a component orchestration optimization system, comprising a processor and a memory.

[0055] The memory is used for storing a program.

[0056] The processor executes the program to realize the method as described above.

[0057] In another aspect, an embodiment of the present application provides a computer readable storage medium, which stores a program, and the program is executed by a processor to realize the method as described above.

[0058] The present application also discloses a computer program product or a computer program, which comprises computer instructions stored in a computer readable storage medium. A processor of a computer device can read the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to make the computer device execute the method as described above.

[0059] The embodiment of the present application first acquires a service request of a target object; establishes a service combination model based on the service request; wherein the definition of the service combination model comprises a task set, a service component set, a service component action set and a service component attribute value set; optimizes based on the service component combination reliability through the service combination model, and obtains a target service component combination; wherein the service component attribute value in the service component attribute value set is associated with the service component combination reliability; and the service component combination reliability comprises service component discovery reliability, service component binding reliability and service component execution reliability. Through the service combination model based on the task set, the service component set, the service component action set and the service component attribute value set, and by taking the enhancement of the service component combination reliability as the primary optimization target, the embodiment of the present application can effectively solve the problem of poor optimization ability in a large-scale service combination scenario, and further realizes the reliable and flexible combination of industrial software components in cloud-edge collaboration, and efficiently and high-quality meets the user demand. BRIEF DESCRIPTION OF DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained based on these drawings without creative labor.

[0061] Figure 1 A flowchart of a component arrangement optimization method provided by an embodiment of the present application is shown in the figure.

[0062] Figure 2 A flowchart of a deep reinforcement learning algorithm provided by an embodiment of the present application is shown in the figure.

[0063] Figure 3 A manufacturing IoT cloud edge collaborative business process scenario provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0064] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0065] First of all, it should be pointed out that the industrial cyber physical system relies on data from Internet of Things (IoT) devices and other systems, and is based on a combination of heterogeneous systems and components from different suppliers, which need to exchange data and integrate services seamlessly at runtime. The architecture of cloud edge collaboration provides efficient and low latency characteristics for the running of industrial software components, and tasks such as production and device control, autonomous decision or monitoring application require better Quality of Service (QoS) indicators, especially high reliability, which puts higher requirements on the arrangement of industrial software components. In the cloud edge collaborative manufacturing scenario, in addition to software services, there are also a considerable number of manufacturing and processing services. The time and quality of these services are uncontrollable, and manufacturing services mainly include processing services, experimental services, logistics services, etc., and the reliability of their execution is directly related to whether the user's demand can be efficiently and high-quality completed. After analyzing the user's demand, a workflow model is dynamically generated, and an input file for generating a service combination algorithm is searched.

[0066] In view of this, on the one hand, with reference to Figure 1 The embodiments of the present application provide a component arrangement optimization method, which comprises:

[0067] S100, obtaining a service request of a target object;

[0068] The service request proposed by the user (i.e. the target object) is obtained through an online system or the like.

[0069] S200, establish a service combination model based on the service request;

[0070] It should be noted that the definition of the service combination model includes a task set, a service component set, a service component action set, and a service component attribute value set; in some embodiments, a Markov decision process-based service combination model is constructed based on the service request; wherein the expression of the definition of the service combination model is:

[0071] T = [t1, t2, t3, …, tn] i , tn] n

[0072]

[0073]

[0074]

[0075] In the formula, T represents a task set, t i represents the i-th task of T; represents a service component set, represents the m-th service component corresponding to the task t i ; represents a service component action set, represents the action corresponding to the m-th service component corresponding to the task t i ; represents a service component attribute value set, represents the service quality attribute value of the m-th service component corresponding to the task t i ;

[0076] Specifically, in some specific embodiments, a workflow is established by an orchestration system in response to a service request, and a Markov decision process-based service combination model corresponding to the task is constructed, which contains the following definitions: T = [t1, t2, t3, …, tn] i , tn] n , wherein t i represents the i-th task of T, and n is the total number of tasks of T;

[0077] represents the m-th service component corresponding to the subtask t i , and n is the total number of service components of ;

[0078] represents the action corresponding to the m-th service component corresponding to the subtask t i , and n is the total number of actions of ;

[0079] ​ subtask t i a corresponding QoS (Quality of Service) attribute value of the mth service component. Wherein, T, and The meaning of n in the total number is inconsistent.

[0080] S300, based on the service component combination reliability, the target service component combination is obtained by optimizing the service combination model.

[0081] It should be noted that the service component attribute value in the service component attribute value set is associated with the service component combination reliability; the service component combination reliability includes service component discovery reliability, service component binding reliability and service component execution reliability; in some embodiments, through the service combination model, the maximum value of the service component combination reliability is solved by using a deep reinforcement learning algorithm under the constraint condition and taking the service component combination reliability as the optimization target; the service component combination corresponding to the maximum value of the service component combination reliability is determined as the target service component combination.

[0082] In some embodiments, the step of presetting the constraint condition comprises the following steps:

[0083] The first constraint condition is that the service component combination cost does not exceed the preset cost;

[0084] The expression of the first constraint condition is:

[0085]

[0086] In the formula, c i represents the cost of a single service component, n represents the number of service component combination subtasks, and C represents the preset cost;

[0087] The second constraint condition is that the service component combination response time does not exceed the preset response time;

[0088] The expression of the second constraint condition is:

[0089]

[0090] In the formula, t i represents the response time of a single service component, n represents the number of service component combination subtasks, and T represents the preset response time.

[0091] In some embodiments, the method further comprises a step of calculating the reliability of a single service component according to the reliability of the service component combination, which comprises the following steps: calculating the service component discovery reliability, the service component binding reliability and the service component execution reliability based on the service component combination model; and calculating the reliability of a single service component according to the service component discovery reliability, the service component binding reliability and the service component execution reliability.

[0092] wherein the expression of the service component discovery reliability is:

[0093]

[0094]

[0095] wherein R i,dis represents the service component discovery reliability of the i th service component in the service component combination, λ represents the parameter of the exponential distribution of the service request arrival, μ represents the parameter of the exponential distribution of the time for processing each search request, t out represents the preset request timeout threshold, represents the probability of the request waiting time exceeding t out .

[0096] The expression of the service component binding reliability is:

[0097]

[0098] wherein R i,bind represents the service component binding reliability of the i th service component in the service component combination, represents the probability of the binding error caused by the error of the service description document, λ represents the first failure intensity parameter, represents the probability of the binding failure caused by the protocol problem negotiation, μ represents the second failure intensity parameter, and n represents the number of sub-tasks of the service component combination.

[0099] The expression of the service component execution reliability is:

[0100] R ij,exc = exp(- λ i,j T i,j )

[0101] wherein R ij,exc represents the service component execution reliability between the i th and j th service components in the service component combination, λ i,j represents the parameter of the Poisson process after the link failure, T i,j represents the total communication time of the exchanged information between the service components i and j.

[0102] In particular, the QoS attribute value in the embodiment of the present application mainly focuses on the reliability value, and the reward value after a certain state execution action in the service combination process is represented by the reliability value. The higher the reliability value of the selected service is, the greater the reward value is, and otherwise the smaller the reward value is. According to the service component discovery reliability, the service component binding reliability and the service component execution reliability, the following calculation is performed:

[0103] R a = R dis * R bind * R exc

[0104] In the formula, R a represents the single service component reliability, R dis represents the service component discovery reliability, R bind represents the service component binding reliability, and R exc represents the service component execution reliability.

[0105] In some embodiments, the step of training the deep reinforcement learning algorithm further includes the following steps:

[0106] initializing target parameters; wherein the target parameters include an experience pool size, a gradient descent sample number, a learning rate, a decay coefficient, a maximum iteration number, first network parameters, second network parameters and a greed coefficient; initializing a task set, and based on the task set, combining a service component set, a service component action set and a service component attribute value set, performing a loop iteration training until the maximum iteration number is reached, and completing the training of the deep reinforcement learning algorithm; wherein the task set is initialized each time, and then loop processing is performed in the task set after each iteration initialization; and the second network parameters are copied and updated through the first network parameters every preset number of iterations.

[0107] In particular, as shown in FIG. 1, the deep reinforcement learning algorithm actually includes the following steps: Figure 2

[0108] Step 1, initializing the experience pool D size m = 10000, the gradient descent sample number Batch Size = 32, the learning rate a = 0.01, the decay coefficient γ = 0.9, the maximum iteration number Max Rpisode = 80000, initializing the action value function Q1 of the Q-Network network and its parameters θ1 (i.e. the first network parameters), initializing the action value function Q2 of the Q-Target network and its parameters θ2 = θ1 (i.e. the second network parameters), the greed coefficient ε = 0.9, and when the experience pool capacity reaches m, the value is gradually reduced by 0.0005 with the iteration number, until ε = 0.1.

[0109] Step 2, initializing the task T = [t1, t2, t3, …, t​i ,…,t n ], let i = 1;

[0110] Step 3, randomly generate a number x, x e (0, 1), if x > ε, randomly select action Otherwise select That is, the action corresponding to the maximum Q value (select the corresponding QoS reliability reward comprehensive average value under the current state);

[0111] Step 4, the selected action a i Output to the environment, record t i Corresponding service component set t i Corresponding service component set All reliability values of service components And the next service component set of t i As a sample Stored in experience pool D;

[0112] Step 5, if the number of samples stored in the experience pool is greater than m, randomly obtain Batch Size Samples from the experience pool update Q-Network, calculate the target value y k , where k = 1, 2, 3, …, m, the formula is as follows:

[0113]

[0114] Its loss function is: L(θ) = E[(y k -Q(S k, a k, θ1)) 2 ];

[0115] In the formula, Indicates the reward value obtained by performing action a k Under the state S k , γ indicates the decay coefficient, Q2 indicates the action value function of Q-Target network, S k+1 Indicates the next state transferred by taking action, a indicates the learning rate, Q indicates the output of Q-Network, θ1 indicates the parameters of the action value function of Q-Network, θ2 indicates the parameters of the action value function of Q-Target network, E indicates mathematical expectation, S k Indicates that the current is the kth subtask, a k Indicates the action performed under the state S k .

[0116] ​Step 6, in step 5, when i = n, then end the loop to enter the next episode, i.e., return to step 2, otherwise let i = i + 1, continue the current episode sampling, i.e., return to step 3;

[0117] In step 5, every C (preset number of times) episodes, the parameters of the Q-Network network are copied to the Q-Target network, the current Q value function is estimated by the Q-Network, and the target Q value is generated by the Q-Target. The Q-Target value is kept unchanged for a period of time, which reduces the correlation between the current Q value and the target Q value to a certain extent and improves the stability of the algorithm. In step 5, the training is iterated for a plurality of times. When the maximum number of iterations is reached, the converged Q-Network network is obtained. The candidate service attributes of each task in the workflow, the user's preferences and constraints on the QoS non-functional attributes are input into the converged network model, and the appropriate workflow candidate service component combination sequence is output.

[0118] As a preferred embodiment, the reliability model can add other QoS attributes as constraint conditions to make the method more universal. The QoS values of the sub-service components are normalized, and then the weighted sum of the normalized QoS values is obtained. The normalized values of each attribute are obtained, and then the weighted sum of the plurality of QoS attributes is obtained as the reward value, and the formula is as follows:

[0119]

[0120]

[0121] In the formula, r i is the reward value corresponding to the QoS index, i.e., the reward value after performing an action in the service combination process, ω i is the weight value corresponding to the index, and k is the total number of QoS indexes.

[0122] Specifically, the constraint conditions include that the service component combination cost does not exceed the preset cost:

[0123]

[0124]

[0125] In the formula, c i represents the cost of a single service component, n represents the number of service component combination sub-tasks, C represents the preset cost, c min represents the minimum value of the cost of all candidate service components of all tasks, and c max represents the maximum value of the cost of all candidate service components of all tasks.

[0126] Optionally, the constraint condition further comprises that the service component combination response time does not exceed a preset response time.

[0127]

[0128]

[0129] In the formula, t i represents a single service component response time, n represents a service component combination subtask number, T represents a preset cost, t min represents a minimum value of response times of all candidate service components of all tasks, t max represents a maximum value of response times of all candidate service components of all tasks.

[0130] In some specific embodiments, as shown in Figure 3 the embodiment of the present application needs to process user requirements into workflows and identify corresponding industrial software components to process the workflows in the cloud-edge collaborative industrial software component orchestration process. The workflow can be divided into two forms, simple and complex. The simple workflow is composed of tasks arranged in sequence. The complex workflow contains sequential, parallel, conditional and loop modes. The complex workflow can be decomposed into multiple simple workflows. Each task can use multiple industrial software components with similar functions, and the deployment in different industrial software components is arranged based on reliability to obtain a set of optimal component combination sequences. The candidate service components are deployed on multiple edge nodes or in the cloud center. The orchestration system gives the candidate service components that can complete the service combination based on user requirements, optimally selects based on the reliability index, and finally returns the optimal service component combination scheme to the user. The reliability index includes service component discovery reliability, service component binding reliability and service component execution reliability. The above reliability calculation function is used for next step optimization selection, and finally the solution with the maximum reliability value is output through the deep reinforcement learning based solving algorithm, that is, the requested optimal industrial software combination orchestration result.

[0131] In summary, the embodiment of the present application takes enhancing service component combination reliability as the primary optimization target, including service component discovery reliability, service component binding reliability and service component execution reliability to build a cloud-edge collaborative industrial software component reliability orchestration optimization model. The deep reinforcement learning based solving algorithm is used to solve the service component combination orchestration scheme, and the deep network is applied to improve the model performance. The problems of poor optimization ability of existing service combination algorithms in large-scale service combination scenarios can be effectively solved. The combination service QoS output in large-scale service combination problems is more optimal, the algorithm running time is shorter, the cloud-edge collaborative industrial software component reliable and flexible combination is realized, and the user requirements are efficiently and high-quality satisfied.

[0132] In another aspect, an embodiment of the present application provides an assembly orchestration optimization system, comprising: a first module configured to obtain a service request of a target object; a second module configured to establish a service combination model based on the service request; wherein the definition of the service combination model comprises a task set, a service component set, a service component action set, and a service component attribute value set; and a third module configured to obtain a target service component combination based on service component combination reliability by optimizing the service combination model; wherein a service component attribute value in the service component attribute value set is associated with the service component combination reliability; and the service component combination reliability comprises service component discovery reliability, service component binding reliability, and service component execution reliability.

[0133] The content of the method embodiments of the present application is applicable to the system embodiments of the present application, the system embodiments of the present application specifically implement the same functions as the above method embodiments, and achieve the same beneficial effects as the above method embodiments.

[0134] In another aspect, an embodiment of the present application further provides an assembly orchestration optimization system, comprising a processor and a memory.

[0135] The memory is configured to store a program.

[0136] The processor executes the program to implement the method as described above.

[0137] The content of the method embodiments of the present application is applicable to the system embodiments of the present application, the system embodiments of the present application specifically implement the same functions as the above method embodiments, and achieve the same beneficial effects as the above method embodiments.

[0138] In another aspect, an embodiment of the present application further provides a computer readable storage medium, the storage medium stores a program, and the program is executed by a processor to implement the method as described above.

[0139] The content of the method embodiments of the present application is applicable to the computer readable storage medium embodiments of the present application, the computer readable storage medium embodiments of the present application specifically implement the same functions as the above method embodiments, and achieve the same beneficial effects as the above method embodiments.

[0140] The embodiment of the present application further discloses a computer program product or a computer program, the computer program product or the computer program comprises computer instructions stored in a computer readable storage medium. The processor of the computer device can read the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to make the computer device execute the method as described above.

[0141] In some alternative embodiments, the function / operations mentioned in the block diagrams can not occur in the order mentioned in the operational illustrations. For example, depending on the involved function / operation, two blocks shown in succession can in fact be executed substantially concurrently or the blocks can sometimes be executed in reverse order, depending upon the functionality / operations involved. Further, embodiments presented and described in the flow diagrams are only examples of implementing the processes. The processes disclosed are not limited to the order shown in the figures. Alternative embodiments are possible, where some operations are changed, and where the sub-operations described as part of a larger operation are performed independently.

[0142] Further, while the present application has been described in the context of functional modules, it is to be understood that one or more of the functions and / or features described can be integrated in a single physical device and / or software module, or one or more functions and / or features can be implemented in separate physical devices or software modules. It will also be appreciated that detailed discussion of the actual implementation of each module is not necessary for an understanding of the application. Rather, the actual implementation of the modules, in light of the attributes, functions, and internal relationships of the various functional modules disclosed herein, will be apparent to one of ordinary skill in the art, given the benefit of this disclosure. Accordingly, the present application is not limited to the specific embodiments illustrated herein, but is capable of being practiced with variations that are within the scope of the claims appended hereto, and equivalents thereof, as defined by the appended claims.

[0143] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the part of the technical solutions that make essential contributions to the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0144] The logic and / or steps represented in the flowcharts and / or described herein, for example, can be embodied in non-transitory computer-readable media, executed by an instruction execution device, apparatus or device, such as a computer-based device, a processor-based device or other device that can fetch the instructions from the instruction execution device, apparatus or device and execute the instructions, or in conjunction with the instruction execution device, apparatus or device. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution device, apparatus or device.

[0145] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can also be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example, via optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

[0146] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution device. For example, if implemented in hardware, as in another embodiment, any of the following techniques, which are known in the art and practice of the present application, can be used alone or in any combination: discrete logic circuitry having logic gates for implementing logic functions upon data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and so forth.

[0147] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like means that the particular feature, structure, material or characteristic being described in connection with the embodiment or example contains in at least one embodiment or example of the present application. The illustrative representations of the above terms in the specification do not necessarily refer to the same embodiment or example. Moreover, the particular features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0148] While the embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary and are not to be construed as limiting the scope of the application. The scope of the application is defined by the appended claims and their equivalents.

[0149] The above is a specific description of the preferred embodiment of the present application, but the present application is not limited to the described embodiment, and those skilled in the art can make various equivalent modifications or replacements without departing from the spirit of the present application, and these equivalent modifications or replacements are all included in the scope defined by the claims of the present application.

Claims

1. An assembly orchestration optimization method, characterized by, The method comprises the following steps: obtaining a service request of a target object; establishing a service combination model based on the service request; wherein the definition of the service combination model comprises a task set, a service component set, a service component action set and a service component attribute value set; optimizing based on service component combination reliability through the service combination model to obtain a target service component combination; wherein the service component attribute value in the service component attribute value set is associated with the service component combination reliability; the service component combination reliability comprises service component discovery reliability, service component binding reliability and service component execution reliability; wherein the optimization based on service component combination reliability through the service combination model to obtain a target service component combination comprises: solving the maximum value of the service component combination reliability under the constraint condition by using a deep reinforcement learning algorithm with the service component combination reliability as the optimization target through the service combination model; determining that the service component combination corresponding to the maximum value of the service component combination reliability is the target service component combination; calculating the single service component reliability according to the service component combination reliability, which comprises the following steps: calculating the service component discovery reliability, the service component binding reliability and the service component execution reliability based on the service combination model; calculating the single service component reliability according to the service component discovery reliability, the service component binding reliability and the service component execution reliability; the expression of the service component discovery reliability is: wherein denotes the service component discovery reliability of the i-th service component in the service component composition, denotes the service component discovery reliability of the i-th service component in the service component composition, denotes the parameter of the exponential distribution to which the service request arrival obeys, denotes the parameter of the exponential distribution to which the time of processing each lookup request obeys, denotes the preset request timeout threshold, denotes the probability that the request waiting time exceeds the preset request timeout threshold.

2. The method of claim 1, wherein, the step of establishing a service component arrangement optimization mathematical model based on the service request comprises: constructing a service combination model based on Markov decision process based on the service request; wherein the expression of the definition of the service combination model is: wherein, represents a task set, represents a first task of the task set; represents a service component set, represents a task corresponding to a first service component of the service component set; represents a service component action set, represents a task corresponding to an action of a first service component of the service component set; represents a service component attribute value set, represents a task corresponding to a quality of service attribute value of a first service component of the service component set.

3. The method of claim 1, wherein, the step of predefining a constraint condition comprises the following steps: taking the service component combination cost not exceeding the predefining cost as the first constraint condition; wherein the expression of the first constraint condition is: In the formula, represents the cost of a single service component, represents the number of sub-tasks of a service component combination, represents a preset cost; taking the service component combination response time not exceeding the predefining response time as the second constraint condition; wherein the expression of the second constraint condition is: In the formula, represents the single service component response time, represents the service component combination subtask number, represents the preset response time.

4. The method of claim 1, wherein, the expression of the service component binding reliability is: wherein denotes the service component binding reliability of the service component of the service component combination, denotes the error probability of the service description document leading to a binding error, denotes a first failure intensity parameter, denotes the probability of a protocol problem negotiation leading to a binding failure, denotes a second failure intensity parameter, denotes the number of service component combination subtasks.

5. The method of claim 1, wherein, the expression of the service component execution reliability is: wherein denotes the service component execution reliability between the i-th and the j-th service component in a service component composition, denotes the parameter of the Poisson process after a link failure, denotes the service component and the total communication time for exchanging information between the service components​ 6. The method of claim 1, wherein, the step of training the deep reinforcement learning algorithm comprises the following steps: initializing target parameters; wherein the target parameters comprise an experience pool size, a gradient descent sample number, a learning rate, a decay coefficient, a maximum iteration number, a first network parameter, a second network parameter and a greed coefficient; initializing a task set, and performing cyclic iteration training based on the task set in combination with the service component set, the service component action set and the service component attribute value set until the maximum iteration number is reached to complete the training of the deep reinforcement learning algorithm; wherein the task set is initialized each time, and then cyclic processing is performed in the task set initialized each time; the second network parameter is copied and updated through the first network parameter every predefining number of iterations.

7. An assembly orchestration optimization system, comprising: The method comprises the following steps: a first module for obtaining a service request of a target object; A second module, based on the service request, establishes a service combination model; Wherein, the definition of the service combination model includes a task set, a service component set, a service component action set and a service component attribute value set; A third module, through the service combination model, optimizes based on service component combination reliability to obtain a target service component combination; Wherein, the service component attribute value in the service component attribute value set is associated with the service component combination reliability; the service component combination reliability includes service component discovery reliability, service component binding reliability and service component execution reliability; Wherein, the third module includes: Through the service combination model, under the constraint condition, taking the service component combination reliability as the optimization target, using a deep reinforcement learning algorithm to solve the maximum value of the service component combination reliability; Determining the service component combination corresponding to the maximum value of the service component combination reliability as the target service component combination; According to the service component combination reliability, calculating a single service component reliability, which includes the following steps: Based on the service combination model, calculating service component discovery reliability, service component binding reliability and service component execution reliability; According to the calculation of service component discovery reliability, service component binding reliability and service component execution reliability, the single service component reliability is calculated; The expression of the service component discovery reliability is: wherein denotes the service component discovery reliability of the service component in the service component combination, denotes the parameter of the exponential distribution to which the service requests arrive, denotes the parameter of the exponential distribution to which the time of processing each lookup request is subject to, denotes the preset request timeout threshold, denotes the probability that the request waiting time exceeds the request timeout threshold.

Images