Abnormal adaptive reconstruction method and system for CPS manufacturing component service composition based on improved HHO algorithm

Abstract

The present invention discloses a method and system for abnormal adaptive reconstruction of CPS manufacturing component service combination based on an improved HHO algorithm. First, the CPS manufacturing component service is determined according to the order task and the execution path and mandatory conditions of the service combination are constructed. When checking whether there are abnormal CPS manufacturing components in the service combination, constrained reconstruction is performed starting from the first abnormal CPS manufacturing component service to obtain the execution path of the reconstructed service combination. The original service combination is updated until the order task is completed. The service combination reconstruction model provided by the present invention adopts multi-constraint features, cluster optimization strategy and population diversity supplement strategy to improve the HHO algorithm. It has superiority in solving the CPS manufacturing component service combination reconstruction model and is an efficient solution algorithm based on the improved HHO algorithm.

Description

Technical Field

[0001] The present invention relates to the field of intelligent manufacturing technology, and in particular to a CPS manufacturing component service combination abnormal adaptive reconstruction method and system based on an improved HHO algorithm. Background Art

[0002] The current research on service composition models and methods in cloud environments is all conducted under an ideal cloud manufacturing environment without exceptions and constraints, without considering actual dynamic constraints (such as manufacturing service exceptions, manufacturing service occupancy time, and manufacturing service strong coupling constraints). Due to the complex dynamic manufacturing environment, some unpredictable service exceptions will inevitably occur from time to time, which will cause the original CPS manufacturing component service to fail to execute smoothly, and thus fail to complete the manufacturing task on time. That is, when a service exception occurs during the execution of the service composition, the traditional service composition model and method cannot dynamically reconstruct and adjust the CPSMCSC, so that the manufacturing task cannot be completed on time with quality and quantity according to the established CPSMCSC execution path.

[0003] In addition, when reconstructing CPSMCSC, the corresponding actual constraints must be considered (mandatory time constraints of the original CPSMCSC execution path, manufacturing service occupancy time, and strong coupling constraints of manufacturing services). Therefore, when component service anomalies occur, dynamically reconstructing and adjusting the CPSMCSC execution path under actual dynamic constraints has become a practical problem that needs to be urgently solved on a cloud platform.

[0004] Research on cloud manufacturing service composition optimization has made great progress, but some issues, such as how to efficiently reconstruct and adjust the service composition when service anomalies occur during the execution of CPS manufacturing component service composition, have not received much attention from researchers. Existing research is often carried out under relatively ideal conditions, and rarely considers the actual constraints in the service execution process. This paper aims to consider the problem of reconstructing and optimizing the service composition when service anomalies occur during the execution of CPS manufacturing component service composition. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a CPS manufacturing component service combination abnormal adaptive reconstruction method based on an improved HHO algorithm.

[0006] The CPS manufacturing component service combination abnormal adaptive reconstruction method based on the improved HHO algorithm provided by the present invention comprises the following steps:

[0007] Determine the CPS manufacturing component services based on the order tasks and construct the execution path and mandatory conditions of the service combination;

[0008] Check whether there are abnormal CPS manufacturing components in the CPS manufacturing component service combination, and determine the first abnormal CPS manufacturing component service;

[0009] According to the mandatory conditions and the constraints of the CPS manufacturing component service occupancy time, the subsequent CPS manufacturing component service combinations are constrainedly reconstructed starting from the first abnormal CPS manufacturing component service, and the execution path of the reconstructed service combination is obtained;

[0010] Update the execution combination of the original service combination according to the execution path of the reconstructed service combination;

[0011] The loop executes the check to see if there are any abnormal CPS manufacturing components until the order task is completed.

[0012] Furthermore, the execution path of the reconstructed service composition is implemented based on the improved Harris Hawk model algorithm, and is specifically performed in the following steps:

[0013] Construct a Harris Hawk population model and initialize the Harris Hawk population;

[0014] The constraints in CPS manufacturing component services are used to control the population encoding boundary, and then the fitness value is calculated;

[0015] Initialize the optimal position X and its fitness value;

[0016] Initialize or update parameters in the Harris Hawk Flock Model;

[0017] Adopt the corresponding HHO search mechanism according to the parameter value, and apply the code update mechanism to update the code;

[0018] Set the Harris Hawk population location and apply constraints to control the population coding boundaries;

[0019] According to the mandatory conditions and service occupancy time constraints, invalid solutions and conflicting solutions are eliminated;

[0020] Calculate and update the cost C and service quality S based on the reconstruction information of the CPS manufacturing component service combination;

[0021] Calculate the new population fitness value and update the optimal fitness and its encoding information;

[0022] The parameters and encoding information in the Harris Hawk Swarm model algorithm are calculated cyclically until the preset optimal solution is obtained and output.

[0023] Further, the encoding information is performed according to the following steps:

[0024] Determine the original code starting from the location of the first CPS manufacturing component service anomaly point;

[0025] Generate random codes and adjust the random codes according to the constraints to obtain reconstructed codes;

[0026] Then compare it with the original code to obtain the code reconstruction information matrix RE;

[0027] According to the constraints, the corresponding HHO search mechanism is used to update the coding information to obtain the updated reconstructed coding;

[0028] The cost matrix C and the service quality matrix S are updated according to the encoded reconstruction information matrix RE.

[0029] Furthermore, the constraints in the CPS manufacturing component service specifically include any one or any combination of the following:

[0030] Constraint 1: The manufacturing service in the reconstructed service composition cannot be an abnormal CPS manufacturing component service; if the abnormal CPS manufacturing component service is a service with a strong coupling constraint, then it is necessary to combine constraint 2 to update its candidate service set;

[0031] Constraint 2: CPS manufacturing component services with strong coupling constraints must be selected at the same time in the adjacent subtask candidate service set to be effective; in addition, one upstream coupled CPS manufacturing component service can correspond to multiple downstream coupled CPS manufacturing component services;

[0032] Constraint 3: The delivery time of the original CPSMCSC execution path remains unchanged before and after the service composition reconstruction, that is, the completion time of the reconstructed service composition execution must be within the delivery time of the original CPSMCSC;

[0033] Constraint 4: During the CPS manufacturing component service matching selection process, the CPS manufacturing component processing time must not conflict with its existing occupied time.

[0034] Further, the cost matrix C and the service quality matrix S are updated according to the encoding reconstruction information matrix RE, and the steps are as follows:

[0035] When abnormal CPS manufacturing components are replaced, the cost, processing quality and service quality after reconstruction are calculated according to the following formula:

[0036]

[0037]

[0038] minC=C M +C L +C Re ;

[0039]

[0040] Among them, C represents cost; Q represents processing quality; S represents service quality;

[0041] The following method is used to standardize it:

[0042] The negative attribute index C is quantified using the following formula:

[0043]

[0044] The indicators Q and S of positive attributes are quantified using the following formula:

[0045]

[0046]

[0047] The optimization objective function is determined according to the following formula:

[0048] minFitness=1 / (w C × c (X)+w Q × Q (X)+w S × S (X));

[0049] Among them, X is the candidate service set, w C , w Q , w S are the weights of cost, processing quality and service quality in the optimization target respectively; each weight is calculated by the hierarchical analysis method.

[0050] Furthermore, the following optimization strategies are also included, as follows:

[0051] Strategy 1: Cluster Optimization Strategy

[0052] After all individuals in the population have completed the update and constraints 1-4 have been imposed, the population fitness calculation is performed;

[0053] Strategy 2: Population diversity supplementation strategy

[0054] After applying the above cluster optimization strategy, the soft encirclement and hard encirclement search mechanisms of the fast dive strategy in the original HHO were applied to supplement the Harris hawk population.

[0055] Furthermore, when calculating the minimum cost of the service combination in the improved Harris Hawk model algorithm, the following formula is used:

[0056]

[0057]

[0058] minC=C M +C L ;

[0059] Among them, C is the total cost, CM and CL represent the manufacturing cost and logistics cost respectively;

[0060] c(Rs i ) means Rs i the manufacturing cost of performing the subtask;

[0061] lc(Rs i ,Rs i+1 ) means from service Rs i to Rs. i+1 logistics costs.

[0062] Furthermore, when calculating the optimal processing quality in the improved Harris Eagle Swarm model algorithm, the following formula is used:

[0063]

[0064] Among them, Q represents the final processing quality of the entire product;

[0065] PS(Rs i ) indicates that for this subtask,

[0066] CPS Manufacturing Component Services Rs. i The achievable processing quality.

[0067] Furthermore, when calculating the best service quality in the improved Harris Hawk model algorithm, the following formula is used:

[0068]

[0069] Among them, S represents the service evaluation of the whole process, SR (Rs i ) means Rs i service quality level.

[0070] The CPS manufacturing component service combination abnormal adaptive reconstruction system based on the improved HHO algorithm provided by the present invention includes a memory and a processor, wherein the memory stores executable code, and the above method is implemented when the processor executes the executable code.

[0071] In order to achieve the above object, the present invention provides the following technical solutions:

[0072] The beneficial effects of the present invention are:

[0073] The CPS manufacturing component service combination abnormal adaptive reconstruction method based on the improved HHO algorithm provided by the present invention overcomes the situation in the previous fog manufacturing paradigm that resource allocation optimization is carried out under ideal conditions without abnormalities and constraints in manufacturing services. This makes the existing models and methods unable to be applied to the optimization of CPS manufacturing component service combination (CPSMCSC) of actual unmanned production lines under various actual constraints, and it is even more impossible to adaptively reconstruct and adjust the original execution path of the CPS manufacturing component when service abnormalities occur during the execution of manufacturing services.

[0074] At the same time, due to the forced time constraints of the original execution path, the manufacturing service occupancy time constraints and the strong coupling constraints of the manufacturing service in the execution process of the CPS manufacturing component combination, a CPS manufacturing component service combination adaptive reconstruction and adjustment model (CMSCAR) under actual constraints is proposed when an abnormality occurs in the service-oriented CPS manufacturing component. In order to solve the model, on the basis of a detailed analysis of the essential characteristics of the problem, a variety of optimization strategies are integrated, and the improved Harris Hawk optimization algorithm (IHHO) is adopted. The actual example shows that the improved model and algorithm provided by the present invention can efficiently (compared with GWO and PSO) adaptively reconstruct and adjust the executing service combination under actual constraints when an abnormality occurs in the manufacturing service, thereby improving the robustness of the execution of the CPS manufacturing component service combination under the fog manufacturing paradigm.

[0075] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] In order to make the purpose, technical solution and beneficial effects of the present invention clearer, the present invention provides the following drawings for illustration:

[0077] Figure 1 The CPS manufacturing component service anomaly occurs when the CPSMCSC dynamically reconfigures the model.

[0078] Figure 2 Flowchart of algorithm reconstruction for improving HHO manufacturing cloud service composition.

[0079] Figure 3 Schematic diagram of the coding update mechanism.

[0080] Figure 4 Iterative solution process for the three algorithms.

[0081] Figure 5Adjust the Gantt chart for service composition execution path reconstruction when exceptions occur.

[0082] Figure 6 This is a schematic diagram of the production line layout before reconstruction.

[0083] Figure 7 This is a schematic diagram of the reconstructed production line layout. DETAILED DESCRIPTION

[0084] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0085] Example 1

[0086] The present embodiment provides a method for adaptively reconfiguring abnormal CPS manufacturing component service combinations based on an improved HHO algorithm. This method is a mathematical model for dynamic reconstruction of CPS manufacturing component service combinations under the fog manufacturing paradigm. During the operation of the cloud manufacturing platform under the fog manufacturing paradigm, CPS manufacturing component manufacturing service demanders continuously submit their own manufacturing needs (orders or manufacturing tasks) to the platform. The platform selects the best CPS manufacturing component service for each user based on their personalized needs and generates a scheduling plan for tasks and services. The service combination process is as follows: Figure 1 As shown in the figure, when a complex manufacturing task is submitted to the service platform, the task is first decomposed into multiple subtasks. At the same time, the corresponding CPS manufacturing component service candidate set is matched to each subtask according to specific requirements, and then an optimal service is selected from each CPS manufacturing component service candidate set. Finally, all selected CPS manufacturing component services are called in sequence to build a service combination execution path.

[0087] For some important products or urgent production of certain products in extraordinary times, the delivery time of some manufacturing tasks is often a mandatory condition to be met during the execution of the service combination. Under this condition, if an abnormality occurs in a node in the CPS manufacturing component service combination, it is necessary to constrain the subsequent CPS manufacturing component service combination starting from the first abnormal CPS manufacturing component service to ensure the smooth execution and on-time delivery of the entire manufacturing service process. Therefore, the fog manufacturing CPS manufacturing component service combination reconstruction must impose the original CMSC execution path mandatory time constraint to ensure the total manufacturing task delivery time. In addition, during the CPS manufacturing component service combination reconstruction process, the reconstructed CPS manufacturing component service combination must not conflict with the occupancy time of the selected service. In the manufacturing field, there is a strong coupling relationship between manufacturing resources, that is, only when coupled manufacturing equipment is selected at the same time can the processing task be completed.

[0088] This embodiment provides an adaptive reconstruction optimization model for the original CPS manufacturing component service execution path under actual constraints when an exception occurs in the CPS manufacturing component service under the fog manufacturing paradigm. This can greatly improve the success rate of the CPS manufacturing component service combination execution under actual constraints under the fog manufacturing paradigm.

[0089] Mathematical model description of dynamic reconstruction of CPS manufacturing component service composition under practical constraints

[0090] The dynamic reconstruction model of CPS manufacturing component service composition when service anomalies occur is based on the following conditions:

[0091] ① When a task is processed in a certain manufacturing service, it is necessary to wait until the task is completely processed before calling the next manufacturing service, that is, the manufacturing activities are not carried out in the assembly line production method.

[0092] ② Manufacturing services execute tasks in a sequential manner, that is, there can only be one task being processed at any time, and the task execution times do not overlap.

[0093] ③The logistics costs and time between manufacturing services in the same location are not taken into account.

[0094] ④Material quality loss caused by machining is not considered.

[0095] ⑤The manufacturing service is locked during the task time period and will be released when the task is completed.

[0096] ⑥When the service is abnormal, the task has not yet been executed to the abnormal point.

[0097] like Figure 1 As shown, Figure 1 Figure 3 is a schematic diagram of the dynamic reconstruction model of the CPS manufacturing component service combination. When a CPS manufacturing component service anomaly occurs under the fog manufacturing paradigm provided in this embodiment, the service anomaly occurs randomly and uncertainly in the actual management process of the actual CPSMCSC execution process by the fog manufacturing platform. The fog manufacturing control mode based on the cloud platform and hierarchical digital twin is used for fine perception. Therefore, the service anomaly in the execution process of the original CPSMCSC is also generated in a random manner. Then, the original CPSMCSC is adaptively reconstructed from the first service anomaly point under the forced time constraints of the original CPSMCSC execution path, the dynamic manufacturing service occupancy time constraints and the strong coupling constraints of manufacturing resources. Then, the execution process of the reconstructed CPSMCSC is finely sensed and controlled through the cloud platform and fog manufacturing control technology.

[0098] Rs = {L, SR, OcT, PS} represents CPS manufacturing component services,

[0099] Where L represents the location of effective output of manufacturing services,

[0100] SR represents the service quality of the CPS manufacturing component service (evaluated with a value between 0 and 10).

[0101] OcT is the known occupancy time of the CPS manufacturing component dynamics,

[0102] PS is the pass rate, which is used to indicate the processing quality of manufacturing services. This value is different for different tasks.

[0103] MT={LS, LE, Am, M, Lmp} represents the processing task.

[0104] Among them, LS represents the location where the manufacturing blank is sent out.

[0105] LE indicates the location where the final product is delivered.

[0106] Am is the quantity required for processing,

[0107] M represents the weight of a single blank.

[0108] Lmp represents the logical processing of the task, which is a task chain composed of processing subtasks.

[0109] Lmp={subtask 1, subtask 2, subtask 3,…, subtask n}.

[0110] Exroute = {Rs1, Rs2, Rs3, ..., Rsn} is the original CPSMCSC execution path, which represents the CPSMCSC that meets the processing task after platform scheduling.

[0111] Rsi(i∈[1,n]) is a CPS manufacturing component service that corresponds one-to-one to the manufacturing subtask i(i∈[1,n]).

[0112] G(i)={Rs i,1 , Rs i,2 , Rs i,3 , …, Rs i,m}(i∈[1,n],m∈N*) represents the candidate service set of the i-th CPS manufacturing component service on the CPSMCSC execution path.

[0113] The most important service optimization selection indicators in service-centric manufacturing systems are summarized into four categories: shortest completion time, optimal processing quality, minimum cost, and best service quality, namely TQCS indicators.

[0114] Use its indicators and calculation methods of indicators

[0115] Shortest completion time

[0116]

[0117] Where T is the total time; n is the total number of subtasks; i is the sequence number of the current subtask;

[0118] TM and TL represent manufacturing time and logistics time, respectively;

[0119] t(Rs i ) means Rs i Manufacturing time;

[0120] lt(Rs i ,Rs i+1 ) means from Rs i to Rs. i+1 Logistics time;

[0121] LS and LE represent the starting and ending positions of logistics, respectively;

[0122] (2) Optimal processing quality:

[0123]

[0124] Among them, Q represents the final processing quality of the entire product;

[0125] PS(Rs i ) indicates that it is for this subtask;

[0126] CPS Manufacturing Component Services Rs. i The achievable processing quality.

[0127] (3) Minimum manufacturing cost:

[0128]

[0129]

[0130] minC=C M +C L (5)

[0131] Among them, C is the total cost, C M and C L represent manufacturing cost and logistics cost respectively;

[0132] c(Rs i ) means Rs i the manufacturing cost of performing the subtask;

[0133] lc(Rs i ,Rs i+1 ) means from service Rs i to Rs. i+1 logistics costs;

[0134] m represents the mass of the part;

[0135] (4) Best service quality:

[0136]

[0137] Among them, S represents the service evaluation of the whole process, SR (Rs i ) means Rs i service quality level.

[0138] Practical constraints in the dynamic reconstruction of CPS manufacturing component service composition

[0139] Through summarizing the deployment and management experience of the cloud platform for machining fog manufacturing paradigm, it is found that the actual CPS manufacturing component service combination reconstruction must meet the corresponding actual constraints, which can be summarized as the following four:

[0140] (1) Constraint 1: The manufacturing service in the reconstructed service composition cannot be an abnormal CPS manufacturing component service. If the abnormal CPS manufacturing component service is a service with a strong coupling constraint, then its candidate service set needs to be updated in combination with constraint 2.

[0141] (2) Constraint 2: CPS manufacturing component services with strong coupling constraints must be selected simultaneously in the adjacent subtask candidate service set to be valid. In addition, one upstream coupled CPS manufacturing component service can correspond to multiple downstream coupled CPS manufacturing component services.

[0142] If <Rs i,j >Then <Rs i+k1,j+1 R i+k2,j+1 …orR i+kn,j+1 >

[0143] (i,k1,k2,…,kn=1,2,…n; j=1,2,…,m)(7)

[0144] Among them, Rs i,j Rs represents the jth service of the i-th manufacturing service candidate set on the cloud manufacturing service composition execution path; i+k1,j+1 represents the j+1th service of the i+k1th manufacturing service candidate set on the execution path of the cloud manufacturing service composition;

[0145] (3) Constraint 3: The delivery time of the original CPSMCSC execution path remains unchanged before and after the service composition reconstruction, that is, the completion time of the reconstructed service composition execution must be within the delivery time of the original CPSMCSC. It can be expressed as:

[0146]

[0147] Among them, t(Exroute)reconstruction represents the total time of the reconstructed execution path;

[0148] t(Exroute) 原CPSMCSC Represents the original CPSMCSC completion time;

[0149] (4) Constraint 4: During the CPS manufacturing component service matching selection process, the CPS manufacturing component processing time must not conflict with its existing occupied time. It can be expressed as:

[0150]

[0151] Where k represents the number of services; OcT represents the service occupancy time;

[0152] Objective function of dynamic reconstruction model of CPS manufacturing component service composition under practical constraints

[0153] Based on the dynamic reconstruction model of CPS manufacturing component service combination under actual constraints, time is no longer used as an evaluation indicator in the model as a mandatory constraint that must be met. Therefore, according to the characteristics of the studied model, the TQCS optimization indicators are revised to cost C (Cost), processing quality Q (Quality) and service quality S (Service Quality).

[0154] In addition, the replacement of CPS manufacturing component services after the original CPSMCSC abnormal point is bound to cause dissatisfaction among the replaced manufacturing service providers, so it is necessary to give appropriate compensation to the replaced CPS manufacturing component service providers based on whether the CPS manufacturing component services are replaced. This is bound to increase the corresponding reconstruction costs and reduce the corresponding CPS manufacturing component service quality evaluation, so the cost and service quality need to be revised.

[0155]

[0156]

[0157] minC=C M +C L +C Re (12)

[0158]

[0159] Among them, SR(Rs i ) indicates service Rs i The processing quality is evaluated by a value between 0 and 10; S 损失 Indicates the loss of service quality; C 补偿 Represents manufacturing cost compensation; C Re represents the manufacturing service reconfiguration cost;

[0160] Considering that the optimization directions, dimensions and ranges of these three optimization indicators are different, the following method is used to standardize them:

[0161] The negative attribute index C is quantified using the following formula:

[0162]

[0163] Among them, r C (X) represents the standardized cost of the dataset; C(X) max represents the maximum cost in the data set; C(X) represents the average cost in the data set; C(X) min represents the minimum cost in the data set;

[0164] The indicators Q and S of positive attributes are quantified using the following formula:

[0165]

[0166]

[0167] Among them, r Q (X) represents the processing quality of the standardized data set; Q(X) max Indicates the highest processing quality in the data set; Q(X) min represents the lowest processing quality in the data set; Q(X) represents the average processing quality in the data set;

[0168] r s (X) represents the quality of service of the standardized data set; S(X) max Indicates the highest quality of service for a data set; S(X) min represents the minimum service quality of the data set; S(X) represents the average service quality of the data set;

[0169] Therefore, the optimization objective function is:

[0170] minFitness=1 / (w C × c (X)+w Q × Q (X)+w S × S (X))(17)

[0171] Among them, X is the candidate service set, w C , w Q , w S are the weights of cost, processing quality and service quality in the optimization objective. Each weight is calculated by the hierarchical analysis method.

[0172] Original Harris Hawk Optimization Algorithm: Heidari et al. (2019) developed the Harris Hawk Optimization Algorithm (HHO) for solving engineering optimization problems by imitating the cooperative hunting behavior of Harris Hawks to prey on escaping prey (usually rabbits). The main feature of this HHO optimization algorithm is that it contains multiple search mechanisms, called "raid" or "seven kill" strategies to attack and capture prey, which can effectively improve the diversity and convergence of solutions. Like other metaheuristic algorithms, HHO is also a population-based algorithm that includes an exploration phase and an exploitation phase. In the exploration phase (|E|>=1), the two predation strategies are switched by randomly generated q. When q<0.5, the position of the Harris Hawk group is updated according to formula (18). When q≥0.5, the Harris Hawk will randomly perch on a tall tree, and its random position in the Harris Hawk group is updated using formula (19).

[0173]

[0174]

[0175] In the HHO development stage, there are four strategies based on the random escape success rate r and the escape energy |E|.

[0176] When the condition (r ≥ 0.5, |E| ≥ 0.5) occurs, the Harris Hawk group will implement a soft encirclement tactic, and its mathematical model is the following formula (20) and formula (21).

[0177]

[0178]

[0179] When r ≥ 0.5 and |E| < 0.5, the rabbit cannot escape because it is exhausted. In addition, the Harris hawks attack the prey through the surprise attack tactics simulated by formula (22).

[0180]

[0181] When r<0.5 and |E|≥0.5 appear, HHO combines the levy flight strategy and gradually selects the best way to pounce on the prey, executing the following strategies respectively. That is, if the next position is better than the current position, the position of the Harris hawk group will be updated by formula (23-27), otherwise the Harris hawk group will perform a rapid dive based on the levy flight strategy according to formula (23,25-28).

[0182]

[0183]

[0184] Z=Y+S×Levy(dim)(25)

[0185]

[0186]

[0187] u and v take random values ​​in the interval (0,1), and β is a default constant value of 1.5.

[0188] When r<0.5, |E|<0.5, the position of the Harris Hawk is updated by formula (26-29).

[0189]

[0190] In addition, HHO realizes the transition from the exploration stage to the development stage by applying formula (29).

[0191]

[0192] CPS manufacturing component service composition reconstruction algorithm based on improved HHO: No free lunch theory reveals that the optimization algorithm search process must match the essence of the problem being studied. In view of the multi-constraint and discretization characteristics of the dynamic reconstruction model of CPS manufacturing component service composition when service anomalies occur, the flowchart of the CPS manufacturing component service composition reconstruction algorithm (IHHO) based on improved HHO is as follows: Figure 2 As shown, Figure 2 Flowchart of algorithm reconstruction for improving HHO manufacturing cloud service composition.

[0193] Initialize the original CPS manufacturing component service combination execution route and its encoding and other parameters according to Table 1;

[0194] Perceive CPS manufacturing component service failure information, strong coupling relationships, time constraints, and service occupancy time;

[0195] Harris hawk population initialization;

[0196] Apply constraint 1-2 to control the population encoding boundary, and then use strategy 1 to calculate the fitness value;

[0197] Initialize the optimal position X and its fitness value;

[0198] Initialize or update HHO parameters (q, r, E0, l);

[0199] Apply the corresponding search mechanism (q, r, |E|, l) according to the parameter value, and apply the code update mechanism to update the code;

[0200] Strategy 2 was applied to supplement the Harris Hawk population location, and Constraints 1-2 were applied to constrain the control coding boundary;

[0201] Determine whether the mandatory time constraint is met. If not, remove invalid solutions. If yes, proceed to the next step.

[0202] Determine whether the service occupancy time constraint is met. If not, remove the conflicting solution; if yes, proceed to the next step;

[0203] Apply the reconstruction information RE to update the cost C and service quality S;

[0204] Calculate the new population fitness value and update the optimal fitness and its encoding information;

[0205] Determine whether the maximum number of iterations has been reached. If not, initialize or update the HHO parameters (q, r, E0, l);

[0206] If not, determine whether the optimal solution is an empty set. If it is an empty set, return to the cloud platform and fog manufacturing control mode; if not, output the code and the best fitness value; the execution process is controlled and input into the cloud platform and fog manufacturing control mode.

[0207] The encoding and its update mechanism are as follows: Since the CPS manufacturing component service composition dynamic reconstruction model is reconstructed under multiple constraints, the encoding method needs to determine the service selection scheme after the first CPS manufacturing component service abnormal point, the manufacturing resource strong coupling constraint information, the original CPSMCSC execution path time constraint and the encoding reconstruction information. The specific encoding and its update method are as follows Figure 3 shown. Figure 3 For the coding update mechanism, Figure 3 In the example, the original CPS manufacturing component service combination execution path is [1 1 33 4 4 3 5]. During the execution process, the 3rd and 7th CPS manufacturing component services fail, and reconstruction needs to be started from the 3rd CPS manufacturing component service failure. Then, the code is randomly generated and constraint 1-2 is applied to adjust it.

[0208] Then, it is compared with the original code to obtain the code reconstruction information matrix RE. Then, under the constraints of constraints 1-2, the corresponding HHO search mechanism is applied to update the position coding information, and the cost matrix C and service quality matrix S are updated according to RE.

[0209] The optimization strategy is as follows: Since the CPS manufacturing component service composition dynamic reconstruction model considers service anomaly constraints (constraint 1), manufacturing resource strong coupling constraints (constraint 2), the original CPSMCSC execution path mandatory time constraints (constraint 3) and manufacturing service occupancy time constraints (constraint 4), the designed IHHO algorithm must adopt corresponding strategies to deal with them. Constraints 1-2 mainly act on the encoding process, and constraints 3-4 are used to screen effective solutions before solving the fitness of multiple populations.

[0210] (1) Strategy 1: Cluster optimization strategy

[0211] Like traditional meta-heuristic intelligent algorithms, the original HHO updates each solution in the population and calculates its fitness at the same time. However, since the service composition dynamic reconstruction model calculates fitness only after the population individuals satisfy constraints 1-4, it is not necessary to calculate the fitness of each solution, otherwise it will greatly reduce the solution efficiency. Therefore, the IHHO algorithm adopts a cluster optimization strategy, that is, after all population individuals have completed the update and constraints 1-4 are applied, the population fitness calculation is performed.

[0212] (2) Strategy 2: Population diversity supplementation strategy

[0213] After applying the above cluster optimization strategy, the soft encirclement and hard encirclement search mechanisms of the fast dive strategy in the original HHO (Formula (24, 25, 28)) are applied to supplement the Harris hawk population, expand the search mechanism, and increase the diversity of feasible solutions.

[0214] like Figure 5 As shown, Figure 5 This is a schematic diagram of a Gantt chart for reconstructing and adjusting the service combination execution path when an exception occurs. The IHHO algorithm flow chart provided in this embodiment includes the following steps:

[0215] Step 1: When a service anomaly occurs during the CPSMCSC execution process, the original CPS manufacturing component service combination execution path and its encoding and other parameters are initialized according to Table 1;

[0216] Step 2: The hierarchical digital twin model finely perceives service fault information (constraint 1) and strongly coupled manufacturing resources (constraint 2);

[0217] Step 3: Set the original CPSMCSC execution path completion time (constraint 3) and the dynamic manufacturing service occupancy time (constraint 4) through the cloud platform under the fog manufacturing paradigm;

[0218] Step 4: Starting from the first abnormal CPS manufacturing component service, the encoding mechanism is applied to initialize the Harris Hawk population;

[0219] Step 5: Apply constraints 1-2 to adjust the Harris Hawk population coding, then adopt strategy 1 and apply formula (10-17) to calculate the fitness value;

[0220] Step 6: Obtain the current optimal position X by comparing the fitness between individuals in the population;

[0221] Step 7: Initialize or update (q, r, E0, l) according to the HHO parameter generation rule;

[0222] Step 8: According to the parameter values ​​(q, r, |E|, l), apply the corresponding search mechanism and code update mechanism to update the code;

[0223] Step 9: Apply strategy 2 to supplement the Harris Hawk population location and apply constraints 1-2 to constrain the control coding boundary;

[0224] Step 10: applying constraints 3-4 to the obtained Harris's Hawk population to screen the Harris's Hawk population;

[0225] Step 11: Update the cost C and service quality S according to the CPS manufacturing component service combination reconstruction information RE and formula (10-13);

[0226] Step 12: Calculate the fitness value of the Harris Hawk population and update the optimal fitness and its encoding information;

[0227] Step 13: Loop calculation. Repeat steps 7 to 12 until the number of iterations reaches the upper limit;

[0228] Step 14: Determine whether there is an optimal solution output at the end. If yes, output the optimal solution, otherwise request the cloud platform to update the CPS manufacturing component resource candidate service set and repeat steps 1 to 14.

[0229] Step 15: Perform refined perception control on the executable reconstructed CMSC execution process through the cloud platform and fog manufacturing control mode. If a service failure occurs, jump to step 1.

[0230] 3 Example verification

[0231] This embodiment uses a forging plant to complete a certain type of hook tail frame production order as a verification scenario. The model and algorithm are programmed using MATLAB 2018b and are performed under a 64-bit Windows 10 operating system with a 1.80GHz processor. In this process, the proposed IHHO is compared with the PSO and GWO optimization algorithms. The parameters of IHHO are set to: high and low limits: lb, ub; problem dimension: dim. The PSO parameters are set to: inertial damping ratio: w damp =0.99; inertia weight: w=1; global learning coefficient: c2=2.0; individual learning coefficient: c1=1.5. The parameters of GWO are set as follows: high and low limits: lb, ub; problem dimension: dim. In addition, the population size and number of iterations of all algorithms are: popsize=100, Max_iter=1000.

[0232] The core productivity of the plant is the 8000t press, and its mechanical processing part needs to be completed through the cloud manufacturing platform under the fog manufacturing paradigm, and the delivery time of the platform scheduling plan is used as the final delivery time. In addition, the processing process information of each part is refined and controlled through the fog manufacturing control mode. The processing task of the hook tail frame is decomposed into 8 subtasks: subtask 1 is to process the two outer sides of the frame; subtask 2 is to process the inner side of the tail; subtask 3 is to process the outer side of the connecting plate; subtask 4 is to process the two inner sides of the frame; subtask 5 is to process the transition between the upper and lower heads of the frame plate and the arc connection of the frame body; subtask 6 is the R13 fillet processing of the tail; subtask 7 is to process the triangular circular hole of the frame tail; subtask 8 is the processing of the hook tail pin hole.

[0233] The entire manufacturing task is described as MT = {C1, C5, 500, 104.2, {subtask 1, subtask 2, subtask 3, ..., subtask n}}. C1 is the starting position, C5 is the target position, 500 is the number of hook tail frames, and 104.2 (unit kg) is the mass of a single hook tail frame. Table 2-5 lists the logistics time and cost, manufacturing time and cost, candidate service set, and 25 CPS manufacturing component service samples involved in processing each subtask of this order. In summary, the relevant parameters of the startup algorithm can be obtained, as shown in Table 1.

[0234] Table 1 Parameters related to the service composition reconstruction process

[0235]

[0236]

[0237] Table 2 The range of processing time (hours) and manufacturing cost (yuan) for each process of 500 hook tail frames

[0238]

[0239] Table 3 Logistics time and cost between locations

[0240]

[0241] Table 4 Manufacturing cloud service information

[0242]

[0243] Table 5 Detailed information of cloud services in each candidate service cluster (manufacturing time (hours) / qualified rate / price ( / unit))

[0244]

[0245] In this embodiment, before reconstructing the original CPS manufacturing component service combination, some parameters need to be determined.

[0246] First, the original execution path of the CPS manufacturing component service combination is: Exroute = {1, 6, 8, 8, 14, 9, 18, 25}. The corresponding solution code is Osolution = [1 1 3 3 4 4 3 5]. The working time of each CPS manufacturing component service in the original CPSMCSC execution path is ExTime = [8 44 44 118 118 167 167 218 218 276 280 390 395 566 570628].

[0247] Therefore, constraint 3 is imposed on the service composition reconstruction model by the end time of the original CPSMCSC execution path (Tlimit=628). In this model, two adjacent CPS manufacturing components one-to-many coupling relationships MRCR and two abnormal CPS manufacturing component services EXCEPTIONS are randomly generated.

[0248] In order to verify the effect of the IHHO algorithm, a comparative test was conducted with the PSO and GWO algorithms. Based on the above-mentioned relevant parameters, each algorithm was independently run 30 times, and the test statistical results are shown in Table 6. Due to the randomness of the optimization algorithm, the Wilcoxon signed rank test with a significance level of 0.05 was used to perform a significance test on the IHHO, where the symbols ("+", "-", "=") respectively represent the relationship between the proposed IHHO and the comparison algorithm (significantly better than, significantly worse than, no significant difference).

[0249] Table 6 Comparison of fitness results of three algorithms in 30 experiments

[0250]

[0251] To better demonstrate the search process of each algorithm, take EXCEPTIONS = [3 4] as an example and draw the algorithm solution process as follows: Figure 4 As shown, Figure 4 The iterative solution process of the three algorithms is shown in Table 6. Figure 4 It can be seen that the optimal fitness value and average fitness value of IHHO are better than those of PSO and GWO. From the perspective of the algorithm's search mechanism, PSO mainly controls the entire search process through global learning factors, personal learning factors and adjustment of inertia. However, due to the large number of candidate services for related manufacturing tasks in the model and the existence of multiple practical constraints, the single update mode of position speed and the static inertia weight adjustment strategy make PSO prone to falling into local optimality and unable to effectively obtain the global optimal solution. GWO introduces alpha wolves of different social levels to make joint decisions and increase the diversity of solutions, making GWO slightly better than PSO in terms of solution effect. However, when solving the irregular reconstruction model of CPS manufacturing component service combination under multiple practical constraints, it still cannot escape the dilemma of falling into local optimality, such as Figure 4As shown. In contrast, based on the model characteristics of HHO and the population diversity supplement strategy, the IHHO algorithm has up to 6 different search mechanisms, which can greatly enhance the diversity of solutions under irregular constraints; in addition, the nonlinear mechanism conversion control strategy in HHO (randomly generated event probabilities q, r and escape energy E calculated by applying formula (29)) enables the algorithm to switch between the search mechanisms of its development process and exploration process in a balanced manner, thereby making the most of the characteristics of the HHO multi-search mechanism and effectively avoiding falling into local optimality in the later stage of the algorithm search, so as to obtain the optimal solution in the global scope. In summary, the IHHO algorithm is more suitable for efficiently solving the CPS manufacturing component service combination reconstruction model when service anomalies occur than PSO and GWO.

[0252] This example successfully applies the proposed model and algorithm to the cloud manufacturing platform deployed under the machining fog manufacturing paradigm. In the above hook tail frame machining production process, under the control of the fog manufacturing mode, the CPS manufacturing component service exception information EXCEPTIONS = [3 4] is perceived in real time, and then the machining cloud platform performs refined perception of the relevant actual constraints (manufacturing resource strong coupling constraint MRCR = {5[5]6[1 2 5]}, original CMSC completion time and manufacturing service occupancy time (Table 4)).

[0253] On this basis, the model algorithm is applied to reconstruct and optimize the original CPSMCSC execution path to obtain the optimal CPS manufacturing component service combination reconstruction execution path under the current constraints. Figure 5 Adjust the Gantt chart for service composition execution path reconstruction when exceptions occur, Figure 5 Intuitively demonstrate the above perceived practical constraints, Figure 5 In the paper, when the third and fourth CPS manufacturing component services are abnormal, the subsequent execution path is reconstructed and adjusted starting from the first manufacturing abnormal service point in the original CPS manufacturing component service combination coding. The reconstructed coding scheme is [1 1 1 5 4 4 5 4], and the optimization target is Q = 5.82, C = 20120, S = 50.6; then the reconstructed CPS manufacturing component service combination continues to be executed under the perception control of the cloud platform and the fog manufacturing management and control mode, ensuring the on-time and efficient completion of the order processing tasks and improving the robustness of the execution of the CPS manufacturing component service combination.

[0254] Example 2

[0255] This embodiment is based on the CPS manufacturing component service combination abnormal adaptive reconstruction method of the improved HHO algorithm to develop a system, embed the improved HHO model algorithm into the system, and realize the algorithm-driven self-organization reconstruction of the unmanned production line. The system includes a memory and a processor. The memory stores executable code. When the processor executes the executable code, the above method is implemented. When the Haier water heater tank model in the new order changes compared with the previous one, the current production line ( Figure 6 ) cannot meet production needs, and the production line is adaptively and quickly reconstructed. Figure 7 ; Figure 6 To reconstruct the front production line layout, Figure 7 This is the layout of the reconstructed production line.

[0256] The above-described embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. Equivalent substitutions or changes made by those skilled in the art based on the present invention are within the protection scope of the present invention. The protection scope of the present invention shall be subject to the claims.

Claims

1. An abnormal adaptive reconstruction method for CPS manufacturing component service combination based on the improved HHO algorithm, characterized by: The following steps are involved: Determine the CPS manufacturing component services based on the order tasks and construct the execution path and mandatory conditions of the service combination; Check whether there are abnormal CPS manufacturing components in the CPS manufacturing component service combination, and determine the first abnormal CPS manufacturing component service; According to the mandatory conditions and the constraints of the CPS manufacturing component service occupancy time, the subsequent CPS manufacturing component service combinations are constrainedly reconstructed starting from the first abnormal CPS manufacturing component service, and the execution path of the reconstructed service combination is obtained; Update the execution combination of the original service combination according to the execution path of the reconstructed service combination; The loop executes the check to see if there are any abnormal CPS manufacturing components until the order task is completed; The execution path of the reconstructed service composition is implemented based on the improved Harris Hawk model algorithm, and is specifically performed in the following steps: Construct a Harris Hawk population model and initialize the Harris Hawk population; The constraints in CPS manufacturing component services are used to control the population encoding boundary, and then the fitness value is calculated; Initialize the optimal position X and its fitness value; Initialize or update parameters in the Harris Hawk Flock Model; Adopt the corresponding HHO search mechanism according to the parameter value, and apply the code update mechanism to update the code; Set the Harris Hawk population location and apply constraints to control the population coding boundaries; According to the mandatory conditions and service occupancy time constraints, invalid solutions and conflicting solutions are eliminated; Calculate and update the cost C and service quality S based on the reconstruction information of the CPS manufacturing component service combination; Calculate the new population fitness value and update the optimal fitness and its encoding information; Circularly calculate the parameters and encoding information in the Harris Hawk Swarm model algorithm until the preset optimal solution is obtained and output; The encoding information is carried out according to the following steps: Determine the original code starting from the location of the first CPS manufacturing component service anomaly point; Generate random codes and adjust the random codes according to the constraints to obtain reconstructed codes; Then compare it with the original code to obtain the code reconstruction information matrix RE; According to the constraints, the corresponding HHO search mechanism is used to update the coding information to obtain the updated reconstructed coding; Update the cost matrix C and service quality matrix S according to the coded reconstruction information matrix RE; The constraints in the CPS manufacturing component service specifically include any one or any combination of the following: Constraint 1: The manufacturing service in the reconstructed service composition cannot be an abnormal CPS manufacturing component service; if the abnormal CPS manufacturing component service is a service with a strong coupling constraint, then it is necessary to combine constraint 2 to update its candidate service set; Constraint 2: CPS manufacturing component services with strong coupling constraints must be selected at the same time in the adjacent subtask candidate service set to be effective; in addition, one upstream coupled CPS manufacturing component service can correspond to multiple downstream coupled CPS manufacturing component services; Constraint 3: The delivery time of the original CPSMCSC execution path remains unchanged before and after the service composition reconstruction, that is, the completion time of the reconstructed service composition execution must be within the delivery time of the original CPSMCSC; Constraint 4: During the CPS manufacturing component service matching selection process, the CPS manufacturing component processing time must not conflict with its existing occupied time; The cost matrix C and the service quality matrix S are updated according to the encoding reconstruction information matrix RE, and are performed according to the following steps: When abnormal CPS manufacturing components are replaced, the cost, processing quality and service quality after reconstruction are calculated according to the following formula: minC=C M +C L +C Re ; Among them, C represents cost; Q represents processing quality; S represents service quality; The following method is used to standardize it: The negative attribute index C is quantified using the following formula: The indicators Q and S of positive attributes are quantified using the following formula: The optimization objective function is determined according to the following formula: minFitness=1 / (w C ×r c (X)+w Q ×r Q (X)+w S ×r S (X)); Among them, X is the candidate service set, w C , w Q , w S are the weights of cost, processing quality and service quality in the optimization target respectively; each weight is calculated by the hierarchical analysis method; The following optimization strategies are also included, as follows: Strategy 1: Cluster Optimization Strategy After all individuals in the population have completed the update and constraints 1-4 have been imposed, the population fitness calculation is performed; Strategy 2: Population diversity supplementation strategy After applying the above cluster optimization strategy, the soft encirclement and hard encirclement search mechanisms of the fast dive strategy in the original HHO were applied to supplement the Harris Hawk population; When calculating the minimum cost of the service combination in the improved Harris Hawk model algorithm, the following formula is used: minC=C M +C L ; Among them, C is the total cost, CM and CL represent the manufacturing cost and logistics cost respectively; c(Rs i ) means Rs i the manufacturing cost of performing the subtask; lc(Rs i ,Rs i+1 ) indicates that the service Rs i to Rs. i+1 logistics costs; When calculating the optimal processing quality in the improved Harris Eagle Group model algorithm, the following formula is used: Among them, Q represents the final processing quality of the entire product; PS(Rs i ) indicates that for this subtask, CPS Manufacturing Component Services Rs. i Achievable processing quality; When calculating the best service quality in the improved Harris Hawk model algorithm, the following formula is used: Among them, S represents the service evaluation of the whole process, SR (Rs i ) means Rs i service quality level.

2. A CPS manufacturing component service combination abnormal adaptive reconstruction system based on an improved HHO algorithm, including a memory and a processor, characterized in that: The memory stores executable codes, and when the processor executes the executable codes, the method described in claim 1 is implemented.

Images