Multi-chain block chain performance optimization method based on manufacturing service cooperation platform

Abstract

The invention discloses a multi-chain block chain performance optimization method based on a manufacturing service cooperation platform, and the method comprises the steps: carrying out the evaluation of the comprehensive capability of a platform participant based on the historical production data of the platform participant, and distributing a block chain organization and authority according to a score; performing grouping and preference analysis on the manufacturing services based on the historical service quality data of the manufacturing services; based on a generalized random Petri network, performing modeling analysis on each stage of service execution and log uplink recording; the model is simulated and evaluated based on discrete event simulation, and key performance parameters of the block chain are optimized through an improved optimization algorithm. According to the method, through automatic modeling, simulation, optimization and testing, the problems of correctly evaluating the capability of a participant and allocating organizations and permissions to the block chain in the aspect of block chain management are solved; the problem that the cost is too high due to the fact that the requirements of different manufacturing services for the block chain log performance are remarkably different, and the block chain performance becomes a bottleneck or far exceeds the requirements is solved.

Description

Technical Field

[0001] The present invention belongs to the fields of blockchain technology and operation management of manufacturing service collaboration platforms, and particularly relates to a multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform. Background Art

[0002] With the integration of digital technology and the manufacturing industry, the manufacturing industry is gradually developing towards intelligence and networking, and the collaboration in the manufacturing industry has gradually shifted from offline resource collaboration to service collaboration on a shared platform. As an important foundation for supporting the collaborative cooperation in the manufacturing industry, the manufacturing service platform plays an increasingly crucial role. Due to the large number of participants in the manufacturing service platform and inconsistent interests, it is necessary to establish mutual trust in aspects such as the personnel, finance, network systems, and order contracts of all parties. This kind of mutual trust usually relies on a centralized system for data storage and processing, but this model has problems such as difficult data sharing, insufficient credibility, and difficult to guarantee system security. Due to its characteristics of decentralization, immutability, and traceability, blockchain technology has gradually become an ideal technical solution to solve these problems.

[0003] In practical applications, how to optimize the performance of the blockchain, especially in the face of the ability differences of different manufacturing service participants and the quality differences of manufacturing services, is still a difficult problem to be solved urgently. The service execution process needs to store and read logs on the blockchain. The operating efficiency of the platform is not only related to the execution efficiency of the manufacturing service itself, but also closely related to system stages such as log recording, proposal endorsement, and message sorting on the chain. For services with a long execution time, a loose execution gap, and a small amount of log data, their log systems have no high requirements for latency and throughput, and the blockchain can be deployed on an edge industrial control computer to save costs; while for services that need to be executed quickly, have a tight execution gap, and a large amount of log data, their log systems have extremely high requirements for latency and throughput. If the performance does not meet the standard, it will cause delays in subsequent manufacturing links. Therefore, this kind of blockchain needs to be deployed on a public cloud or a private cloud, which requires a high cost. How to reasonably optimize the performance of each system link has become a key issue for improving the overall service quality of the platform and reducing the operating cost. The blockchains of existing manufacturing service collaboration platforms rarely design and optimize for different participants and different service qualities, cannot adapt to the needs of different accounting contents and accounting performances, and cannot be dynamically adjusted according to the actual service execution performance on the platform every quarter. Summary of the Invention

[0004] Aiming at the deficiencies of the existing methods, the present invention proposes a multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform, aiming to optimize the blockchain performance, comprehensively improve the system performance, and improve the service efficiency of the platform through automated modeling, simulation, optimization, and testing, so as to meet the needs of modern intelligent manufacturing.

[0005] The blockchain used in the present invention is based on Hyperledger Fabric and is a multi-chain architecture including several blockchains, where the blockchains are used to record the key logs of service execution. Based on the historical production data of the platform participants, the entropy weight method and TOPSIS scoring are used to evaluate the task execution ability, resource allocation ability, service performance ability, and quality consistency, etc., to score the comprehensive ability of the participants, determine the organizational allocation on the chain and the corresponding services; based on the historical service quality data of manufacturing services, an adaptive clustering algorithm based on evolution and dynamic splitting and merging is used for clustering, and the manufacturing services are divided into several groups to ensure that the services within the group have a relatively consistent service quality level, and on this basis, a blockchain is constructed for each group; based on the generalized stochastic Petri net, by analyzing the conditions and times of the occurrence of various stages of service execution and the corresponding log records on the blockchain under high concurrency in each service group, a system state transition model is established for subsequent system performance analysis; based on the above models, discrete event simulation is used to simulate the occurrence of events in each stage of the system to analyze the system performance under different key system parameters; based on the improved grey wolf optimization algorithm with a hierarchical hybrid strategy, by iterating the target system performance function under different performance parameters, the optimal blockchain performance parameters are obtained to reduce the blockchain deployment cost while ensuring efficient service execution.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform, comprising the following steps:

[0008] Step S110, based on the historical production data of the platform participants, score the comprehensive ability of the platform participants to determine the permissions and responsible services of the platform participants for the platform chain organization to which the enterprise belongs;

[0009] Step S120, based on the historical service quality data of manufacturing services, divide the manufacturing services into several service groups to ensure that services with similar service quality form a group as the optimization basis for the corresponding blockchain performance;

[0010] Step S130, based on the generalized stochastic Petri net, model the manufacturing service execution process and its service log packaging, transaction proposal, proposal endorsement, message sorting, and block submission stages in each service group, determine the conditions and time factors for the occurrence of its events, and establish a system state transition model for subsequent system performance analysis;

[0011] Step S140: Based on the system state transition model, discrete event simulation is used to simulate the state transitions of each stage of the system, analyze and determine the key performance indicators, establish the optimization objective function, adopt an improved optimization algorithm, iterate different parameter groups to obtain the optimal target performance parameters, and construct blockchains belonging to each service group according to the target performance parameters. By conducting performance tests on the blockchains under alternative deployment scenarios, select the plan with the lowest cost when the requirements are met.

[0012] The beneficial effects of the present invention compared with the prior art are as follows:

[0013] Aiming at the scenario where there are large gaps in the capabilities of participants and significant differences in the quality of manufacturing services in the existing manufacturing service platform, through automated modeling, simulation, optimization, and testing, the present invention solves the problem of how to correctly evaluate the capabilities of participants and allocate organizations and permissions for the blockchain in blockchain management; it also solves the problem that due to the significant differences in the blockchain log performance requirements of different manufacturing services, the blockchain performance becomes a bottleneck or far exceeds the requirements, resulting in excessively high costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a flowchart of a multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform of the present invention;

[0015] Figure 2 It is a flowchart of the improved grey wolf optimization algorithm proposed in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0017] As Figure 1 shown, a multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform of the present invention specifically includes the following steps:

[0018] Step S110: Based on the historical production data of platform participants, evaluate the comprehensive capabilities of platform participants to determine the permissions and responsible services of platform participants for the platform chain organization to which the enterprise belongs. Evaluating the comprehensive capabilities of platform participants includes: evaluating the task execution ability, resource allocation ability, service fulfillment ability, and quality consistency of platform participants using the entropy weight method and TOPSIS scoring. Divide the platform chain organization to which the enterprise belongs into a management organization and an operating organization. Determining the permissions and responsible services of platform participants for the platform chain organization to which the enterprise belongs includes: platform participants with high total scores are responsible for the management organization, and platform participants with high single scores are responsible for the operating organization. The former is responsible for message sorting, and the latter is responsible for transaction proposals and endorsement services;

[0019] Step S120: For manufacturing services, divide the manufacturing services into several service groups based on historical service quality data to ensure that services with similar service quality form a group, that is, the service quality within the service group has a similar level, as the optimization basis for the corresponding blockchain performance; among them, dividing the manufacturing services into several service groups includes: using an adaptive clustering algorithm based on evolution and dynamic splitting and merging to divide the manufacturing services into several service groups;

[0020] Step S130: Based on the generalized stochastic Petri net, model the execution process of manufacturing services within each service group, as well as the stages of service log packaging, transaction proposal, proposal endorsement, message sorting, and block submission, determine factors such as the conditions and times of event occurrence, and establish a system state transition model for subsequent system performance analysis;

[0021] Step S140: Based on the system state transition model, use discrete event simulation to simulate the state transitions of each stage, analyze and determine key performance indicators, establish an optimization objective function, use an improved optimization algorithm, iterate over different parameter groups to obtain the optimal target performance parameters, and construct a blockchain belonging to each service group according to the target performance parameters. This step may also include: performing performance tests on the blockchain under different deployment scenarios through an automated framework, and when the requirements are met, adopting the plan with the lowest cost. The improved optimization algorithm includes: using performance parameters related to service quality as known conditions, using the performance parameters of the blockchain log recording process as parameters to be optimized, and using an improved grey wolf optimization algorithm based on a hierarchical hybrid strategy.

[0022] Furthermore, in step S110, based on the historical production data of platform participants, evaluate the task execution ability, resource allocation ability, service fulfillment ability, and quality consistency, etc. using the entropy weight method and TOPSIS scoring. Evaluating the comprehensive capabilities of participants may include:

[0023] Step S110-1: For the statistically obtained historical production data, perform data preprocessing, such as deleting abnormal rows, deleting outliers, filling missing values, and normalizing. After data preprocessing, calculate the Chatterjee correlation coefficient for the statistically historical production data. For data with too high correlation, only keep one of them. For the remaining data, use a comprehensive scoring mechanism based on the entropy weight method and TOPSIS for evaluation, where the total score evaluation index is updated by combining single-factor sensitivity analysis, index elimination analysis, and random permutation analysis to ensure that the evaluation index for each period best conforms to the current data characteristics.

[0024] Step S110-2: Use TOPSIS to calculate the ranking for each score and the total score. Divide the on-chain organizations into management organizations and operating organizations. Among them, the management organization's permissions are given to the participants with good total score performance, responsible for message sorting services, while the operating organizations are selected from the participants with high single scores and rotated, responsible for transaction proposal and endorsement services.

[0025] Furthermore, step S120 includes:

[0026] Step S120-1: Based on the historical logs of manufacturing services provided by platform participants, statistically calculate the manufacturing service quality data. The manufacturing service quality data mainly includes: taking service function A as an example, the average time taken to complete the service , the maximum continuous usage time of the service desk (indicating that maintenance is required after exceeding this time) , the maximum repair time of the service desk (indicating that the service can resume normal execution) , the average service log packaging time , the average service log size ; These data are all obtained by taking the mean value after preprocessing the historical data in the historical logs of manufacturing services.

[0027] Step S120-2: Based on the manufacturing service quality data mentioned in step S120-1, design an adaptive clustering algorithm based on evolution and dynamic splitting and merging. This clustering algorithm combines the framework of the evolutionary algorithm on the basis of the traditional K-means algorithm. In the crossover, exchange the cluster labels to which the elements belong. In the mutation, select the cluster with the largest sample distance for further splitting, and at the same time select the clusters closest to the center to be combined into the same cluster, and try to perform local label fine-tuning on discrete data. The optimization objective can be set as multiple clustering indicators such as the silhouette coefficient or indicators related to the actual meaning of the solution.

[0028] Step S120-3: Cluster the manufacturing services based on the adaptive clustering algorithm mentioned in Step S120-2, and use the cluster center as the mean value of the service quality level of a service group to reflect service quality differences, such as different failure rates, different execution speeds, and different log sizes, etc. These differences often mean different devices and processes used in the manufacturing services. For each service group, construct a blockchain belonging to the group and use its service quality mean value as a known condition in subsequent modeling and optimization.

[0029] Furthermore, Step S130 includes:

[0030] Step S130-1: Perform log recording based on the Hyperledger Fabric blockchain, and use the generalized stochastic Petri net for modeling in combination with the manufacturing service execution process. The transaction process of the Hyperledger Fabric blockchain is mainly divided into four stages: transaction proposal, proposal endorsement, message ordering, and block submission. Each stage can be divided into 3 states: waiting, idle, and in execution after simplification. Regarding the state of the manufacturing service execution system, taking the service execution process as an example, its states are divided into service waiting, service in execution, service desk idle, service desk repair, and service desk failure. The system state can be modeled by the generalized stochastic Petri net. The overall process includes service execution, log packaging, transaction proposal, proposal endorsement, message ordering, and block submission.

[0031] Step S130-2: Based on the modeling method in Step S130-1, consider the situation where the services in a service group on the platform are executed and logged through the blockchain simultaneously, and conduct modeling analysis under the maximum log recording concurrency. This process can include: splitting the Petri net into a service execution subnet Petri-MS and a log recording subnet Petri-BC. Among them, Petri-MS is a subnet containing three parts: service execution, log packaging, and transaction proposal, while Petri-BC is a subnet containing parts of proposal endorsement, message ordering, and block submission. Connect multiple Petri-MS with Petri-BC to simulate the scenario where multiple services are executed concurrently and logged through the same blockchain. The rates and probabilities of system state transitions in Petri-MS can be calculated from the manufacturing service quality mean value in Step S120-3.

[0032] Furthermore, Step S140 includes:

[0033] Step S140-1: Simulate the system state transition model based on the discrete event simulation method. For the system state transition model obtained in step S130, record the moments when each token enters and leaves each place in the Petri-BC, and the average delay and total delay of each link can be calculated. In the system state transition model, each system state is a vector of the number of tokens in all places. An infinite number of tokens are placed in the initial place to simulate the production situation at the saturated steady state.

[0034] Step S140-2: Based on the discrete event simulation method mentioned in step S140-1, combined with the optimization objective function, substituting the average service quality level in step S120-3 as a condition, the following parameters can be optimized: the maximum number of messages in a block , the average delay of proposal endorsement , the average delay of message sorting and the average delay of message submission , the total delay.

[0035] Establish the optimization objective function (only retain the trend term) as follows:

[0036] (1)

[0037] where is the weight, and are the number of times and occurrence rate of proposal endorsement, and are the number of times and occurrence rate corresponding to the sorting and submission links; represents the ratio of the time a token stays in the subnet Petri-BC in step S130-2 to the time it takes to complete the entire Petri net, and there is , is the number of messages waiting at the message sorting at each moment, is the total simulation time, is the maximum number of messages in a block, is the total number of manufacturing services successfully executed and logged at the end of the simulation.

[0038] Step S140-3: Based on the system state transition model and discrete event simulation method mentioned in the previous steps, use the improved grey wolf optimization algorithm based on the hierarchical hybrid strategy to optimize the target performance parameters, and its process is as Figure 2As shown. This algorithm can effectively jump out of the local optimal solution, aiming at obtaining a better solution with fewer iteration times (i.e., fewer discrete event simulation times) as the target performance parameter, and at the same time, there is also a significant improvement in stability compared with the current similar algorithms. Based on the original Grey Wolf Optimization algorithm, this algorithm uses low-discrepancy sequences for initialization, sorts and stratifies according to the objective function values obtained by wolf individuals, and adopts different update strategies. Among them, the layer with the worst objective value is responsible for exploration, and the M-flight random walk mechanism is adopted. Among them, the wolf individuals explore in an oscillating manner on the tangent plane of the line connecting their positions with the alpha wolf's position and gradually approach the alpha wolf, and its local trajectory is in the shape of the English letter "M". Its update method is as follows:

[0039] (2)

[0040] Wherein, is a constant, represents the position of the wolf in the round of iteration, represents the vector from this wolf to the alpha wolf's position, is its tangential vector. is a modification quantity with the same dimension as , and half of its dimensions are randomly taken as 1, and the other half conforms to the bimodal distribution, as shown in the following formula:

[0041] (3)

[0042] Wherein, and are the probability density functions of the left peak and the right peak of the bimodal distribution respectively, is the function, , and are the Gamma functions of the corresponding parameters, is the random variable of the probability distribution, is the integral temporary variable, , , , , and are the shape parameters in the corresponding probability density functions respectively, and their values can be modified according to requirements.

[0043] Step S140-4, based on the optimal target performance parameters obtained in step (4.3), select multiple deployment plans from the elastic deployment alternative plans, test them using the Hyperledger Caliper framework, and select the plan with the lowest cost for deployment within a certain performance error range.

[0044] The process steps described above can be built into an automatic application module, which is automatically executed in each cycle to ensure that the platform blockchain always meets the requirements of the current cycle.

[0045] The content not described in detail in the specification of the present invention belongs to the prior art well-known to those skilled in the art.

[0046] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform, characterized in that: The steps include: Step S110, based on the historical production data of the platform participants, the comprehensive capabilities of the platform participants are scored to determine the authority and business responsibilities of the platform participants in the platform chain organization to which the enterprise belongs; Step S120, based on the historical service quality data of manufacturing services, the manufacturing services are divided into several service groups, ensuring that services with similar service quality constitute a group, which serves as a basis for optimizing the performance of the corresponding blockchain; Step S130, based on the generalized stochastic Petri net, model the manufacturing service execution process in each service group and its service log packaging, transaction proposal, proposal endorsement, message sorting and block submission stages, determine the conditions and time factors for the occurrence of its events, and establish a system state transition model for subsequent system performance analysis; Step S140, based on the system state transition model, discrete event simulation is used to simulate the state transition of each stage, key performance indicators are analyzed and determined, the optimization objective function is established, and an improved optimization algorithm is used to iterate different parameter groups to obtain the optimal target performance parameters. The blockchain belonging to each service group is constructed according to the target performance parameters, and the blockchain under the alternative deployment scheme is performance tested. When the requirements are met, the lowest cost scheme is adopted.

2. According to claim 1, a multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform is characterized in that: The step S110 includes: Step S110-1, perform data preprocessing on the various historical production data obtained by statistics, and calculate the Chatterjee correlation coefficient for the statistical historical production data. For data with too high correlation, only one is retained, and the remaining data is evaluated by a comprehensive scoring mechanism based on entropy weight method and TOPSIS, in which the total score index is updated by combining single factor sensitivity analysis, indicator elimination analysis and random shuffling analysis; Step S110-2, ranking the platform participants by using TOPSIS for each score and the total score.

3. The multi-chain blockchain performance optimization method based on the manufacturing service collaboration platform according to claim 1 is characterized in that: In step S110, the comprehensive capabilities of the platform participants are scored, including: the task execution capability, resource allocation capability, service fulfillment capability and quality consistency of the platform participants are evaluated using the entropy weight method and TOPSIS scoring.

4. The multi-chain blockchain performance optimization method based on the manufacturing service collaboration platform according to claim 1 is characterized in that: In step S110, the authority and business responsibilities of the platform participants to the platform chain organization to which the enterprise belongs are determined: platform participants with high total scores are responsible for managing the organization, and platform participants with high individual scores are responsible for operating the organization. The former are responsible for message sorting, and the latter are responsible for transaction proposals and endorsement services.

5. According to claim 1, a multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform is characterized in that: The step S120 includes: Step S120-1, collecting manufacturing service quality data based on the manufacturing service history logs provided by the platform participants; Step S120 - 2 , based on the manufacturing service quality data mentioned in step S120 - 1 , designing an adaptive clustering algorithm based on evolution and dynamic split-merge; Step S120-3, cluster the manufacturing services based on the adaptive clustering algorithm mentioned in step S120-2, and then use the cluster center as the service quality level average of a service group, build a service chain unique to each service group, and use its service quality average as a known condition in subsequent modeling and optimization.

6. According to claim 5, a multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform is characterized in that: In step S120-2, the adaptive clustering algorithm is based on the traditional K-means algorithm and combined with the framework of the evolutionary algorithm. The optimization goal of the evolution is set to multiple clustering indicators including the silhouette coefficient and indicators related to the practical significance of the solution.

7. According to claim 1, a multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform is characterized in that: The step S130 includes: Step S130-1, log recording based on Hyperledger Fabric blockchain, combined with the manufacturing service execution process, and modeling using generalized stochastic Petri nets; Step S130 - 2 , based on the modeling method in step S130 - 1 , considering the situation where services belonging to a service group in the platform are executed simultaneously and logged through a service chain, modeling analysis is performed under the maximum logging concurrency.

8. The multi-chain blockchain performance optimization method based on the manufacturing service collaboration platform according to claim 6 is characterized in that: Step S130-2 includes: splitting the Petri net into a service execution subnet Petri-MS and a logging subnet Petri-BC, wherein Petri-MS is a subnet including service execution, log packaging and transaction proposal, and Petri-BC is a subnet including proposal endorsement, message sorting and message submission, and connecting multiple Petri-MS with Petri-BC to simulate the scenario where multiple services are executed concurrently and logged through a blockchain.

9. The multi-chain blockchain performance optimization method based on a manufacturing service collaboration platform according to claim 1 is characterized in that: The step S140 includes: Step S140-1, based on the discrete event simulation method, simulate the system state transition model. For the system state transition model obtained in step S130, record the time when each token in Petri-BC enters and leaves each library, and calculate the average delay and total delay of each link; Step S140-2, based on the discrete event simulation method mentioned in step S140-1, combined with the optimization objective function, the service quality level average in step S120-3 is substituted as a condition to optimize the following parameters: the maximum number of messages packaged in a block, the average delay of proposal endorsement, the average delay of message sorting, the average delay of message submission, and the total delay; Step S140-3, based on the system state transition model and discrete event simulation method mentioned in the previous steps, an improved grey wolf optimization algorithm based on a hierarchical hybrid strategy is used to optimize the parameters to obtain a better solution as the target performance parameter; Step S140-4: Based on the target performance parameters obtained in step S140-3, multiple deployment schemes are selected from the existing elastic deployment alternatives, and the Hyperledger Caliper framework is used to test them, and the scheme with the lowest cost among the qualified schemes is selected for deployment.

10. The multi-chain blockchain performance optimization method based on the manufacturing service collaboration platform according to claim 1 is characterized in that: In step S140, the improved optimization algorithm includes: taking the performance parameters related to the quality of service as known conditions, taking the performance parameters of the blockchain logging process as parameters to be optimized, and adopting an improved grey wolf optimization algorithm based on a hierarchical hybrid strategy.

Images