CN115032952A - Marine weight-related part processing production workshop integrated scheduling differential evolution control method, device and system based on differentiation transfer strategy - Google Patents

Marine weight-related part processing production workshop integrated scheduling differential evolution control method, device and system based on differentiation transfer strategy Download PDF

Info

Publication number
CN115032952A
CN115032952A CN202210520668.3A CN202210520668A CN115032952A CN 115032952 A CN115032952 A CN 115032952A CN 202210520668 A CN202210520668 A CN 202210520668A CN 115032952 A CN115032952 A CN 115032952A
Authority
CN
China
Prior art keywords
cnc
rgv
population
individual
code
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
CN202210520668.3A
Other languages
Chinese (zh)
Inventor
汤可宗
姜元昊
陈紫薇
李芳�
高尚
段梓怡
钱煜
钱宇轩
陈小斌
危雄飞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jingdezhen Hanyuan Technology Development Co ltd
Jiangsu University of Science and Technology
Jingdezhen Ceramic Institute
Original Assignee
Jingdezhen Hanyuan Technology Development Co ltd
Jiangsu University of Science and Technology
Jingdezhen Ceramic Institute
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jingdezhen Hanyuan Technology Development Co ltd, Jiangsu University of Science and Technology, Jingdezhen Ceramic Institute filed Critical Jingdezhen Hanyuan Technology Development Co ltd
Priority to CN202210520668.3A priority Critical patent/CN115032952A/en
Publication of CN115032952A publication Critical patent/CN115032952A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00Programme-control systems
    • G05B19/02Programme-control systems electric
    • G05B19/418Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
    • G05B19/41865Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by job scheduling, process planning, material flow
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/12Computing arrangements based on biological models using genetic models
    • G06N3/126Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00Program-control systems
    • G05B2219/30Nc systems
    • G05B2219/32Operator till task planning
    • G05B2219/32252Scheduling production, machining, job shop
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Biology (AREA)
  • General Physics & Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Automation & Control Theory (AREA)
  • Data Mining & Analysis (AREA)
  • Physiology (AREA)
  • Genetics & Genomics (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Quality & Reliability (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Manufacturing & Machinery (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention belongs to the technical field of ship and marine equipment manufacturing, and discloses a differential evolution control method, a device and a system for integrated dispatching and differential evolution of a ship critical part processing and production workshop based on a differentiation transfer strategy, wherein the differential evolution method (DE-TS) based on the differentiation transfer strategy comprises the following steps: accelerating the convergence speed of the algorithm by using a material transfer sequence generation mechanism based on the attenuation factor; dividing the population into a light injury population, a heavy injury population and a death population according to the fitness, and respectively undertaking the tasks of storing an optimal individual set, maintaining population diversity and eliminating poorer individuals; when the individual in the serious injury population executes the differentiation transfer strategy, the individual is regenerated on the basis of storing a part of key information so as to strengthen the searching performance. Data in the examples show that: compared with a plurality of comparison algorithms such as GA, PSO and the like, the DE-TS described by the invention has better optimizing capability and can be effectively applied to the integrated scheduling control process of a ship critical part processing production workshop.

Description

Marine weight-related part processing production workshop integrated scheduling differential evolution control method, device and system based on differentiation transfer strategy
Technical Field
The invention relates to the technical field of manufacturing of ships and marine engineering equipment, in particular to a method, a device and a system for integrated dispatching differential evolution control of a ship critical part processing production workshop based on a differentiation transfer strategy.
Background
Currently, the current state of the art commonly used in the industry is such that:
with the high-speed development of the fields of digital twinning, electromechanical integration and the like, integration, unmanned and intellectualization become the main development direction of the production and processing system of the marine critical parts. To achieve this goal, Rail-Guided Vehicle (RGV) devices are widely used, and are used to replace manpower to perform material handling operations such as loading and unloading, material handling, and material cleaning, so as to achieve fine control of material circulation state (Zhang X, Wang z. Dynamic Scheduling Model of Intelligent Rail-Guided Vehicles Based on Dynamic Programming [ C ].2019International Conference on Intelligent Computing, Automation and Systems (ICICAS),2019: 11-15.). The instability existing in the material circulation process under the manual operation is reduced, and the further development of the digital twinning technology provides a base support. In recent years, application of integrated scheduling control of high-precision computer numerical control machines by using RGV equipment has become very common (Worutating, Yuyitong RGV intelligent scheduling strategy based on dynamic programming algorithm [ J ]. industrial technology innovation, 2018,05(06): 44-47.). In this operation scenario, the RGV device performs integrated control on a Computer Numerical Control (CNC) machine, so that the CNC machine can stably and efficiently complete a predetermined processing task under a proper material supply condition, which is called as an rdsp (RGV dynamic scheduling protocol) problem. The research of the RDSP problem considers the supply and circulation process of materials, converts the workshop processing problem into the scheduling control problem of the RGV, and is more closely related to the real processing scene of the ship weight-related parts.
However, in terms of cost, the marine heavy-part workshop cannot be stopped in a short time after being started up once, and the continuous processing time is generally long, which causes the dimension of the RGV coding to be too high, so that the RDSP problem is converted into a high-dimensional optimization problem, and the optimization difficulty is high. On the other hand, since each interaction between the CNC and the RGV causes a complex change of the plant logistics state, this phenomenon is reflected in the RDSP problem that a series of material processing operations occurring first can have a continuous and complex effect on the material processing operations occurring later, that is, a previous dimension value in the RGV encoding can have an unpredictable effect on the selection of a subsequent dimension value, which makes it difficult for an optimization algorithm for implementing a high-dimensional problem by a decision variable clustering method to achieve a better effect on the RDSP problem (Ma X, Liu F, Qi Y, et al.a. objective based on utilization of multiple optimization schemes with multiple variables [ J ]. IEEE Transactions evaluation algorithm, 275 (2): 298). In addition, before determining the RGV code, the type of machining tool installed by each CNC in the workshop needs to be determined, which is associated with the type of process processed by the CNC, and this operation is the determination process of the CNC code. If the CNC code is not properly set, the production efficiency of the whole processing workshop is affected, and the CNC code is difficult to change again before the ship-used heavy-part closing workshop is stopped. Many existing evolutionary optimization algorithms are prone to fall into local extremum solutions on the RDSP problem, so that a ship heavy part workshop faces the bottleneck problem that the output efficiency is not high.
For example, in the workshop operation scene of RGV, H.Fan et al have designed an Improved Genetic Algorithm (Improved Genetic Algorithm for Solving the RDSP, IGAR) based on a cyclic perturbation coding method, which achieves significant efficiency improvement compared to the conventional Genetic Algorithm (Fan H, Yu X, Shu D-Q, et al. Compared with the traditional genetic algorithm, the IGAR selects the initial population by adopting a greedy strategy and designs a more effective evaluation index aiming at the RDSP problem, so that the method has more excellent performance on the RDSP problem. However, under the condition of long duration processing time of the ship critical part workshop, the introduced new mechanism cannot well deal with the problem of high-dimensional optimization and is easy to fall into a local extreme value solution. Due to the fact that the solution space is huge, stagnation phenomena are prone to occurring in the later operation stage of the algorithm.
The Kraskar selection operator is added into the greedy algorithm by H.Xiao et al, and a set of dynamic scheduling simulation model (Xiao H.RGV dynamic scheduling model based on krusk algorithm [ J ]. IOP Conference Series: Materials Science and Engineering,2019, 612 (3)) for RGV material processing is designed, which provides a feasible scheme for path selection among CNC. The method combines the thought of the graph theory with the evolution calculation method, and the optimal path is quickly found under the scene that the RGV accesses each CNC only once. Thereafter, the machining of the remaining heavy parts can be completed by repeating this path. However, this method does not dynamically adjust the machining load levels of the different CNC's according to the time required for completion of the different processes, and the number of times each CNC is accessed by the RGV is the same. When the time difference required for finishing the machining among different processes is increased, the CNC idle time of the process with shorter completion time is increased, and the production efficiency of a heavy part workshop is also limited. Also, under the control of this method, there may be large differences in the load levels of different process CNC, which may result in parts of the machining job intensive CNC being more prone to failure.
Li, et al, calculated the weights using entropy weight method and expert ranking method, established a feasible intelligent scheduling model of RDSP problems (Li Y.scheduling analysis of organic mapping system based on combined weights [ J ]. IOP Conference Series: Materials Science and Engineering,2019,493.). In the processing process, when considering whether to carry out loading and unloading for a certain CNC, the method takes the distance between the CNC and the RGV, the time for carrying out loading and unloading operation for the CNC and the three indexes of the RGV deviating from the center after the loading and unloading operation for the CNC into consideration, and obtains the weight values of the RGV between the current position and different CNC through an entropy weight method and an expert sorting method, thereby determining the CNC for carrying out material processing operation in the next step. The method can effectively solve the problem that the load levels of CNC of different procedures are possibly greatly different. However, the expert ranking method used in the algorithm depends on the experience and knowledge level of the expert, and it is difficult to verify whether the weight combination is still applicable when the required processing materials are changed.
Ding et al designed a forward-looking stepping model for optimizing RGV scheduling schemes and introduced chaotic particle swarm optimization algorithm into the model, thereby proposing a hybrid scheduling control method that combines multi-step processing mechanism with evolutionary algorithm (Ding C, He H, Wang W, et al. However, the method does not consider the influence of the installation of the CNC tool on the machining efficiency of the heavy part workshop, and because the installation of the CNC tool determines the upper limit of the machining efficiency of the heavy part workshop, the method cannot automatically adjust the installation method of the CNC tool according to the machining parameters of machining materials, and the efficiency bottleneck under CNC coding is difficult to break through optimization of the algorithm.
Based on the above discussion, it is easy to find that, in the prior art, when solving the RDSP problem, a preset rule method, such as an FIFO scheduling policy method, a look-ahead stepping policy method, a greedy policy method, etc., or an evolutionary optimization method, such as a genetic algorithm, a chaotic particle swarm algorithm, an improvement method thereof, etc., is generally adopted, and meanwhile, a hybrid method combining the two methods also exists. Because the continuous processing time of the RGV workshop is very long and the problem scale of the RDSP series problem is very large, the direct searching and optimization are very difficult, and the optimization algorithm for realizing the high-dimensional problem by using the decision variable clustering method is also difficult to obtain a better effect on the RDSP problem. In summary, the main problems of the prior art are: optimization obstacles caused by high-dimensional decision variables are difficult to solve, the algorithm effect is closely related to the result of population initialization, and the method is easy to fall into a local optimal solution. The direct factor influencing the problem lies in that a simplified mode of appropriate decision variables is not found, so that the search space of the algorithm is reduced under the condition of not reducing the solution space as much as possible, and the search effect of the algorithm is improved. However, the improper processing method may deteriorate the machining efficiency of the workshop, for example, each CNC is accessed once to be used as a decision group, it is easy to optimize the optimal scheduling strategy in one decision group, and after that, the local optimal scheduling strategy is repeated to obtain good algorithm effect, which is also a common processing method. However, this method cannot accommodate the wide gap between the completion times required for the different processes. In this case, although the number of times of completion of each CNC is the same, there may be a large difference in the processing load between the CNC for processing different processes, resulting in a low overall efficiency of the heavy-duty plant.
On the other hand, the continuous machining time of the ship heavy-duty workshop is generally long due to cost considerations, and it is difficult to readjust the machining tool type of the CNC during the machining process. The improper CNC tool allocation method will disadvantageously limit the machining efficiency of the heavy-duty workshop. Many of the prior art omit the optimization of the CNC tool mounting method, so that CNC cannot adaptively determine the optimal procedure and its tool mounting for different critical part machining parameters.
In summary, the problems of the prior art are as follows:
(1) the pressure of large-scale optimization in the RDSP problem is not effectively alleviated. Optimization is difficult due to the very large scale of the RDSP problem. The current academic world also fails to research a universal and effective evolutionary optimization method of high-dimensional decision variables, and due to the complex interaction relationship among different dimensions of the decision variables, the optimization algorithm for realizing the high-dimensional problem by using a decision variable clustering method is difficult to obtain a better effect on the RDSP problem;
(2) the machining load levels between different CNC are not effectively balanced. The optimization method using all the CNC as a decision group has achieved certain effect, and although the number of times of completion of each CNC is the same, the processing may cause great difference in the machining load level among the CNC of different procedures. In order to balance the processing load level among different CNC machines, a dynamic decision group processing method capable of self-adaptively adjusting according to the processing parameters of the current processing materials is needed;
(3) a method for determining the closest matching CNC code based on the machining parameters is not provided. The machining process to be handled by each CNC needs to be predetermined before starting machining. In the machining process, it is difficult to change the machining process to be processed by the CNC. In order to determine the most matched CNC code according to the machining parameters, a proper CNC distribution ratio needs to be determined according to the completion time ratios of different working procedures, and then a proper installation method is further determined in the ratio, so that an effective CNC code is formed.
The difficulty and significance for solving the technical problems are as follows:
in order to effectively solve the technical problems, the invention designs a Differential Evolution Algorithm (DE-TS) Based on a differentiation transfer Strategy. In order to effectively relieve the pressure brought by large-scale optimization in the RDSP problem, an RGV material transfer sequence generation mechanism based on a decay factor is designed in a DE-TS algorithm, so that the convergence speed of the algorithm in the early stage is accelerated on the premise of not compressing a solution space as much as possible. In the process, according to the operation parameter of the algorithm of the attenuation factor, each individual in the population obtains a length value of the RGV code according to the probability, and when the length value is smaller than the expected value of the RGV code, the sequence is repeated to achieve the expected dimension value, thereby realizing the purpose of reducing the dimension; when this value is greater than or equal to the expected value for RGV encoding, the sequence is truncated to make it equal to the expected dimension value. The processing method can relieve the calculation pressure of high-dimensional optimization by combining part of individuals with the dimensionality reduction method of indefinite length circulation on the premise of not compressing a decoding space as much as possible, and meanwhile, the indefinite length circulation lengths distributed by different individuals are possibly different, so that conditions are provided for maintaining population diversity.
In DE-TS, in order to effectively balance the processing load level between different CNC, in the process of generating an indefinite length cycle, a cycle with the smallest difference in processing load level between CNC is obtained according to the processing parameters of the current processing material, which is called an atomic work cycle. In an atomic duty cycle, the CNC is allowed to be repeatedly accessed in certain situations to achieve the goal of minimizing the CNC total idle time. When each individual is assigned an indefinite length cycle, it is required that it must be an integer multiple of the atomic duty cycle. It should be noted that there is no strict limitation between atomic work cycles between indefinite long cycles, and the atomic work cycle is only used to assist in determining the dimension value of indefinite long cycle and the optimal access times of each CNC, and does not limit the access order of the CNC in indefinite long cycle.
Further, a method for determining the best-matching CNC code based on the machining parameters is also provided in DE-TS. Firstly, the number of CNC machines allocated to each process is determined according to the ratio of the completion time of different processes. At this time, since the number of the CNC must be an integer, and the ratio of the completion time may not perfectly match the number of the CNC, it is allowed to vary to both sides of the current ratio according to the probability, thereby obtaining the number of the CNC allocated to each process whose result is an integer. And then, randomly initializing CNC codes according to the CNC numbers distributed in different procedures, and splicing the CNC codes with the RGV codes to obtain decision vectors, thereby realizing the self-adaptive optimization of the CNC codes.
Disclosure of Invention
In order to solve the technical problems, the invention provides a method, a device and a system for integrated dispatching differential evolution control of a marine heavy part processing production workshop based on a differentiation transfer strategy.
The invention provides a differential evolution control method for integrated scheduling of a marine heavy part processing and producing workshop based on a differentiation and transfer strategy, which comprises the following steps:
s1: and obtaining the juvenile population through random initialization according to the quantity information of the CNC of the computer numerical control machine tool contained in the ship weight-related part processing production workshop. In the population, each individual in the juvenile period only stores CNC code information, and the CNC code is used for storing tool type information of the computer numerical control machine tool. In CNC coding, the machining task category which can be processed by each CNC is represented by a digit;
s2: calculating the expected dimension value E of the RGV code of each individual in the population according to the processing parameter information of the current processing material and the material processing operation parameter information of the rail shuttle trolley RGV RGV . Unlike juvenile individuals, each mature individual is composed of two parts, CNC coding and RGV coding. At this time, the RGV encoding of each individual is randomly initialized so that the length of the RGV encoding of each individual is equal to E RGV . At this time, the juvenile population is changed to the mature population. And storing a loading and unloading task list of the RGV in the RGV code, wherein each loading and unloading task is identified by a number of a corresponding CNC. Transferring the obtained maturity population as a parent population to step S3;
s3: and obtaining the filial population by the transmitted parent population through a standard Differential Evolution (DE) algorithm. Then, each individual in the offspring population generates a new RGV code under the influence of the attenuation factor mu, and replaces the original RGV code of the individual with the new RGV code;
s4: checking the length of the RGV code of each individual in the offspring population when the length of the RGV code of an individual exceeds the expected dimension value E of the RGV code RGV Is truncated from the first bit so that the length of the RGV encoding is exactly equal to E RGV (ii) a When the length of RGV code of a certain body is less than the expected dimension value E of RGV code RGV Repeatedly filling the RGV code so that the length of the RGV code is just equal to E RGV
S5: and combining the parent population and the child population, and calculating the fitness value according to the CNC coding and RGV coding information carried by each individual. Then, in the combined population, carrying out internal sequencing according to the fitness value of each individual from high to low;
s6: and for the population sequenced according to the fitness value, dividing the population into three sub-populations of a light injury population, a heavy injury population and a death population according to the fitness value from high to low, wherein the ratio of the number of individuals in the three sub-populations is 1-alpha: α:1, where α is a parameter called ambient pressure;
s7: a differentiation transfer strategy was performed for the heavily wounded population. In this process, each individual in the severely injured population will return to the juvenile stage from the mature stage and back to the mature stage again through a differentiation-metastasis strategy. Weighing the wounded population, and performing a differentiation transfer strategy to obtain a new population;
s8: combining the light injury population with the new population, and taking the combined population as a parent population of a new iteration;
s9: the number of times of evaluation of the linear function is consumed each time the fitness value of one individual is calculated. Comparing the consumed function evaluation times FE with a preset maximum function evaluation time maxFE: when FE < maxFE, jumping to step S3, and transferring the parent population obtained in step S8 into step S3; when FE is more than or equal to maxFE, jumping to step S10;
s10: and drawing an image of which the population fitness value changes along with the function evaluation times, and storing the obtained CNC codes and RGV codes of the individuals with the highest fitness. The algorithm execution then ends and exits.
In order to achieve the effect of determining the most suitable RGV encoding length according to the current processing parameters, in the step S2, the expected dimension value E of the RGV encoding of each individual in the population is calculated according to the following formula RGV
Figure RE-GDA0003749433700000071
In the above formula, E (D) is the expected value of the individual dimension D, n CNC For the number of CNC machines in a manufacturing shop, D I And D The upper limit of the dimension is considered in the case of only the first step or the second step.
Figure RE-GDA0003749433700000072
In order to complete the time required for one-time process-material processing,
Figure RE-GDA0003749433700000073
the time required for finishing the processing of the second material in the first working procedure.
In order to achieve the effect of optimizing the fitness level of the population, in said step S3, a new RGV code is generated for each individual in the offspring population under the influence of the attenuation factor μ according to the following steps: step 3-1: the probability vectors are initialized. The value of the first dimension of the probability vector is assigned to 1, after which the value of each dimension is equal to the product of the value of the previous dimension and the attenuating factor μ. Continuously amplifying dimensionality for the probability vector according to the method, and turning to the step 3-2 until the value of a new dimensionality to be generated next time is lower than the set calculation precision;
step 3-2: all values in the probability vector are summed and the probability vector is divided by this sum, the resulting sum of all values in the probability vector being 1. The step realizes the normalization of the probability vector;
step 3-3: and selecting an individual in the offspring population which has not been subjected to the step, taking the value in each dimension in the probability vector as the probability value corresponding to the dimension, and selecting one dimension from the probability vector by using a standard roulette method. And assigning the number of the dimension in the probability vector to the number NL of the circulation rounds of the individual, and calculating the new RGV encoding length of the individual according to the following formula:
Figure RE-GDA0003749433700000081
in the above formula, length (newRGVCode) is the length of new RGV code, AWC is the length of the current individual's atomic duty cycle, Tool (I) The number of CNC's for the machining process one, Tool, specified in the current individual's CNC code (Ⅱ) The number of CNC used for the second machining process is specified in the current individual CNC code; step 3-4: extracting the previous length (in the original RGV coding of the current descendant individual)newRGVCode) bit and performs a repair operation on the extracted portion of the RGV encoding. Replacing the original RGV code of the individual with the repaired RGV code;
step 3-5: and if the child population has the individuals which have not performed the step 3-3, returning to the step 3-3, and if not, ending.
In order to achieve the effect that the machining load levels of the respective CNC are as consistent as possible, in said step 3-4, a repair operation is performed for the RGV encoding according to the following steps:
step 4-1: and counting the dimension value RD of the RGV encoding needing to execute the repairing operation, and generating an incremental matrix with the same dimension as the RGV encoding to be repaired. In the incremental matrix, the value stored in the first dimension is 1, and then the value stored in each dimension is equal to the value stored in the last dimension plus one;
step 4-2: and randomly disordering the sequence of all elements in the increasing matrix on the premise of keeping the dimension of the increasing matrix unchanged, thereby obtaining a disordered array A. Initializing the accessible times of each CNC, and setting the accessible times of each CNC for the first process to NL. Tool (Ⅱ) The accessible times of each CNC for the first process is set to NL. Tool (I) . Let the value of the temporary variable i be 1;
step 4-3: the CNC number stored in the a (i) th bit of the RGV encoding to be repaired is extracted as the CNC to be accessed. If the access times of the CNC to be accessed are positive values, subtracting one from the access times corresponding to the CNC to be accessed, otherwise, randomly selecting a CNC with the same process type as the CNC and the positive access times to replace the original CNC, and subtracting one from the access times corresponding to the replaced CNC; step 4-4: let the temporary variable i add 1. And if the accessible times of all the CNC are 0, ending, otherwise, returning to the step 4-3.
For the purpose of facilitating the evaluation of each individual in the population, in step S5, the fitness value F is calculated for any individual according to the following formula:
Figure RE-GDA0003749433700000091
in the above formula, L real Installing CNC of corresponding category according to the CNC code carried by the individual, and executing feeding and discharging operation for the corresponding CNC according to the RGV code carried by the individual in sequence, and then actually outputting the material quantity in the maximum continuous operation time length T by the processing and production workshop. L is max Is the upper limit of the system output, t hi Denotes the time, t, required for the CNC system numbered i to complete a machining operation li And the time required by the RGV to complete one feeding and discharging operation on the CNC with the number i is shown. P is the set of all CNC constructs used to machine the first pass, and Q is the set of all CNC constructs used to machine the second pass.
In order to achieve the effects of maintaining the diversity of the population and avoiding the local extremum solution, in step S7, a differentiation and transfer strategy is performed for the seriously damaged population according to the following steps:
step 6-1: each individual in the severely injured population is returned from the mature stage to the juvenile stage. At this time, for each individual in the seriously damaged population, the RGV code is deleted, and only the CNC code is reserved;
step 6-2: and for each individual in the seriously damaged population, randomly disordering the installation positions of all the CNC in the processing production workshop on the premise of keeping the total number of the CNC in any processing procedure unchanged, and synchronously generating new CNC codes. Replacing the original CNC code with a new CNC code;
step 6-3: each individual in the severely injured population is returned to the mature stage from the juvenile stage. RGV codes were regenerated for each individual in the heavy injury population in the same manner as steps S3 and S4.
The invention also provides a differential evolution control device for integrated scheduling of the marine heavy part processing and production workshop based on the differentiation and transfer strategy, which comprises terminal equipment, wherein the terminal equipment adopts internet terminal equipment and comprises a processor and a computer readable storage medium, and the processor is used for realizing each instruction; the computer readable storage medium is used for storing a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the differential evolution control method for the integrated scheduling of the ship critical part processing production shop based on the differentiation transfer strategy according to any one of claims 1-6.
The invention also provides a differentiation transfer strategy-based ship weight processing production workshop integrated scheduling differential evolution control system, which comprises a computer-readable storage medium, wherein a plurality of instructions are stored, and the instructions are suitable for being loaded by a processor of a terminal device and executing the differentiation transfer strategy-based ship weight processing production workshop integrated scheduling differential evolution control method according to any one of claims 1-6.
Compared with the prior art, the method, the device and the system for integrated scheduling differential evolution control of the marine critical heavy part processing production workshop based on the differentiation transfer strategy have the following beneficial effects:
compared with the prior art, the DE-TS method has better overall optimization searching efficiency on the basis of the RGV scheduling control optimization problem in the production process of the ship weight-related part workshop, and can obtain better overall path effect by planning in the RGV optimization searching process.
The invention aims to realize the improvement of the production efficiency of the marine heavy part workshop through the optimization of the RGV scheduling control process. The main contributions of the present invention include:
1) in order to accelerate the convergence rate, the invention designs an RGV material transfer sequence generation mechanism based on the attenuation factor. Firstly, the expected length of the atomic working cycle is calculated according to the current processing parameters, and then a sequence with the random integral multiple length of the atomic working cycle is selected by a statistical means. Next, the sequence will be automatically subjected to repair operations so that the load levels of each CNC within the sequence are as close as possible with as few variations as possible. If the length of the generated sequence is shorter than the desired length of the desired RGV material transfer sequence, the sequence is repeated until the requirements are met. The method can accelerate the convergence speed in the early stage of the operation of the algorithm on the premise of not sacrificing the size of the solution space as much as possible.
2) In the iterative process of the population, the invention uses a differentiation transfer strategy to provide guidance for the evolution of the population. For this reason, all individuals are firstly sorted according to fitness values, a part of individuals with the worst fitness are directly eliminated in the natural selection process, a part of individuals with the best fitness survive and enter the next iteration, and the rest of individuals are considered to be injured in a non-lethal way, and therefore part of data is lost. At this point, these individuals enter the stage of differentiation and metastasis. On the premise of ensuring that the installation quantity of various cutters is not changed, the installation positions of the cutters of the individuals are randomly disturbed, and the RGV material transfer sequence is automatically regenerated. The strategy is helpful for maintaining the diversity of the population and strengthening the exploration capability of the algorithm for the solution space.
3) The invention also provides a method for calculating the mathematical expected value of the RGV material transfer sequence dimension, which prevents the potential problem caused by the mismatching of the preset sequence length with the dimension: too short a sequence may result in premature termination of the algorithm, while too long a sequence may waste computing resources.
The DE-TS algorithm designed by the invention carries out simulation experiments of continuous processing for 30 days in a heavy part workshop, and is repeatedly operated to take an average value so as to prevent the influence of contingency on simulation results. This simulation experiment contained 9 examples consisting of cartesian products of the 3 test questions RDSP1-3 with three different machining scenarios with CNC quantities of 8, 10 and 12. Among these 3 test problems in RDSP1-3 are derived from the RGV and CNC Machining Parameter Dataset (RCMPD) (Wang L, Mu Y, Gao H, et al. the research on internal RGV dynamic scheduling base on hybrid genetic algorithm [ J ]. Journal of Physics: Conference Series,2019,1311 (1)). All simulation experiments were performed on the same equipment equipped with an Intel i7-7600U @2.80GHz dual-core processor. The operation environment of the simulation experiment further comprises an operation memory with the size of 16G and a display card with the model number of Intel HD Graphics 620. The software platform on which the algorithm runs is Matlab 2020b, which also includes a PlatEMO component with version 3.3. In the DE-TS algorithm, the parameters are set to: the attenuation factor is 0.8 and the ambient pressure is 0.5. In addition, the population size of all algorithms during the experiment was set to 50. In order to ensure the fairness of the experiment, the maximum evaluation times of all the algorithms are set to be 500.
The results of the simulation experiments have been summarized in the following table. Wherein PSO (Eberhart R, Kennedy J.A new optimizer using a particulate design of the C. MHS'95.Proceedings of the six International Symposium on Micro Machine and Human Science,1995:39-43.) and GA (Holland J H.Adaptation in natural and scientific systems: an integer analysis with algorithms of biology, control, and integer to integer analysis [ M ] MIT, 1992.) are two classic algorithms, while the IGAR Algorithm proposed by H.Fan et al (Fan H, Yu X, Shu D-Q, inherent Algorithm for simulation V. is a modified version of the C. general Algorithm 012133. J.A. A. a. Both the Success-history based parameter adaptation for the temporal evolution (SHADE) and the DE-TS algorithm proposed herein are improved algorithms based on standard DE algorithms (Tanabe R, Fukunaga A. Success-history based parameter adaptation for the temporal evolution [ C ].2013IEEE contract on evolution, 2013: 71-78.).
TABLE 1 loss Rate loss and standard deviation thereof
Figure RE-GDA0003749433700000111
Figure RE-GDA0003749433700000121
N in the above table CNC For representing the number of CNC and D for representing the dimension of the order decision vector to be optimized in the current problem. Each row represents an embodiment where the best performing algorithm has been identified with a gray shading. According to the table, in 9 embodiments, the effects of the DE-TS algorithm are all optimal, which shows that the DE-TS algorithm designed by the invention has significant positive influence on the RDSP series problem.
Drawings
Fig. 1 is a flowchart of a method, an apparatus, and a system for integrated scheduling differential evolution control in a marine heavy object processing production workshop based on a differentiation and transfer strategy according to a preferred embodiment of the present invention.
Fig. 2 is a schematic encoding diagram of the DE-TS algorithm provided in the embodiment of the present invention.
Fig. 3 is a schematic diagram of a marine heavy part processing production workshop according to an embodiment of the present invention.
Fig. 4 is a flow chart of a marine heavy part processing method according to an embodiment of the present invention.
FIG. 5 is an algorithm evolution curve of DE-TS and DE-WTS under different CNC quantity scenarios provided by an embodiment of the present invention.
FIG. 6 is an algorithm evolution curve of DE-TS and DE-WTS under different test problems according to an embodiment of the present invention.
FIG. 7 is a loss distribution diagram of the contrast algorithm on the extended set according to the embodiment of the present invention.
In the figure: (a) loss distribution diagram of DE-TS on the expansion set; (b) loss distribution graph of SHADE on the expansion set; (c) loss distribution diagram of IGAR on the expansion set; (d) loss distribution graph of GA on the expansion set; (e) loss profile of PSO on the extended set.
FIG. 8 is a distribution diagram of the loss difference between DE-TS and IGAR in the extended set according to an embodiment of the present invention.
Detailed Description
The invention is further described with reference to the following figures and embodiments.
Please refer to fig. 1, fig. 2, fig. 3, fig. 4, fig. 5, fig. 6, fig. 7 and fig. 8 in combination.
As shown in FIG. 1, a method for controlling differential evolution of integrated dispatching in a ship-used heavy-part processing production workshop based on a differentiation transfer strategy comprises the following steps S1 to S10:
s1: and obtaining the juvenile population through random initialization according to the quantity information of the CNC of the computer numerical control machine tool contained in the ship weight-related part processing production workshop. In the population, each individual in the juvenile period only stores CNC code information, and the CNC code is used for storing tool type information of the computer numerical control machine tool. In CNC coding, the machining task category which can be processed by each CNC is represented by a digit;
s2: according to the current processed objectProcessing parameter information of the material and material processing operation parameter information of the rail shuttle trolley RGV are calculated, and an expected dimension value E of the RGV code of each individual in the population is calculated RGV . Unlike juvenile individuals, each mature individual is composed of two parts, CNC coding and RGV coding. At this time, the RGV encoding of each individual is randomly initialized so that the length of the RGV encoding of each individual is equal to E RGV . At this time, the juvenile population is converted to the mature population. And storing a loading and unloading task list of the RGV in the RGV code, wherein each loading and unloading task is identified by a number of a corresponding CNC. The obtained maturity population is transferred to step S3 as a parent population.
Further, in order to achieve the effect of determining the most suitable RGV encoding length according to the current processing parameters, in step S2, a desired dimension value E of the RGV encoding of each individual in the population is calculated according to the following formula RGV
Figure RE-GDA0003749433700000131
In the above formula, E (D) is the expected value of the individual dimension D, n CNC For the number of CNC machines in a manufacturing shop, D I And D The upper limit of the dimension is considered in the case of only the first step or the second step.
Figure RE-GDA0003749433700000132
In order to complete the time required for one-step material processing,
Figure RE-GDA0003749433700000133
the time required for finishing the material processing of the first procedure and the second procedure;
s3: and obtaining the filial population by the transmitted parent population through a standard Differential Evolution (DE) algorithm. Each individual in the offspring population then generates a new RGV code under the influence of the attenuating factor μ and replaces the original RGV code of that individual with the new RGV code.
Further, to achieve the effect of optimizing the fitness level of the population, in step S3, a new RGV code is generated for each individual in the progeny population under the influence of the wilting factor μ according to steps 3-1 to 3-5:
step 3-1: a probability vector is initialized. The value of the first dimension of the probability vector is assigned to 1, after which the value of each dimension is equal to the product of the value of the previous dimension and the attenuation factor μ. Continuously amplifying dimensionality for the probability vector according to the method, and turning to the step 3-2 until the value of a new dimensionality to be generated next time is lower than the set calculation precision;
step 3-2: all values in the probability vector are summed and the probability vector is divided by this sum, the resulting sum of all values in the probability vector being 1. The step realizes the normalization of the probability vector;
step 3-3: and selecting an individual in the offspring population which has not been subjected to the step, taking the value in each dimension in the probability vector as the probability value corresponding to the dimension, and selecting one dimension from the probability vector by using a standard roulette method. And assigning the number of the dimension in the probability vector to the number NL of the circulation rounds of the individual, and calculating the new RGV encoding length of the individual according to the following formula:
Figure RE-GDA0003749433700000141
in the above formula, length (newRGVCode) is the length of new RGV code, AWC is the length of the current individual's atomic duty cycle, Tool (I) The number of CNC's for the machining process one, Tool, specified in the current individual's CNC code (Ⅱ) The number of the CNC used for the second machining procedure specified in the current individual CNC code;
step 3-4: the first length (newRGVCode) bit in the original RGV code of the current child individual is extracted, and the repair operation is performed on the extracted part of the RGV code. Replacing the original RGV code of the individual with the repaired RGV code;
step 3-5: and if the child population has the individuals which have not performed the step 3-3, returning to the step 3-3, and if not, ending.
Further, in order to achieve the effect that the machining load levels of the respective CNC are as uniform as possible, in said step 3-4, a repair operation is performed for the RGV encoding according to steps 4-1 to 4-4:
step 4-1: and counting the dimension value RD of the RGV encoding needing to execute the repairing operation, and generating an incremental matrix with the same dimension as the RGV encoding to be repaired. In the incremental matrix, the value stored in the first dimension is 1, and then the value stored in each dimension is equal to the value stored in the last dimension plus one;
step 4-2: and randomly disturbing the sequence of all elements in the increasing matrix on the premise of keeping the dimensionality of the increasing matrix unchanged, thereby obtaining a disordered array A. Initializing the accessible times of each CNC, and setting the accessible times of each CNC for the first process to NL. Tool (Ⅱ) The accessible times of each CNC for the first process is set to NL. Tool (I) . Let the value of the temporary variable i be 1;
step 4-3: the CNC number stored in the a (i) th bit of the RGV encoding to be repaired is extracted as the CNC to be accessed. If the access times of the CNC to be accessed are positive values, subtracting one from the access times corresponding to the CNC to be accessed, otherwise, randomly selecting a CNC with the same process type as the CNC and the positive access times to replace the original CNC, and subtracting one from the access times corresponding to the replaced CNC; step 4-4: let the temporary variable i add 1. If the accessible times of all the CNC are 0, ending, otherwise, returning to the step 4-3;
s4: checking the length of the RGV code of each individual in the offspring population when the length of the RGV code of an individual exceeds the expected dimension value E of the RGV code RGV Is truncated from the first bit so that the length of the RGV encoding is exactly equal to E RGV (ii) a When the length of RGV code of a certain body is less than the desired dimension value E of RGV code RGV Repeatedly filling the RGV code so that the length of the RGV code is just equal to E RGV
S5: and combining the parent population and the child population, and calculating the fitness value according to the CNC coding and RGV coding information carried by each individual. Subsequently, in the combined population, the fitness values of each individual are internally sorted from high to low.
For the purpose of facilitating the evaluation of each individual in the population, in step S5, the fitness value F is calculated for any individual according to the following formula:
Figure RE-GDA0003749433700000151
in the above formula, L real Installing CNC of corresponding category according to the CNC code carried by the individual, and executing feeding and discharging operation for the corresponding CNC according to the RGV code carried by the individual in sequence, and then actually outputting the material quantity in the maximum continuous operation time length T by the processing and production workshop. L is max Is the upper limit of the system output, t hi Indicates the time, t, required for the CNC with the number i to complete one machining operation li The time required for the RGV to complete one feeding and discharging operation on the CNC with the number i is shown. P is the set of all CNC components used for the first process, and Q is the set of all CNC components used for the second process;
s6: and for the population sequenced according to the fitness value, dividing the population into three sub-populations of a light injury population, a heavy injury population and a death population according to the fitness value from high to low, wherein the ratio of the number of individuals in the three sub-populations is 1-alpha: α:1, where α is a parameter called ambient pressure;
s7: a differentiation transfer strategy was performed for the heavily wounded population. In this process, each individual in the severely injured population will return from maturity to juvenile stage and back again to maturity through a differentiation and transfer strategy. Weighing the wounded population, and performing a differentiation transfer strategy to obtain a new population.
In order to achieve the effects of maintaining the diversity of the population and avoiding the local extremum solution, in step S7, a differentiation and transfer strategy is performed for the seriously damaged population according to steps 6-1 to 6-3:
step 6-1: each individual in the severely injured population is returned from the mature stage to the juvenile stage. At this time, for each individual in the seriously damaged population, the RGV code is deleted, and only the CNC code is reserved;
step 6-2: and for each individual in the seriously damaged population, randomly disordering the installation positions of all the CNC in the processing production workshop on the premise of keeping the total number of the CNC in any processing procedure unchanged, and synchronously generating new CNC codes. Replacing the original CNC code with a new CNC code;
step 6-3: each individual in the severely injured population is returned to maturity from juvenile stage again. Regenerating the RGV codes for each individual in the severe injury population in the same manner as steps S3 and S4;
s8: combining the light injury population with the new population, and taking the combined population as a parent population of a new iteration; s9: the number of times of evaluation of the linear function is consumed each time the fitness value of one individual is calculated. Comparing the consumed function evaluation times FE with a preset maximum function evaluation time maxFE: when FE < maxFE, jumping to step S3, and transferring the parent population obtained in step S8 into step S3; when FE is more than or equal to maxFE, jumping to step S10;
s10: and drawing an image of which the population fitness value changes along with the function evaluation times, and storing the obtained CNC codes and RGV codes of individuals with the highest fitness. The algorithm execution then ends and exits.
According to another aspect of one or more embodiments of the present disclosure, there is also provided a device for controlling integrated scheduling differential evolution of a ship-used heavy-part processing production workshop based on a differentiation transfer strategy, including a terminal device, where the terminal device is an internet terminal device, and includes a processor and a computer-readable storage medium, and the processor is configured to implement instructions; the computer readable storage medium is used for storing a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the differential evolution control method for the integrated scheduling of the ship critical part processing production shop based on the differentiation transfer strategy according to any one of claims 1-6.
According to another aspect of one or more embodiments of the present disclosure, there is further provided a system for integrated dispatch differential evolution control of a ship weight process plant based on a differentiation transfer strategy, which is characterized by comprising a computer-readable storage medium, wherein a plurality of instructions are stored, and the instructions are adapted to be loaded by a processor of a terminal device and execute the method for integrated dispatch differential evolution control of a ship weight process plant based on a differentiation transfer strategy according to any one of claims 1 to 6.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product, and as such, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The invention is further described with reference to specific examples.
The DE-TS algorithm designed by the invention carries out a simulation experiment of continuous processing for 30 days in a marine heavy-duty workshop. In the DE-TS algorithm, the parameters are set to: the attenuation factor is 0.8 and the ambient pressure is 0.5. In addition, the population size of all algorithms during the experiment was set to 50. In order to ensure the fairness of the experiment, the maximum number of evaluations of all the algorithms is set to 500. RDSP1-3, the 3 test questions obtained from the RGV and CNC Machining Parameter Dataset (RCMPD) Dataset (Wang L, Mu Y, Gao H, et al. the research on inner vibration RGV dynamic scheduling based on hybrid genetic algorithm [ J ]. Journal of Physics: Conference Series,2019,1311 (1)). is used in the examples. All simulation experiments were performed on the same equipment equipped with an Intel i7-7600U @2.80GHz dual-core processor. The operation environment of the simulation experiment also comprises an operation memory with the size of 16G and a display card with the model number of Intel HD Graphics 620. The software platform on which the algorithm runs is Matlab 2020b, which also includes a PlatEMO component with version 3.3.
Fig. 1 is a flowchart of a method, an apparatus and a system for integrated scheduling differential evolution control in a ship critical part processing production workshop based on a differentiation and transfer strategy according to a preferred embodiment of the present invention. As shown in FIG. 1, the DE-TS algorithm designed by the present invention mainly comprises 10 steps S1-S10.
In step S1, each individual in the population formed by the DE-TS algorithm will be initialized with CNC coding, at which time the status of these individuals is defined as juvenile. Subsequently, these individuals will be initialized with RGV encoding in step S2, at which time the status of these individuals is defined as maturity.
For the convenience of understanding the steps S1 and S2, please refer to fig. 2, fig. 3, and fig. 4 in combination. Fig. 2 is a schematic encoding diagram of the DE-TS algorithm provided in the embodiment of the present invention. As shown in FIG. 2, in the DE-TS algorithm designed by the invention, the scheduling strategy of the ship heavy-part workshop is abstractly coded and stored by using the method shown in FIG. 2. Specifically, as shown in fig. 2, in the population iteration process of the DE-TS algorithm provided by the embodiment of the present invention, each individual in the population is a feasible solution, and each solution can be represented by a one-dimensional vector, and the one-dimensional vector is composed of two parts, namely CNC coding and RGV coding. Wherein in CNC coding, for the presence of n CNC For the ship weight-closing workshop of the CNC, use n CNC The CNC code of the bit represents the process type of each CNC process, and the 1 st bit in the CNC code is used for representing the CNC with the number 1, the 2 nd bit is used for representing the CNC with the number 2, and so on. In CNC coding, the number 0 is used to indicate that the CNC is used to process the first pass, and the number 1 is used to indicate that the CNC is used to process the second pass.
Fig. 3 is a schematic diagram of a marine heavy part processing production workshop according to an embodiment of the present invention. As shown in fig. 3, the ship heavy part machining and producing workshop comprises a rail-guided shuttle school bus RGV, a feeding conveyor belt, a discharging conveyor belt and a computer numerical control machine tool CNC, wherein the number of each CNC is formed by adding a symbol "#", for example, CNC #1 represents CNC with the number of 1. Accordingly, the length-8 CNC code "10100111" as shown in fig. 2 can be understood as: in a ship heavy parts machining shop with only 8 CNC machines, CNC #2, #4, #5 are used to process the first pass, while CNC #1, #3, #6, #7, #8 are used to process the second pass. Referring to fig. 3, the position relationship of the CNC with different numbers in this embodiment is shown in fig. 3.
FIG. 2 is a DE-T according to an embodiment of the present inventionThe coding diagram of the S algorithm, where a representation of the RGV coding is also given. The codes for processing the CNC of the first process and the second process are alternately stored in the RGV codes because each complete process of the ship heavy object processing production shop requires the two processes of the first process and the second process to be completed in sequence. Accordingly, the RGV encoding "02030407050802060501" as shown in fig. 2 can be understood as: in a ship heavy part closing machining production workshop, in the 1 st complete machining process, the RGV firstly accesses a CNC #2 to complete the machining of the 1 st heavy part closing procedure I, and then accesses a CNC #3 to complete the machining of the heavy part closing procedure II; in the 2 nd complete machining process, the RGV firstly accesses the CNC #4 to complete the machining of the 2 nd procedure I of the heavy part, and then accesses the CNC #7 to complete the machining of the procedure II of the heavy part; in the 3 rd complete machining process, the RGV firstly accesses the CNC #5 to complete the machining of the 3 rd heavy part in the first procedure, and then accesses the CNC #8 to complete the machining of the heavy part in the second procedure; in the 4 th complete machining process, the RGV firstly accesses the CNC #2 to complete the machining of the 3 rd heavy part in the first procedure, and then accesses the CNC #6 to complete the machining of the heavy part in the second procedure; during the 5 th full machining process, the RGV first accesses CNC #5 to complete the machining of the 5 th critical part, process one, and then accesses CNC #1 to complete the machining of the critical part, process two. It should be noted that the RGV encoding shown in FIG. 2 is only for the purpose of aiding understanding, and in fact the length of the RGV encoding is very long, e.g., the degree E of RGV encoding in the 9 embodiments listed in the present invention "loss Rate loss and its standard deviation" of Table 1 RGV Most remain in dimensions 30000 to 50000. The calculation method of the RGV encoding length value ERGV and its detailed formula have been given in claim 2 and will not be described herein.
Fig. 4 is a flowchart of a processing method of a marine heavy object according to an embodiment of the present invention, which includes a detailed description of 12 steps for a complete processing. For further explanation, please refer to fig. 2 and fig. 3 for understanding. Referring to CNC used for the first process as class I CNC and CNC used for the second process as class II CNC, in the RGV coding given in fig. 2, each complete process is described by the number of one class I CNC and the number of one class II CNC, for example, next, the first complete process part "0203" in the RGV coding in fig. 2 is exemplified. Referring to fig. 4, during a complete process, the RGV is first moved to the class I CNC to be processed, which in this embodiment is CNC #2, as shown in step 1 of fig. 4. The RGV then grabs a piece of raw batch with the A-side gripper and rotates the gripper 180, as shown in step 2 of FIG. 4. Next, the RGV grabs the semi-clinker (if any) on the CNC #2 processing table with an empty B-side robot gripper, as shown in fig. 4, step 3. The RGV is then immediately rotated 180 ° to swap the positions of the two side gripper arms, as shown in step 4 of fig. 4. After the rotation is complete, the RGV places the raw meal on the a-side gripper on the CNC #2 table, at which point the CNC #2 will automatically perform the process operation of procedure one for the raw meal, as shown in step 5 of fig. 4. The RGV will then move to the class II CNC to be processed, which in this embodiment is CNC #3, as shown in step 6 of figure 4. The RGV will grab the clinker (if any) on the CNC #3 processing table with an empty gripper as shown in step 7 of fig. 4 and immediately swap the positions of the two grippers by flipping the gripper 180 deg., as shown in step 8 of fig. 4. The RGV then places the semi-clinker held on the gripper on the CNC #3 processing table as shown in step 9 of fig. 4 and again swaps the positions of the two grippers by rotation as shown in step 10 of fig. 4. Finally, the RGV places the clinker held by the gripper in the grog wash tank of the RGV for a washing operation (if any), as shown in step 11 of fig. 4, and after washing is complete places the washed stock on a blanking conveyor, which will then be automatically taken off the marine heavy parts processing system, as shown in step 12 of fig. 4.
Next, in step S3, the incoming parent population will get the child population through the standard differential evolution DE algorithm, and then each individual in the child population generates a new RGV code under the influence of the attenuation factor μ and replaces the original RGV code of the individual with the new RGV code. For the sake of understanding, the embodiment in the case where the reduction factor μ is 0.8 and the calculation accuracy ∈ is 0.01 will be further described here. At this point, the probability vector is initialized, with the value of the first dimension of the probability vector assigned to 1, and then the value of each dimension is equal to the product of the value of the previous dimension and the attenuation factor μ. According to the method, dimensionality is continuously expanded for the probability vector until the value of a new dimensionality to be generated next time is lower than the set calculation precision. Since the 21 st power of the attenuation factor μ is larger than the calculation accuracy ∈ and the 22 nd power of the attenuation factor μ is just smaller than the calculation accuracy ∈, the dimension of the probability vector is 21 dimensions in this case. Table 2 shows the case of the probability vector in this embodiment, and each weight value in table 2 is reserved with 4 decimal places for convenience of illustration.
Table 2 probability vector table in the case where the attenuation factor μ is 0.8 and the calculation accuracy ∈ is 0.01
Figure RE-GDA0003749433700000201
The normalization of the probability vectors is performed next. And summing all values in the probability vector, and dividing the probability vector by the sum value, wherein the sum of all values in the probability vector obtained at the moment is 1, and the value obtained after each weight is normalized is the probability value of the corresponding dimension. Table 2 shows the results of the probability vector normalization in this embodiment, with the attenuation factor μ being 0.8 and the calculation accuracy ∈ being 0.01. For ease of illustration, each probability value in table 3 retains 2 decimal places.
Table 3 probability vector normalization table in the case where the attenuation factor μ is 0.8 and the calculation accuracy ∈ is 0.01
Figure RE-GDA0003749433700000202
Next, for each individual, a standard roulette method is used to select a dimension from the probability vector using the value in each dimension in the probability vector as a probability value corresponding to the dimension, and the number of dimensions of the dimension in the probability vector is assigned to the number NL of rounds of rotation of the individual. For example, in an embodiment where the reduction factor μ is 0.8 and the calculation accuracy ∈ is 0.01, if the random number obtained by a certain individual in the population by the standard roulette method is 0.3728, since the value is greater than the sum of the probabilities in the first 2 dimensions but less than the sum of the probabilities in the first 3 dimensions, the number NL of rounds of the certain individual is assigned to 3 by the standard roulette method. This step is repeated so that each individual obtains a value of the number of loop rounds NL corresponding thereto.
For an individual whose value of the number of rounds NL was obtained by the method described above, the new RGV-encoding length for that individual is calculated by the following formula:
Figure RE-GDA0003749433700000211
in the above formula, length (newRGVCode) is the length of new RGV code, AWC is the length of the current individual's atomic duty cycle, Tool (I) Tool, the number of CNC's for the first machining process specified in the current individual's CNC code (Ⅱ) The number of CNC's for the second machining process specified in the current individual CNC code. For example, for the embodiment shown in FIG. 2, the individual CNC code in this embodiment is "10100111", since the number of bits is 8, contains 3 "0" bits, represents 3 CNC's for the first machining step, contains 5 "1" bits, represents 5 CNC's for the second machining step, and at this time, corresponding n is CNC Is 8, Tool (I) Is 3, Tool (Ⅱ) Is 5. If the number NL of the individual cycle is 3, the length NL · AWC · (2 · Tool) of the new RGV code of the individual can be obtained (I) ·Tool (Ⅱ) ) =3*2*3*5=90。
Then, the first length (newrgvcode) bit in the original RGV code of the current child individual is extracted, and a repair operation is performed on the extracted portion of the RGV code. Replacing the original RGV encoding of the individual with the repaired portion of RGV encoding. Repair and replacement operations are performed once for each individual. Wherein, in the repairing operation, the following procedures are also included:
firstly, the dimension value RD of the RGV coding needing to execute the repairing operation is counted, and an incremental matrix with the same dimension as the RGV coding to be repaired is generated. In the incremental matrix, the value stored in the first dimension is 1, after which the value stored in each dimension is equal to the value stored in the last dimension plus one. And then, on the premise of keeping the dimensionality of the increasing matrix unchanged, randomly disordering the sequence of all elements in the increasing matrix, thereby obtaining a disordered array A. For ease of understanding, table 4 presents one case of the incremental matrix and its unordered array for an embodiment for the case where RD ═ 10.
Table 4 one case of an incremental matrix and its unordered array for the case of RD 10
Figure RE-GDA0003749433700000212
Next, the number of times each CNC can be accessed is initialized, and the number of times each CNC can be accessed for the first process is NL _ Tool (Ⅱ) The accessible times of each CNC for the first process is set to NL. Tool (I) And let the value of the temporary variable i be 1. For ease of understanding, table 5 presents an accessible times table for CNC codes as shown in fig. 2, and NL ═ 3.
Table 5 NL — 3 accessible times table for CNC
Figure RE-GDA0003749433700000221
Subsequently, a single CNC access fix operation is performed. Specifically, the CNC number stored in the a (i) th bit of the RGV code to be repaired is first extracted as the CNC to be accessed. And if the access times of the CNC to be accessed are positive values, subtracting one from the access times corresponding to the CNC to be accessed, otherwise, randomly selecting a CNC with the access times being positive values and the same process type as the CNC to be processed to replace the original CNC, and subtracting one from the access times corresponding to the replaced CNC. After completion, the temporary variable i is incremented by 1. And repeatedly executing the single CNC access repairing operation until the access times of all the CNC are 0. Upon completion of step S3, each individual will obtain the RGV encoding after being repaired.
Subsequently, in step S4, the length of the RGV code of each individual in the child population is subjected to a verification operation.Specifically, when the length of the RGV code of an individual exceeds the expected dimension value E of the RGV code RGV Is truncated from the first bit so that the length of the RGV encoding is exactly equal to E RGV (ii) a When the length of RGV code of a certain body is less than the expected dimension value E of RGV code RGV Repeatedly filling the RGV code so that the length of the RGV code is just equal to E RGV
Subsequently, in step S5, the parent population is merged with the child population, and the fitness value is calculated according to the CNC coding and RGV coding information carried by each individual. Subsequently, in the combined population, the fitness values of each individual are internally sorted from high to low. The populations sorted according to the fitness value are divided into three sub-populations of a light injury population, a heavy injury population and a death population according to the fitness value from high to low in step S6, and the ratio of the number of individuals in the three sub-populations is 1-alpha: α:1, where α is a parameter called ambient pressure. Wherein the fitness value F is calculated for any individual according to the following formula:
Figure RE-GDA0003749433700000222
in the above formula, L real Installing corresponding types of CNC according to the CNC codes carried by the individuals, and executing feeding and discharging operations for the corresponding CNC according to the RGV codes carried by the individuals in sequence, wherein the quantity of finished materials actually output by the processing and production workshop within the maximum continuous operation time T. L is max Is an upper limit value of the system output, t hi Indicates the time, t, required for the CNC with the number i to complete one machining operation li And the time required by the RGV to complete one feeding and discharging operation on the CNC with the number i is shown. P is the set of all CNC constructs used to machine the first pass, and Q is the set of all CNC constructs used to machine the second pass. For example, in one embodiment of the CNC code shown in fig. 2, the maximum continuous operation time is 30 days, i.e., T is 2592000 seconds, the time required for the CNC to complete the first and second processes is 560 seconds and 400 seconds respectively, and the amount of the material actually produced is 12173. Mechanical paw defaults due to RGVFacing the feeding conveyor belt side, therefore, the time required for the RGV to complete one feeding and discharging of CNC #1, #3, #5, #7 close to the feeding conveyor belt side is 28 seconds, and the time required for the RGV to complete one feeding and discharging of CNC #2, #4, #6, #8 close to the discharging conveyor belt side is 31 seconds. In this case, the yield upper limit value of each CNC
Figure RE-GDA0003749433700000231
As shown in table 6.
TABLE 6 yield Upper limit values for each CNC
Figure RE-GDA0003749433700000232
At this time, there are
Figure RE-GDA0003749433700000233
(6056+6056+6013+6056+6013)) ═ min (13178,30194) ═ 13178, at this time, when two digits after the decimal point are retained, the fitness value F ═ L of the individual can be calculated real /L max ×100%= 12173/13178*100%=92.37%。
Subsequently, a differentiation transfer strategy is performed for the heavily wounded population in step S7. In this process, each individual in the severely injured population will return from maturity to juvenile stage and back again to maturity through a differentiation and transfer strategy. Weighing the wounded population, and performing a differentiation transfer strategy to obtain a new population. Specifically, each individual in the severely injured population is first returned from the mature to juvenile stage. At this time, for each individual in the seriously damaged population, the RGV code is deleted, and only the CNC code is reserved; and then, randomly disordering the installation positions of all the CNC in the processing production workshop and synchronously generating new CNC codes for each individual in the seriously damaged population on the premise of keeping the total number of the CNC in any processing procedure unchanged. Replacing the original CNC code with a new CNC code; finally, each individual in the severely injured population is returned to the mature stage from the juvenile stage, and RGV codes are regenerated for each individual in the severely injured population in the same manner as in steps S3 and S4. Next, in step S8, the light injury population will be merged with the new population, and the merged population will be the parent population for a new iteration.
Subsequently, in step S9, the number of times of evaluation of the linear function is consumed each time the fitness value of one individual is calculated. Comparing the consumed function evaluation times FE with a preset maximum function evaluation time maxFE: when FE < maxFE, jumping to step S3, and transferring the parent population obtained in step S8 into step S3; when FE ≧ maxFE, the process proceeds to step S10. For example, when the maximum function evaluation number of times maxFE is set to 500, if the number of times FE of function evaluation that has been consumed is 200, so there is FE < maxFE, then go to step S3; if the number of function evaluation times FE that has been consumed is 500, and there is FE ≧ maxFE, the process proceeds to step S10.
Finally, in step S10, an image in which the population fitness value varies depending on the number of times of function evaluation is drawn, and the obtained CNC code and RGV code of the individual with the highest fitness are saved. The algorithm execution then ends and exits. In one embodiment, the images of population fitness values as a function of the number of evaluations are shown in fig. 5 and 6. Specifically, fig. 5 is an algorithm evolution curve of the DE-TS and the DE-WTS under the scenario of different CNC quantities, which is provided by the embodiment of the present invention. FIG. 5 uses circles and triangles to represent the two algorithms DE-TS and DE-WTS, respectively, and as the number of iterations increases, the algorithm of DE-WTS in any case shows a significantly lower lifting speed than DE-TS. Although the loss of the two algorithms is basically consistent in the initial situation, the DE-TS algorithm shows more excellent optimization capability in the iterative process and can realize quick convergence. Simulation experiments show that the differentiation transfer strategy can effectively help the algorithm to improve performance. Compared with DE-WTS, the performance of the CNC machining system is remarkably improved in a machining scene with different CNC quantities. FIG. 6 is an algorithm evolution curve of DE-TS and DE-WTS under different test problems according to an embodiment of the present invention. The words "P1", "P2", and "P3" in fig. 3 are used to refer to test questions RDSP1, RDSP2, and RDSP3, respectively. The loss values of all initial points in fig. 6 are closer compared to fig. 5, which illustrates just that the number of CNC has a certain influence on the formation of the initial population, while the different processing parameters have a smaller influence on the fitness of the initial population. Simulation experiment results show that after the influence of the CNC quantity on the results is eliminated, the performance of the DE-TS algorithm is still remarkably superior to that of the DE-WTS algorithm. Under different test problems, the system loss of the DE-TS is reduced by 45.80% on average compared with that of the DE-WTS algorithm, and the simulation experiment further verifies the superiority of the effect of the DE-TS algorithm. FIGS. 5 and 6 compare the effect of the DE-TS algorithm including the differentiation and metastasis strategy with the effect of the DE-WTS algorithm not including the differentiation and metastasis strategy, and the displayed simulation results verify the effectiveness of the differentiation and metastasis strategy designed in the proposed DE-TS algorithm.
Further, all algorithms have been subjected to extensive simulation experiments on an extensive set of standard experiments in order to explore the ultimate capabilities of the proposed algorithms. First, upper and lower limit values of the time for which the CNC completes the primary process-one machining and the secondary process-two machining are extracted from examples RDSP1, RDSP2, and RDSP3, and these data are used to constitute an experimental range of the extended simulation experiment. Subsequently, repeated experiments were performed on all test problems, with the number of CNC in each experiment taking the average of all possible values. Finally, the average of all the test results at each data point is used as the value of the corresponding data point, thereby obtaining fig. 7. The purpose of this part of the experiment is to eliminate the potential influence that a fixed test problem may have on the experimental result, and clearly show the distribution of the resolving power levels of different algorithms in the problem space.
FIG. 7 is a loss distribution diagram of the contrast algorithm on the extended set according to the embodiment of the present invention. In the figure: (a) loss distribution diagram of DE-TS on the expansion set; (b) loss profile of SHADE on extended set; (c) loss distribution diagram of IGAR on the expansion set; (d) loss distribution graph of GA on the expansion set; (e) loss profile of PSO on the extended set. FIG. 7 shows the performance distribution of the comparison algorithm on the extended set in the form of thermodynamic diagram, in which the lighter the color is, the higher the loss of the algorithm is and the worse the algorithm effect is; conversely, the darker the color, the lower the loss of the algorithm and the better the algorithm effect. As shown in fig. 7, loss distributions of the three algorithms, i.e., the shadow, the GA, and the PSO, are relatively similar and are less affected by the processing parameters. When the time required by the processing in the first step and the time required by the processing in the second step are both reduced rapidly, the material requirement in the processing system is increased rapidly, and meanwhile, the effects of the three algorithms are reduced to a certain extent. This phenomenon is reflected in the thermodynamic diagram as a lightening of the color of the upper left part of the image. Furthermore, the performance of the three algorithms over the entire extended set is significantly inferior to the other two algorithms. In fig. 7, (a) and (c) show the loss distribution of DE-TS and IGAR on the extended set, respectively, and it can be seen that the effect of DE-TS is significantly better than that of IGAR, and the more the right-lower approach of thermodynamic diagram, the more dispersed the logistics demand obtained by the processing plant, the better the algorithm effect.
Further, in order to analyze the difference between the effects of the two algorithms, namely DE-TS and IGAR, the loss difference between IGAR and DE-TS is shown in FIG. 8. FIG. 7 is a distribution diagram of the loss difference between DE-TS and IGAR in the extended set according to an embodiment of the present invention. It can be seen that the whole image is above the xoy plane, demonstrating that the algorithm effect of DE-TS is consistently better than that of IGAR. The gap between these two algorithms is not significant when logistics demand is relatively dense. However, when the stream demand is more sparse, although the effect of the two algorithms is obviously improved, the loss difference of the two algorithms is larger and larger, which shows that the DE-TS has better solution space exploration capability, and the influence of the DE-TS on the processing parameters is obviously lower than that of IGAR.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention.
The above description is only an embodiment of the present invention, and is not intended to limit the scope of the present invention, and all equivalent structures or equivalent processes performed by the present invention or directly or indirectly applied to other related technical fields are included in the scope of the present invention.

Claims (8)

1. A differential evolution control method for integrated scheduling of a marine heavy part processing and production workshop based on a differentiation and transfer strategy is characterized by comprising the following steps:
s1: and (4) obtaining juvenile population through random initialization according to the quantity information of CNC (computerized numerical control) machine tools in the marine heavy part processing production workshop. In the population, each individual in the juvenile period only stores CNC code information, and the CNC code is used for storing tool type information of the computer numerical control machine tool. In CNC coding, the machining task category which can be processed by each CNC is represented by a digit;
s2: calculating the expected dimension value of the RGV code of each individual in the population according to the processing parameter information of the current processing material and the material processing operation parameter information of the rail shuttle trolley RGVE RGV . Unlike juvenile individuals, each mature individual is composed of two parts, CNC coding and RGV coding. At this time, the RGV encoding of each individual is randomly initialized so that the length of the RGV encoding of each individual is equal toE RGV . At this time, the juvenile population is converted to the mature population. And storing a loading and unloading task list of the RGV in the RGV code, wherein each loading and unloading task is identified by a number of a corresponding CNC. Transferring the obtained maturity population as a parent population into step S3;
s3: and obtaining the filial population by the transmitted parent population through a standard Differential Evolution (DE) algorithm. Subsequently, each individual in the progeny population is at the debilitating factorμGenerating a new RGV code and replacing the original RGV code of the individual with the new RGV code;
s4: checking the length of the RGV code of each individual in the offspring population when the length of the RGV code of an individual exceeds the expected dimension value of the RGV codeE RGV When starting from the first bit, the length of the RGV code is just equal toE RGV (ii) a When the length of RGV code of a certain body is less than the desired dimension value of RGV codeE RGV The RGV encoding is repeatedly padded so that the length of the RGV encoding is exactly equal toE RGV
S5: and combining the parent population and the child population, and calculating the fitness value according to the CNC coding and RGV coding information carried by each individual. Then, in the combined population, carrying out internal sequencing according to the fitness value of each individual from high to low;
s6: for the population sorted according to the fitness value, the population is divided into three sub-populations of a light injury population, a heavy injury population and a death population from high to low according to the fitness value, and the proportion of the number of individuals in the three sub-populations is 1-αα:1, whereinαIs a parameter known as ambient pressure;
s7: a differentiation transfer strategy was performed for the heavily wounded population. In this process, each individual in the severely injured population will return from maturity to juvenile stage and back again to maturity through a differentiation and transfer strategy. Weighing the injured population, and performing a differentiation transfer strategy to obtain a new population;
s8: combining the light injury population with the new population, and taking the combined population as a parent population of a new iteration;
s9: the number of times of evaluation of the linear function is consumed each time the fitness value of one individual is calculated. Comparing the number of function evaluations that have been spentFEEvaluation times of preset maximum functionmaxFEThe size of (2): when in useFE<maxFEWhen the method is used, jumping to the step S3, and transferring the parent population obtained in the step S8 into the step S3; when in useFEmaxFEIf yes, jumping to step S10;
s10: and drawing an image of which the population fitness value changes along with the function evaluation times, and storing the obtained CNC codes and RGV codes of the individuals with the highest fitness. The algorithm execution then ends and exits.
2. The method for integrated dispatching differential evolution control of marine heavy object processing production workshop based on differentiation transfer strategy according to claim 1, wherein in the step S2, the expected dimension value of RGV code of each individual in the population is calculated according to the following formulaE RGV
Figure 39425DEST_PATH_IMAGE002
In the above formula, the first and second carbon atoms are,
Figure 888695DEST_PATH_IMAGE004
for individual dimension
Figure 707265DEST_PATH_IMAGE006
The expected value of (c) is,
Figure 89704DEST_PATH_IMAGE008
for the number of CNC machines in the manufacturing shop,
Figure 676806DEST_PATH_IMAGE010
and
Figure 59377DEST_PATH_IMAGE012
the upper limit of the dimension is considered in the case of only the first step or the second step.
Figure 990293DEST_PATH_IMAGE014
In order to complete the time required for one-time process-material processing,
Figure 584829DEST_PATH_IMAGE016
the time required for finishing the processing of the second material in the first working procedure.
3. The method for integrated dispatching differential evolution control in marine critical part processing production workshops based on differentiation transfer strategy according to claim 1, characterized in that in step S3, the weakening factors are determined according to the following stepsμGenerating a new RGV code for each individual in the population of progeny:
step 3-1: a probability vector is initialized. The value of the first dimension of the probability vector is assigned to 1, after which the value of each dimension equals the value of the previous dimension and the attenuation factorμThe product of (a). According to the method, the continuous probability direction is adoptedMeasuring the amplification dimensionality, and turning to the step 3-2 until the value of a new dimensionality to be generated next time is lower than the set calculation precision;
step 3-2: all values in the probability vector are summed and the probability vector is divided by this sum, the resulting sum of all values in the probability vector being 1. The step realizes the normalization of the probability vector;
step 3-3: and selecting an individual in the offspring population which has not been subjected to the step, taking the value in each dimension in the probability vector as the probability value corresponding to the dimension, and selecting one dimension from the probability vector by using a standard roulette method. The number of the dimensionality in the probability vector is assigned to the number of the circulation rounds of the individualNLCalculating the new RGV encoding length of the individual according to the following formula:
Figure 60940DEST_PATH_IMAGE018
in the above-mentioned formula, the compound has the following structure,
Figure 914496DEST_PATH_IMAGE020
for the length of the new RGV encoding,AWCis the length of the atomic duty cycle for the current individual,
Figure 947305DEST_PATH_IMAGE022
the number of CNC for process one specified in the current individual's CNC code,
Figure 835495DEST_PATH_IMAGE024
the number of the CNC used for the second machining procedure specified in the current individual CNC code;
step 3-4: extracting the pre-in RGV encoding of the current descendant individual script
Figure 13667DEST_PATH_IMAGE026
Bits and performs a repair operation on the extracted portion of the RGV encoding. Replacing the original RGV code of the individual with the repaired RGV code;
step 3-5: and if the child population has the individuals which have not performed the step 3-3, returning to the step 3-3, and if not, ending.
4. The method for integrated scheduling differential evolution control in a ship's heavy object processing and production workshop based on the differentiation and transfer strategy as claimed in claim 1, wherein in the steps 3-4, repair operation is performed for RGV coding according to the following steps:
step 4-1: counting dimension values of RGV encoding requiring repair operationsRDAn incremental matrix is generated that has the same dimension as the RGV encoding to be repaired. In the incremental matrix, the value stored in the first dimension is 1, and then the value stored in each dimension is equal to the value stored in the last dimension plus one;
step 4-2: and randomly disordering the sequence of all elements in the increasing matrix on the premise of keeping the dimension of the increasing matrix unchanged, thereby obtaining a disordered array A. Initializing the accessible times of each CNC, and setting the accessible times of each CNC for processing the first procedure as
Figure 136211DEST_PATH_IMAGE028
Setting the accessible times of each CNC used for processing the first procedure as
Figure 205667DEST_PATH_IMAGE030
. Let temporary variablesiHas a value of 1;
step 4-3: extracting the A (A) of the RGV code to be repairedi) The CNC number stored in the bit serves as the CNC to be accessed. If the access times of the CNC to be accessed are positive values, subtracting one from the access times corresponding to the CNC to be accessed, otherwise, randomly selecting a CNC with the same process type as the CNC and the positive access times to replace the original CNC, and subtracting one from the access times corresponding to the replaced CNC;
step 4-4: let temporary variablesiAnd adding 1. And if the accessible times of all the CNC are 0, ending, otherwise, returning to the step 4-3.
5. The method for integrated dispatching differential evolution control in marine heavy-duty part processing and production workshops based on differentiation and transfer strategies according to claim 1, wherein in step S5, the fitness value of any individual is calculated according to the following formulaF
Figure 761413DEST_PATH_IMAGE032
In the above formula, the first and second carbon atoms are,
Figure 392377DEST_PATH_IMAGE034
installing the CNC of the corresponding category according to the CNC code carried by the individual, and executing the feeding and discharging operation for the corresponding CNC according to the RGV code carried by the individual in sequence, wherein the processing and production workshop has the maximum continuous operation time lengthTThe amount of the finished product actually produced.
Figure 423787DEST_PATH_IMAGE036
In order to obtain an upper limit value of the system output,
Figure 782087DEST_PATH_IMAGE038
is indicated by the reference number
Figure 330356DEST_PATH_IMAGE040
The CNC of (1) completes the time required for one machining operation,
Figure 115909DEST_PATH_IMAGE042
indicates RGV pair number of
Figure 854189DEST_PATH_IMAGE044
The CNC finishes the time required by one feeding and discharging operation.PFor the set of all CNC formations used for the first pass of the machining process,Qis a set of all CNC used for the second pass of the process.
6. The method for integrated dispatching differential evolution control in a ship critical piece processing and production workshop based on the differentiation transfer strategy according to claim 1, wherein in the step S7, the differentiation transfer strategy is executed for the severe damage population according to the following steps:
step 6-1: each individual in the severely injured population is returned from maturity to juvenile stage. At this time, for each individual in the seriously damaged population, the RGV code is deleted, and only the CNC code is reserved;
step 6-2: and for each individual in the seriously damaged population, randomly disordering the installation positions of all the CNC in the processing production workshop on the premise of keeping the total number of the CNC in any processing procedure unchanged, and synchronously generating new CNC codes. Replacing the original CNC code with a new CNC code;
and 6-3: each individual in the severely injured population is returned to the mature stage from the juvenile stage. RGV codes were regenerated for each individual in the heavy injury population in the same manner as steps S3 and S4.
7. A marine heavy part processing production workshop integrated scheduling differential evolution control device based on a differentiation transfer strategy is characterized by comprising terminal equipment, wherein the terminal equipment adopts internet terminal equipment and comprises a processor and a computer readable storage medium, and the processor is used for realizing instructions; the computer readable storage medium is used for storing a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the differential evolution control method for the integrated scheduling of the ship critical part processing production shop based on the differentiation transfer strategy according to any one of claims 1-6.
8. A differentiation transfer strategy-based ship weight processing production workshop integrated scheduling differential evolution control system is characterized by comprising a computer-readable storage medium, wherein a plurality of instructions are stored, and the instructions are suitable for being loaded by a processor of a terminal device and executing the differentiation transfer strategy-based ship weight processing production workshop integrated scheduling differential evolution control method according to any one of claims 1-6.
CN202210520668.3A 2022-05-13 2022-05-13 Marine weight-related part processing production workshop integrated scheduling differential evolution control method, device and system based on differentiation transfer strategy Withdrawn CN115032952A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210520668.3A CN115032952A (en) 2022-05-13 2022-05-13 Marine weight-related part processing production workshop integrated scheduling differential evolution control method, device and system based on differentiation transfer strategy

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210520668.3A CN115032952A (en) 2022-05-13 2022-05-13 Marine weight-related part processing production workshop integrated scheduling differential evolution control method, device and system based on differentiation transfer strategy

Publications (1)

Publication Number Publication Date
CN115032952A true CN115032952A (en) 2022-09-09

Family

ID=83121105

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210520668.3A Withdrawn CN115032952A (en) 2022-05-13 2022-05-13 Marine weight-related part processing production workshop integrated scheduling differential evolution control method, device and system based on differentiation transfer strategy

Country Status (1)

Country Link
CN (1) CN115032952A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116307646A (en) * 2023-05-23 2023-06-23 科大智能物联技术股份有限公司 One-rail double-vehicle scheduling method based on two-stage dynamic partitioning algorithm

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116307646A (en) * 2023-05-23 2023-06-23 科大智能物联技术股份有限公司 One-rail double-vehicle scheduling method based on two-stage dynamic partitioning algorithm
CN116307646B (en) * 2023-05-23 2023-09-01 科大智能物联技术股份有限公司 One-rail double-vehicle scheduling method based on two-stage dynamic partitioning algorithm

Similar Documents

Publication Publication Date Title
CN110796355B (en) Flexible job shop scheduling method based on dynamic decoding mechanism
CN112286149A (en) Flexible workshop scheduling optimization method and system considering crane transportation process
CN112348314A (en) Distributed flexible workshop scheduling method and system with crane
CN112147960A (en) Optimized scheduling method and device for flexible manufacturing system
CN115032952A (en) Marine weight-related part processing production workshop integrated scheduling differential evolution control method, device and system based on differentiation transfer strategy
CN115130789A (en) Distributed manufacturing intelligent scheduling method based on improved wolf optimization algorithm
CN112286152A (en) Distributed flow shop scheduling method and system with batch delivery constraint
He et al. Energy-efficient open-shop scheduling with multiple automated guided vehicles and deteriorating jobs
CN111210125A (en) Multi-target workpiece batch scheduling method and device based on historical information guidance
CN112148446B (en) Evolutionary strategy method for multi-skill resource limited project scheduling
CN113050644A (en) AGV (automatic guided vehicle) scheduling method based on iterative greedy evolution
Keung et al. A genetic algorithm approach to the multiple machine tool selection problem
Jiang et al. A control system of rail-guided vehicle assisted by transdifferentiation strategy of lower organisms
CN116184941A (en) Multi-target method and system for fuzzy workshop scheduling
Zheng et al. Solving multi-objective two-sided assembly line balancing problems by harmony search algorithm based on pareto entropy
CN115034143A (en) Multi-resource cooperative intelligent workshop equipment configuration optimization method
Zhou et al. Imperialist competitive algorithm based on VNSOBL optimization for distributed parallel machine scheduling problem
CN114648232A (en) Cloud resource flexible job scheduling method based on improved chimpanzee optimization algorithm
CN114415615A (en) Mixed-flow assembly line balance distribution method and device under uncertain demand
Li et al. A metaheuristic to solve a robotic cell job-shop scheduling problem with time window constraints
Li et al. Improved cuckoo algorithm for the hybrid flow-shop scheduling problem in sand casting enterprises considering batch processing
CN116795054B (en) Intermediate product scheduling method in discrete manufacturing mode
Wang et al. A feedback-based artificial bee colony algorithm for energy-efficient flexible flow shop scheduling problem with batch processing machines
CN110852500B (en) Resource-limited hybrid flow shop optimization method
Zhang et al. Research on flexible job shop scheduling problem based on improved discrete particle swarm optimization

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WW01 Invention patent application withdrawn after publication

Application publication date: 20220909

WW01 Invention patent application withdrawn after publication