A ship section manufacturing workshop multi-objective scheduling optimization method

Abstract

The present application belongs to the field of ship manufacturing production management, and discloses a ship sub-assembly manufacturing workshop multi-objective scheduling optimization method and system. The method comprises the following steps: converting a non-buffer scheduling problem covering the whole process of plate preprocessing, cutting and decomposition, part processing and sub-assembly assembly into a quantum population evolution problem; constructing a mixed integer programming model, quantifying the non-buffer blocking constraint through the time sequence feasible region boundary, and taking the minimum completion time and total processing energy consumption as the target; solving by using a multi-strategy quantum heuristic iterative greedy algorithm, including three-layer hybrid coding, quantum rotation gate updating, destruction and reconstruction based on quantum observation, and energy consumption perception local adjustment. The present application effectively avoids the dynamic blocking conflict of the flat sub-assembly assembly line, realizes the collaborative optimization of the workshop production capacity utilization rate and the processing energy efficiency, and provides decision support for the green and lean construction of large ships.

Description

Technical Field

[0001] This invention belongs to the field of shipbuilding production management, and specifically relates to a multi-objective scheduling optimization method and system for ship section manufacturing workshops. Background Technology

[0002] The manufacturing of large cruise ship sections involves a fully integrated process of plate cutting and disassembly, component processing, and section assembly. Its production efficiency is highly dependent on the dynamic coupling between the flow of heterogeneous components generated after plate disassembly and the downstream assembly cycle. Due to the enormous size of cruise ship sections, each process stage in the workshop is constrained by physical space, often exhibiting extreme limitations without buffer zones. This makes the production line prone to in-situ blockages, leading to production cycle delays. Furthermore, addressing the differences in rated power of heterogeneous equipment throughout the manufacturing process, and how to reduce total processing energy consumption through optimized job assignment while meeting rigid timing blockage boundaries, is a core bottleneck in achieving green shipbuilding.

[0003] To address the end-to-end integrated scheduling problem, existing technical literature 1, "Mixed-integer linear programming and composed heuristics for three-stage remanufacturing system scheduling problem," Wang et al., Engineering Applications of Artificial Intelligence 137:109257 (2024), proposes a mixed-integer programming model to solve the three-stage remanufacturing scheduling problem involving disassembly, processing, and assembly. Although this research covers the three-stage integrated logic, its model construction is based on unrelated parallel machines and assumes an infinite buffer, failing to characterize the heterogeneous component flow characteristics generated by sheet metal cutting in the cruise ship section manufacturing line and the congestion interference in an unbuffered environment. Existing technical literature 2, “Mathematical model and knowledge-based iterated greedy algorithm for distributed assembly hybrid flow shopscheduling problem with dual-resource constraints,” Yu et al., Expert Systems With Applications 239: 122434 (2024), explores distributed assembly scheduling under multiple resource constraints. However, the model lacks the front-end plate cutting stage and is based on the non-blocking assumption, which cannot meet the integrated scheduling requirements of cruise ship manufacturing starting from steel plate feeding.

[0004] In summary, existing technologies are unsuitable for the "end-to-end + unbuffered + heterogeneous energy efficiency" production environment of large cruise ship flat-section manufacturing. On the one hand, there is a lack of precise mathematical models capable of simultaneously quantifying the coupled logic of "decomposition-processing-assembly" and dynamic congestion effects; on the other hand, existing methods struggle to optimize energy consumption during processing while maintaining production cycle time, taking into account the power characteristics of heterogeneous equipment. Therefore, there is an urgent need to research a Pareto-optimal scheduling scheme that can output a short production cycle and low total processing energy consumption, given complex process logic, timing congestion boundaries, and known equipment rated power. Summary of the Invention

[0005] In view of the deficiencies in the existing technology, this invention proposes a multi-objective scheduling optimization method and system for ship section manufacturing workshops.

[0006] The technical solution adopted in this invention: This invention discloses a multi-objective scheduling optimization method for ship section manufacturing workshops, comprising the following steps: S1: Construct a mixed integer programming model, the model including: setting optimization conditions, defining corresponding decision variables, and determining constraints; S1.1: Construct a mixed integer programming model covering the entire process of sheet pretreatment, cutting and decomposition, component processing and segmented assembly. Transform the unbuffered discrete scheduling problem covering the entire process of sheet pretreatment, cutting and decomposition, component processing and segmented assembly into a quantum population cooperative evolution and steady-state optimization problem in a multidimensional solution space. The goal is to maximize the completion time and the total processing energy consumption of the equipment, and the optimization conditions are to minimize the maximum completion time and minimize the total processing energy consumption of the equipment. S1.2: Define decision variables, including the order of raw material decomposition, the order of segmented assembly, and the assignment relationship between processes and equipment; S1.3: Determine the constraints, including the uniqueness constraint of process assignment, the process sequence constraint on the same equipment, the timing constraint of materials before and after the process, and the no-buffer blocking constraint; S2: Based on a multi-strategy quantum heuristic iterative greedy algorithm, the mixed integer programming model is solved, and a Pareto optimal scheduling scheme set is output; S2.1: Encode the solution of the mixed integer programming model into quantum population individuals, each individual consisting of a state vector representing the discrete scheduling scheme and a quantum probability magnitude vector representing the decision uncertainty; S2.2: Based on the Pareto dominance relation, perform non-dominated sorting of individuals in the quantum population and divide them into energy level levels; S2.3: Through the quantum rotation gate update mechanism, the population is driven to evolve in a direction towards a low-energy-consumption, high-efficiency solution region based on the phase information of the best individual in the current Pareto front. S2.4: Perform destruction and reconstruction operations based on quantum observation, and combine them with energy-sensing operators for local search; S2.5: Determine whether the actual running time of the algorithm has reached the dynamically set termination time threshold, which is calculated from the current scheduling problem size parameter; if yes, stop the evolution and output the current Pareto optimal scheduling scheme set; if no, return to S2.2 to continue iterating.

[0007] Preferably, in step S1.1, the quantification of the maximum completion time and the total processing energy consumption of the equipment includes: the potential energy at the maximum completion time. Its expression is: ,in, For the final segmented product collection, For products When the assembly phase is completed and the system is offline. For the final segmented product collection; Total processing energy consumption of equipment Its expression is: ,in For all processes, For equipment collection, For process Assign to device Decision variables, For processing time, Rated power of the equipment For process Energy consumption attribute coefficient; defining individual The multidimensional fitness potential function is It includes two minimization objectives: maximum completion time potential energy. Total processing energy consumption of the equipment .

[0008] Preferably, the bufferless blocking constraint in S1.3 is defined as: workpiece Equipment at the current stage departure time It must be done no earlier than the subsequent process in the downstream equipment. The start time of the plan ,Right now ,in, For workpiece The subsequent processes in downstream equipment The start time of the plan; specifically, when the workpiece When it is a raw material, its departure time on the decomposition equipment is limited by the start time of all the parts it has decomposed on the processing equipment.

[0009] S1.3 further includes: when the individual solution violates When constraints are met, a cyclic correction mechanism is used to correct them into feasible solutions: ,in, This indicates that the workpiece was in a state of emergency before the blocking correction was performed. Equipment at the current stage The initial departure time is calculated based on its actual processing time; This indicates that after the blocking logic correction, the workpiece is in the equipment. The actual release time that was forcibly postponed.

[0010] Preferably, in S2.1, the state vector , For raw material decomposition sequence, This is a segmented assembly sequence. Assign vectors to workstation equipment; quantum probability amplitude vector: ; For the first The phase angle of the decision variable, the square of its sine value The mapping represents the probability of selecting a specific sorting position or a specific device for that dimension.

[0011] In S2.1, the state vector Its expression is: ; in, For raw material decomposition sequence, This is a segmented assembly sequence. Assign vectors to workstation equipment; The quantum probability amplitude vector Its expression is: , in, For the first The phase angle of the decision variable, the square of its sine value The mapping represents the probability of selecting a specific sorting position or a specific device for that dimension.

[0012] Furthermore, in S2.2, the non-dominated sorting rule is: for two individuals in the population... and If the following conditions are met: , Then determine the individual The energy level is better than ,Right now Dominate The algorithm divides the population into different Pareto level fronts based on this dominance relationship. Preferably, in S2.3, the quantum rotating gate update rule is: No. The phase angle update formula for the generation is: , Among them, the rotation angle increment The calculation combines optimal solution guidance with a dynamic decay strategy: , in, The phase angle of the current highest-ranking non-dominant individual; This is the initial rotation step size; Maximum number of iterations; via sign function Determine the rotation direction to drive the population probability amplitude towards a low-energy steady-state convergence.

[0013] Preferably, in S2.4, the destruction and reconstruction operations based on quantum observation include: The destruction and reconstruction process based on quantum observation and the execution logic of the energy consumption sensing operator are as follows: (1) Destruction operation: Determine the number of workpieces to be removed ,satisfy: , in, For sequence length, The predetermined destruction ratio coefficient; randomly removed from the individual raw material sequence and assembly sequence. Each workpiece node is stored in a candidate set; (2) Quantum probability-guided reconstruction operation: for the workpiece to be inserted in the candidate set Calculate the insertion position for each candidate. Recommended score : , in, For workpiece In the current iteration step The quantum phase angle below, The disturbance factor is a Gaussian random perturbation factor; a probability sampling strategy is adopted for the workpiece. Insert to score The highest candidate position; (3) Energy consumption sensing microstructure adjustment: In the reconstructed scheduling scheme, the retrieval process Optional device set And execute the equipment assignment change, satisfying the following judgment logic: ; , in, For the originally assigned equipment, Low-power devices are the candidates; The rated power of the corresponding equipment; The process is assigned to the original equipment. Maximum completion time and total processing energy consumption; They attempted to replace the process with low-power equipment. The maximum completion time and total processing energy consumption were then recalculated. This is the time loss tolerance factor; (4) Iteration and Output: Repeat the above process until the preset dynamic time termination condition is met. The dynamic time termination condition satisfies: ,in This represents the current cumulative running time of the algorithm. This refers to the number of segments and components. This refers to the number of process stages. This represents the total number of heterogeneous equipment in the workshop. The time consumption coefficient is calculated in units; the algorithm dynamically updates the archive set in real time during the iteration process, and finally outputs the Pareto optimal schedule set after the termination condition is met.

[0014] This invention discloses a multi-objective scheduling optimization system for a ship section manufacturing workshop, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The system is characterized in that the processor implements the steps of the method when executing the program.

[0015] The method is applicable to flat section manufacturing workshops for large cruise ships, large container ships, or liquefied natural gas carriers.

[0016] This invention discloses a multi-objective scheduling optimization system for a ship section manufacturing workshop, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The system is characterized in that the processor executes the program to implement the steps of any of the methods described above.

[0017] Compared with the prior art, the present invention has the following beneficial effects: By applying the technical solution of this invention, the unbuffered pipeline congestion constraint is quantified into the temporally feasible domain boundary, and a mixed integer programming model with the goal of minimizing completion time and total processing energy consumption is constructed. Then, a multi-strategy quantum heuristic iterative greedy algorithm is used to solve the problem, which effectively avoids the dynamic congestion conflict of the flat segmented pipeline, realizes the synergistic optimization of workshop capacity utilization and processing energy efficiency, and provides decision support for the green and lean construction of large ships. Attached Figure Description

[0018] Figure 1 This is an overall flowchart of the present invention; Figure 2 A schematic diagram of the integrated decomposition and assembly scheduling scheme for the straight sections of a cruise ship. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be further described clearly and completely below with reference to the accompanying drawings. It should be noted that the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0020] To address the deeply coupled multi-stage processes of "plate decomposition - component processing - section assembly" in the manufacturing of large cruise ship straight sections, this invention proposes a multi-objective scheduling optimization method and system for ship section manufacturing workshops. This method is based on the rigid physical constraints of the straight section assembly line without buffer zones. It aims to optimize the production cycle and processing efficiency by finely assigning heterogeneous workstation resources and collaboratively rearranging work sequences, effectively avoiding dynamic congestion effects on the assembly line. The specific implementation process is as follows: S1: Construct a mixed integer programming model covering the entire process of sheet pretreatment, cutting and decomposition, component processing and segmented assembly. The model aims to minimize the maximum completion time and minimize the total processing energy consumption of the equipment, and uses pipeline blockage caused by no buffer as a constraint. S1.1: Quantify the optimization objective, where the maximum completion time potential energy is... : , For the final segmented product Completion time; total processing energy consumption of the equipment : , in For all processes, For equipment collection, For process Assign to device Decision variables, For processing time, Rated power of the equipment For process Energy consumption attribute coefficient; defining individual The multidimensional fitness potential function is It includes two minimization objectives: maximum completion time potential energy. Total processing energy consumption of the equipment .

[0021] S1.2: Define decision variables, including the order of raw material decomposition, the order of segmented assembly, and the assignment relationship between processes and equipment; S1.3: Determine the constraints, including the uniqueness constraint of process assignment, the process sequence constraint on the same equipment, the timing constraint of materials before and after the process, and the no-buffer blocking constraint.

[0022] Preferably, the bufferless blocking constraint is defined as: workpiece Equipment at the current stage departure time It must be done no earlier than the subsequent process in the downstream equipment. The start time of the plan ,Right now .

[0023] When individual solutions violate When constraints are met, a cyclic correction mechanism is used to correct them into feasible solutions: .

[0024] S2: Design a multi-strategy quantum heuristic iterative greedy algorithm to solve the mixed integer programming model and output a Pareto optimal scheduling scheme set.

[0025] S2.1: Encode the solution of the mixed integer programming model into quantum population individuals, each individual consisting of a state vector representing the discrete scheduling scheme and a quantum probability magnitude vector representing the decision uncertainty; S2.2: Based on Pareto dominance, perform non-dominated ranking of individuals in the population and divide them into energy level hierarchies; S2.3: Through the quantum rotation gate update mechanism, the population is driven to evolve in a direction towards a low-energy-consumption, high-efficiency solution region based on the phase information of the best individual in the current Pareto front. S2.4: Perform destruction and reconstruction operations based on quantum observation, and combine them with energy-sensing operators for local search; S2.5: Determine whether the actual running time of the algorithm has reached the dynamically set termination time threshold, which is calculated from the current scheduling problem size parameter; if yes, stop the evolution and output the current Pareto optimal scheduling scheme set; if no, return to S2.2 to continue iterating.

[0026] This invention is an integrated decomposition and assembly multi-objective congestion flow scheduling optimization method applied to the manufacturing workshop of flat sections of large cruise ships.

[0027] A mixed-integer programming model is constructed, covering the entire process of sheet metal pretreatment, cutting and decomposition, component processing, and segmented assembly. This transforms the unbuffered discrete scheduling problem into a quantum population cooperative evolution and steady-state optimization problem within a multidimensional solution space. The model aims to minimize the maximum completion time and the total processing energy consumption of the equipment, with pipeline congestion caused by the lack of a buffer as a constraint. Specifically, it includes: S1.1 transforms the dual objectives of minimizing completion time and minimizing processing energy consumption into a vector evaluation problem of finding the Pareto front steady state of the population in a multidimensional potential energy field, specifically including: In the manufacturing of flat sections for large cruise ships, production efficiency and processing energy efficiency are mutually constraining physical properties without a unified dimension. This invention introduces multidimensional potential energy field theory, simulating the solution process as a continuous evolution of the system towards a lower energy and more structurally stable state (i.e., the Pareto optimal steady state). The evaluation system constructs a multidimensional fitness function, mapping the position of an individual in the discrete solution space to a vector energy level. The total potential energy of the system depends on the operating time and energy release efficiency of each process on heterogeneous equipment. By quantifying the contribution of equipment assignment schemes to overall energy efficiency, the evolutionary priority of individuals in the population evolution is defined.

[0028] Based on the process characteristics of segmented manufacturing, the following vector evaluation function and energy level delimitation formula are constructed: (1) (2) (3); in, This represents the potential energy value for the maximum completion time. For the final segmented product The moment when assembly is completed and the worker leaves the workstation; This represents the total processing energy consumption potential value of the system. For process Distributed to equipment The actual working time at that time For equipment Rated processing power, To correspond to the inherent energy consumption attribute coefficients of the sheet metal or components, this formula only counts the electrical energy consumption generated during the effective processing operation, defining individual... The multidimensional fitness potential function is It includes two minimization objectives: maximum completion time potential energy. Total processing energy consumption of the equipment .

[0029] This invention transforms the production decisions of the entire flat segmented workshop process into quantum individual representations within an algorithm. Quantum heuristic logic originates from the simulation of the superposition characteristics and coherent evolution of quantum individuals. It utilizes the wave function characteristics of probability amplitudes to induce a parallel evolutionary process in the entire population within a multidimensional discrete solution space, thereby obtaining a Pareto optimal solution. Based on the research problem of this invention, the production decisions of the entire flat segmented workshop process are transformed into quantum individuals within an algorithm. The gene sequence of an individual is mapped to the sequence of segmented processing and assembly, and the uncertainty of the decision variables is mapped to the phase angle of the qubits. The evolution of the probability amplitude represents the change in the equipment allocation scheme.

[0030] Initialization includes A quantum population of individuals, each represented by a state vector indicating a discrete scheduling scheme. and the phase vector representing the probability distribution composition( The corresponding individual characteristic vectors and quantum state formulas are expressed as follows: (4) (5) (6) (7) in, For the first The discrete state vector of each individual. For raw material decomposition sequence, This is a segmented assembly sequence. Assign vectors to workstation equipment; For the corresponding quantum phase vector, For the first The phase angle of the decision variables is uniformly set to an initial value of 1. For the first The quantum superposition state of dimensional decision-making, its probability amplitude This is mapped to the probability of selecting a specific heterogeneous device in a multi-camera environment.

[0031] S1.3 transforms the pipeline congestion constraint caused by the lack of buffer zones between different stages of the workshop into the temporal feasible domain boundary in the individual evolution process, serving as a criterion for eliminating or repairing states that violate temporal logic. Specifically, this includes: In a flat, segmented manufacturing production line, due to the large physical dimensions of segments and components, it is impossible to configure work-in-process buffers between various workstations, including preprocessing, cutting and decomposition, component processing, and segment assembly. This invention transforms physical space constraints into logical thresholds in the algorithm evolution process by constructing a temporally feasible domain boundary. When a segment task is completed at the current workstation, but its corresponding downstream equipment is still occupied, the task will exhibit a "situational delay," meaning it will forcibly occupy the current workstation's resources until the downstream equipment is ready to receive it. This invention defines scheduling schemes that violate this physical logic as illegal states and introduces a recursive cyclic correction algorithm in the decoding and reconstruction process to adjust the temporal parameters of individuals in the population, ensuring that the evolutionary path always remains within the feasible domain boundary.

[0032] The corresponding blocking boundary definition and timing correction formula are expressed as follows: (8) (9) in ， This is a set of tasks covering the entire process of decomposition, processing, and assembly. and These refer to the workstation at the current process stage and the workstation at the next stage. For workpiece In the equipment The actual departure time on the device, which characterizes the device's... The actual point of resource release; For workpiece The subsequent processes in downstream equipment The moment the plan began, This indicates that the workpiece was in a state of emergency before the blocking correction was performed. Equipment at the current stage The initial departure time is calculated based on its actual processing time; This indicates that after the blocking logic correction, the workpiece is in the equipment. The actual release time that was forcibly postponed.

[0033] Formula (8) establishes the lower bound of the feasible region in an unbuffered environment, meaning that the release time of any process in the current stage equipment must not be earlier than the planned start time of its downstream stage equipment; specifically, in the transition between the decomposition stage (Stage 2) and the processing stage (Stage 3), the departure time of the raw material sheet on the decomposition equipment is limited by the earliest start time of all its decomposed components on the processing equipment. If the observed individual state satisfies If the timing violates the physical blocking logic, the correction procedure shown in formula (9) is triggered. By comparing the original departure time with the downstream planned start time, the maximum value is taken as the corrected actual departure time. Through this forward delay mechanism, the present invention simulates the "blockage rebound" effect caused by the lack of buffer in the pipeline, ensuring that the generated integrated scheduling scheme conforms to the real operation rules of the flat section workshop of a large cruise ship.

[0034] Based on the variable mapping, evaluation system, and time-series correction mechanism defined above, this embodiment further constructs a complete mixed integer programming (MILP) model to accurately define the evolution space of the quantum population in the integrated decomposition and assembly process environment: (1) Multi-objective optimization function: In this embodiment, the Pareto optimal frontier of the production system is found by simultaneously optimizing the following dual objectives. The optimization function system is composed of formula (10) and formula (11): (10) (11) in, Used to ensure the delivery cycle of segmented products. This is used to reduce power consumption during operation by optimizing the heterogeneous device assignment scheme, and the power consumption statistics do not include standby power consumption of the device when it is in a blocked state.

[0035] (2) Uniqueness constraint for heterogeneous workstation allocation: Ensure that each production task in each process stage is assigned to one and only one physical device from its set of available heterogeneous parallel devices, so as to eliminate assignment ambiguity: (12) In the formula, These are the allocation decision variables for raw materials, components, and segmented assembly stages, respectively.

[0036] (3) Logical mutual exclusion constraint of sequence variables: In order to ensure the uniqueness of the work sequence decision on the same equipment and to prevent logical overlap or loops, the sequence variables of each stage are mutually exclusive. The following logical inequalities must be satisfied: Preprocessing and decomposition stage: (13) Parts processing stage: (14) Segmented assembly stage: (15) The above constraints ensure that when two tasks are assigned to the same workstation, the algorithm must make an exclusive decision on their order of priority.

[0037] (4) Equipment-level operation non-overlapping and blocking / delay constraints Based on the Big-M linearization concept, the timing boundaries of each process when occupying equipment resources are defined as follows: (16) Combining the blocking logic defined in formula (8), Includes process The processing time and the in-situ dwell time occupied by downstream workstations. This constraint forces the equipment. Before the preceding workpiece has entered the next stage, it is strictly forbidden to start the operation of the subsequent workpiece.

[0038] (5) Constraints on end-to-end transfer and process integration Material transfer: (17) Decomposition phase integration (1 to N): (18) Assembly phase integration (N becomes 1): (19) Formulas (17)-(19) establish a global time-series chain from board feeding to segment completion, quantifying the material flow logic between each stage.

[0039] (6) Logical correlation and consistency constraints of decision variables To eliminate illegal assignment states in the solution space, a low-level linkage between assignment variables, order variables, and time variables is established: Variable activation constraints: (20) Spatiotemporal attribute binding logic: (twenty one) Formula (20) ensures that the order decision takes effect only when the task is jointly assigned, and Formula (21) ensures that the time attribute of the task only contributes a value to the activated device.

[0040] (7) Variable range and domain constraints (twenty two) Formula (22) defines the discrete properties of the decision scheme and ensures that the scheduling scheme satisfies causal positivity on the time axis.

[0041] In summary, the mathematical model composed of formulas (1) to (22) provides a complete physical constraint framework for the manufacturing workshop of flat sections of large cruise ships. For this highly coupled, nonlinear, and complex optimization space, this embodiment further employs the quantum probability-guided optimization strategy disclosed in section S2 below to iteratively search for the Pareto optimal solution set.

[0042] For this highly coupled, nonlinear, and complex optimization space, this embodiment further employs a quantum probability-guided optimization strategy to iteratively search for the Pareto optimal solution set.

[0043] S2.1 designs a quantum population representation strategy with a three-layer hybrid encoding structure to achieve integrated expression of multidimensional decision variables, specifically including: In the designed three-layer coding structure, the first layer is the raw material decomposition sequence, the length of which is equal to the total number of boards, used to define the feeding priority of boards entering the decomposition station. For example, for the 11 raw materials RM1 to RM11 in this embodiment, the coding sequence... This maps the physical sequence of the sheet metal processing by the cutting machine. The second layer is the segmented assembly sequence code, the length of which is equal to the total number of final segmented products, representing the operation sequence of the final assembly station; for example... Figure 2 As shown, segmented The arrangement sequence directly maps to the cycle stability at the end of the pipeline. The third layer assigns vectors to flexible devices, with each dimension determined by the quantum phase angle. Mapping is used for probabilistic assignment among heterogeneous devices in processing areas F1 to F3. A specific coding example logic is as follows: Figure 1 As shown.

[0044] Formula (3) in S2.2 defines the rules for defining energy level hierarchy (i.e., the non-dominated ordering rule): for two individuals in a population and ,when It is no less than in all target dimensions And it is superior to in at least one dimension At that time, the judgment Those orbiting at lower energy levels have a higher evolutionary survival rate. The algorithm uses this logic to build a Pareto front archive to reconcile the balance between production cycles and energy-efficient processing.

[0045] S2.3 introduces a quantum rotation gate update mechanism to drive the population to evolve in a directed manner towards a low-energy-consumption, high-efficiency solution region, specifically including: This embodiment utilizes a quantum rotation gate to dynamically adjust the phase angle of the population. The phase angle differentiates significantly with iteration, driving quantum individuals to shift towards lower energy levels. For example, the phase angle of raw material RM3 evolves to 1.5608 (normalized position mapping is 0.9936), indicating that the algorithm recognizes that this task should be executed at the end of the sequence to optimize the overall timing; while the phase angle of RM2 converges to 0.0100 (normalized position 0.0064), establishing its extremely high feeding priority. The specific phase angle evolution operator is shown in formula (23): (twenty three) In the formula, Indicates the current iteration step. The phase angle of the individual is guided by the selected Pareto front. This is achieved through the sign function. The direction of the phase deviation is determined, and the driving probability amplitude is directionally shifted towards the low-energy-consumption steady-state region in the multidimensional potential energy field.

[0046] The process of performing destruction and reconstruction based on quantum observation and energy-aware local optimization includes: The destruction and reconstruction process based on quantum observation and the execution logic of the energy consumption sensing operator are as follows: (1) Destruction operation: Determine the number of workpieces to be removed ,satisfy: , in, For sequence length, The predetermined destruction ratio coefficient; randomly removed from the individual raw material sequence and assembly sequence. Each workpiece node is stored in a candidate set; (2) Quantum probability-guided reconstruction operation: for the workpiece to be inserted in the candidate set Calculate the insertion position for each candidate. Recommended score : , in, For workpiece In the current iteration step The quantum phase angle below, The disturbance factor is a Gaussian random perturbation factor; a probability sampling strategy is adopted for the workpiece. Insert to score The highest candidate position; (3) Energy consumption sensing microstructure adjustment: In the reconstructed scheduling scheme, the retrieval process Optional device set And execute the equipment assignment change, satisfying the following judgment logic: , in, For the originally assigned equipment, Low-power devices are the candidates; The rated power of the corresponding equipment; These are the completion time and total processing energy consumption, respectively. This is the time loss tolerance factor; The process is assigned to the original equipment. Maximum completion time and total processing energy consumption; They attempted to replace the process with low-power equipment. The maximum completion time and total processing energy consumption were then recalculated. (4) Iteration and Output: Repeat the above process until the preset dynamic time termination condition is met. The dynamic time termination condition satisfies:

[0047] ,in This represents the current cumulative running time of the algorithm. This refers to the number of segments and components. This refers to the number of process stages. This represents the total number of heterogeneous equipment in the workshop. The time consumption coefficient is calculated in units; the algorithm dynamically updates the archive set in real time during the iteration process, and finally outputs the Pareto optimal schedule set after the termination condition is met.

[0048] The algorithm continuously improves the quality of the solution through iterative search. The specific steps are as follows: Step 1: State Observation and Timing Correction. The population is measured using the qubit probability distribution, discrete codes are extracted, and recursive corrections are performed using blocking constraints. The assignment probability of tasks on heterogeneous devices exhibits a non-uniform distribution; for example, the assignment probability of RM1 on device F1_DECM_s1_m2 evolves to 0.8000. The algorithm utilizes this probability advantage to guide individual state collapse. Step 2: Gene Destruction. Based on a preset destruction ratio (e.g., ...). Several key nodes are randomly removed from the individual sequence and stored in the candidate selection set to form a fragmented gene chain with state vacancies. Step 3: Probabilistic-Guided Reconstruction. For the nodes to be inserted in the candidate set, the fitness recommendation score for each vacancy is calculated using the quantum phase angle. The nodes are then backfilled to the gene positions with the highest probability density. Step 4: Energy Consumption Sensing Adjustment. While maintaining the original congestion boundary, the energy consumption sensing operator is used to traverse the selectable workstation equipment. For example, for part P1, its power on F1_PROCM_s1_m1 is 1565.6, while... The power consumption is 644.3. The algorithm uses logical judgment to replace the task with a low-power device. To achieve local energy level downregulation; Step 5: Steady-state determination and output. Repeat the above steps, maintaining the non-dominated solution set in real time, until the iteration termination condition is met, and output the final Pareto optimal scheduling scheme.

[0049] Figure 2 shows the final scheduling Gantt chart generated in this embodiment. This invention uses a high-end cruise ship straight section manufacturing workshop built by a shipyard as a construction case for verification. This production line covers the entire process of plate pretreatment, cutting and decomposition, parts processing, and section assembly. Experimental results show that, under the premise of ensuring 100% compliance with the unbuffered congestion constraint, the method of this invention effectively solves the "ship jamming" phenomenon. The diagonally filled areas in the figure (such as task segments P13 and P20 in workstation F2_PROCM_s1_m2) visually demonstrate the in-situ stagnation and congestion effect caused by downstream occupation after workpiece completion. Through the collaborative optimization of this invention, the total power consumption of the section processing process is reduced by 16.2%~22.7% compared to the traditional random assignment scheme, significantly improving the green construction level of the production line.

[0050] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A multi-objective scheduling optimization method for a ship section manufacturing workshop, characterized in that, Includes the following steps: S1: Construct a mixed integer programming model, set optimization conditions, define corresponding decision variables, and determine constraints; S1.1: Construct a mixed integer programming model covering the entire process of sheet pretreatment, cutting and decomposition, parts processing and segmented assembly, with the maximum completion time and total energy consumption of equipment processing as indicators, and the optimization objectives being to minimize the maximum completion time and minimize the total energy consumption of equipment processing. S1.2: Define decision variables, including the order of raw material decomposition, the order of segmented assembly, and the assignment relationship between processes and equipment; S1.3: Determine the constraints, including the uniqueness constraint of process assignment, the process sequence constraint on the same equipment, the timing constraint of materials before and after the process, and the no-buffer blocking constraint; S2: Based on a multi-strategy quantum heuristic iterative greedy algorithm, the mixed integer programming model is solved, and a Pareto optimal scheduling scheme set is output; S2.1: Encode the solution of the mixed integer programming model into quantum population individuals, each individual consisting of a state vector representing the discrete scheduling scheme and a quantum probability magnitude vector representing the decision uncertainty; S2.2: Based on the Pareto dominance relation, perform non-dominated sorting of individuals in the quantum population and divide them into energy level levels; S2.3: Through the quantum rotation gate update mechanism, the population is driven to evolve in a direction towards a low-energy-consumption, high-efficiency solution region based on the phase information of the best individual in the current Pareto front. S2.4: Perform destruction and reconstruction operations based on quantum observation, and combine them with energy-sensing operators for local search; S2.5: Determine whether the actual running time of the algorithm has reached the dynamically set termination time threshold, which is calculated from the current scheduling problem size parameter; if yes, stop the evolution and output the current Pareto optimal scheduling scheme set; if no, return to S2.2 to continue iterating.

2. The method according to claim 1, characterized in that, In S1.1, the maximum completion time and total energy consumption of equipment processing are quantified, including: potential energy at maximum completion time. Its expression is: ， in, For the final segmented product collection, For products When the assembly phase is completed and the system is offline. For the final segmented product collection; Total processing energy consumption of equipment Its expression is: ,in For all processes, For equipment collection, For process Assign to device Decision variables, For processing time, Rated power of the equipment For process The energy consumption attribute coefficient.

3. The method according to claim 1, characterized in that, The bufferless blocking constraint in S1.3 is defined as: workpiece Equipment at the current stage departure time It must be done no earlier than the subsequent process in the downstream equipment. The start time of the plan ,Right now ,in, For workpiece The subsequent processes in downstream equipment The moment the plan began.

4. The method according to claim 1, characterized in that, S1.3 also includes: When individual solutions violate When constraints are met, a cyclic correction mechanism is used to correct them into feasible solutions. Its expression is: ， in, Indicates violation Individual solutions to constraints.

5. The method according to claim 1, characterized in that, In S2.1, the state vector Its expression is: ， in, For raw material decomposition sequence, This is a segmented assembly sequence. Assign vectors to workstation equipment; The quantum probability amplitude vector Its expression is: ， in, For the first The phase angle of the decision variable, the square of its sine value The mapping represents the probability of selecting a specific sorting position or a specific device for that dimension.

6. The method according to claim 1, characterized in that, In S2.2, the non-dominated sorting rule is: for two individuals in the population... and If the following conditions are met: ， Then determine the individual The energy level is better than ,Right now Dominate The algorithm divides the population into different Pareto level fronts based on this dominance relationship.

7. The method according to claim 1, characterized in that, In S2.3, the quantum rotation gate update rule is as follows: No. The phase angle update formula for the generation is: ， Among them, the rotation angle increment The calculation combines optimal solution guidance with a dynamic decay strategy: ， in, The phase angle of the current highest-ranking non-dominant individual; This is the initial rotation step size; Maximum number of iterations; via sign function Determine the rotation direction to drive the population probability amplitude towards a low-energy steady-state convergence.

8. The method according to claim 1, characterized in that, In S2.4, the destruction and reconstruction operations based on quantum observation include: (1) Destruction operation: Determine the number of workpieces to be removed ,satisfy: ， in, For sequence length, The predetermined destruction ratio coefficient; randomly removed from the individual raw material sequence and assembly sequence. Each workpiece node is stored in a candidate set; (2) Quantum probability-guided reconstruction operation: for the workpiece to be inserted in the candidate set Calculate the insertion position for each candidate. Recommended score : , in, For workpiece In the current iteration step The quantum phase angle below, The disturbance factor is a Gaussian random perturbation factor; a probability sampling strategy is adopted for the workpiece. Insert to score The highest candidate position; (3) Energy consumption sensing microstructure adjustment: In the reconstructed scheduling scheme, the retrieval process Optional device set And execute the equipment assignment change, satisfying the following judgment logic: ， , in, For the originally assigned equipment, Low-power devices are the candidates; The rated power of the corresponding equipment; The process is assigned to the original equipment. Maximum completion time and total processing energy consumption; They attempted to replace the process with low-power equipment. The maximum completion time and total processing energy consumption were then recalculated. This is the time loss tolerance factor; (4) Iteration and Output: Repeat the above process until the preset dynamic time termination condition is met. The dynamic time termination condition satisfies: ,in This represents the current cumulative running time of the algorithm. This refers to the number of segments and components. This refers to the number of process stages. This represents the total number of heterogeneous equipment in the workshop. The time consumption coefficient is calculated in units; the algorithm dynamically updates the archive set in real time during the iteration process, and finally outputs the Pareto optimal schedule set after the termination condition is met.

9. A multi-objective scheduling optimization system for a ship section manufacturing workshop, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1-8.

Images