FLNG pipeline layout method based on multi-strategy optimization
By employing a multi-strategy optimization method for FLNG pipeline layout, utilizing Latin hypercube initialization and five optimization methods, the problems of insufficient global search capability and slow convergence speed in FLNG ship compartment pipeline layout are solved, achieving efficient and automated pipeline path optimization.
Patent Information
- Application Number
- CN202511897362.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-16
- Publication Date
- 2026-02-10
AI Technical Summary
Existing technologies for FLNG ship compartment piping layout suffer from insufficient global search capability, slow convergence speed, the need for manual path correction, and difficulty in balancing multiple objectives under multiple constraints.
A multi-strategy optimization-based FLNG pipeline layout method is adopted, including Latin hypercube initialization, orthogonalization processing, five individual position optimization methods (long-distance migration, burrowing, foraging, avoiding predators, and elite guidance) and dynamic energy factor switching, combined with weight parameters to optimize path length and number of bends.
It improves the convergence speed and path optimization efficiency of FLNG pipeline layout, and can balance multiple objectives under complex constraints to achieve automated layout and reduce manual correction.
Smart Images
Figure CN121502970A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of automatic pipeline layout, and more particularly to a FLNG pipeline layout method based on multi-strategy optimization. BACKGROUND
[0002] As the "blood vessel network" of FLNG, the pipeline system is responsible for the transportation of multiple media such as natural gas, liquefied natural gas, refrigerant, and cooling water, and directly determines the operation efficiency, safety and stability, and service life of the device. However, the internal space of the FLNG cabin is extremely limited and complex in structure, containing a large number of equipment such as storage tanks, pump groups, heat exchangers, and valves, and the pipeline needs to meet multiple engineering requirements such as high-pressure / high-temperature transportation of multiple media, process continuity, equipment maintenance accessibility, and anti-seismic and anti-impact in a narrow space, making the pipeline layout of the FLNG cabin one of the most technically challenging and time-consuming key links in device design.
[0003] Layout scheme based on traditional intelligent optimization algorithm: the industry attempts to apply genetic algorithm, particle swarm algorithm, and original artificial lemming algorithm (ALA) to pipeline layout, but there are multiple defects: insufficient global search capability, easy to fall into local optimum, unable to cover high-dimensional and multi-constrained solution space; slow convergence speed, a large number of invalid searches in iteration, difficult to meet the engineering timeliness requirements; no special mechanism designed for the orthogonal constraints of FLNG pipeline, the output path often contains diagonal segments and needs to be manually corrected; the fitness function design is single and cannot balance multiple objectives such as shortest path, least number of bends, and good obstacle avoidance effect, and the layout scheme has poor practicality. In view of this, we propose a FLNG pipeline layout method based on multi-strategy optimization. SUMMARY
[0004] 1. Technical problem to be solved The purpose of the present application is to provide a FLNG pipeline layout method based on multi-strategy optimization to solve the problems raised in the background art.
[0005] 2. Technical scheme The present application is implemented by the following technical scheme: A FLNG pipeline layout method based on multi-strategy optimization, comprising the following steps: S1, initializing the number of individuals and individual positions in the population; S2, orthogonalizing the individual positions in the population; S3, calculating the fitness of the individual according to the fitness, and determining the specified number of elite individual positions according to the fitness; if the individual position interferes with the obstacle, the fitness is set to infinity; otherwise, the fitness of the individual is calculated using the following formula: f=W1×L+W2×re+W2×EL; Wherein, W1, W2, W2 are weight parameters; L is path length; re is path point repetition number; EL is elbow number; S4, using long-distance migration, digging, foraging, avoiding natural enemies and elite guidance five ways to optimize the individual position: S5, repeat S2-S4, until the maximum iteration number is reached; S6, according to the individual position with the best fitness to pipeline layout.
[0006] As an optional solution of the technical scheme of the present application file, in S1, Latin hypercube initialization is used to determine the position of the individual, the formula is as follows: , k = 1, 2…N; i = 1, 2…dim; Wherein, Aik is the value of the kth individual in i dimension, Uik is a random number in the range of 0 to 1, N is the population size, LBj is the lower limit of the jth dimension, UBj is the upper limit of the jth dimension, dim is the total number of dimensions.
[0007] As an optional solution of the technical scheme of the present application file, In S4, the optimization formula corresponding to long-distance migration is: ; Wherein, Zi (t+1) represents the position of the ith individual after the t+1 optimization; Zbest (t) represents the elite individual position; F is a flag to change the search direction, taking the value of 1 or -1; Random number vector representing Brownian motion, the value range is [0,1]; R is a vector with size 1 x Dim, whose elements are random numbers uniformly distributed in the interval[ 1,1]; Zi (t) represents the position of the ith individual after the tth optimization; Za (t) represents the position of the individual after the tth optimization randomly selected from the population; The formula corresponding to digging is: ; Wherein, Zb(t) represents the position of the individual after the tth optimization randomly selected from the population; L is a random number related to the current iteration number, calculated by the following formula: ; The formula corresponding to foraging is: ; Wherein, spiral represents the shape of the random search in the foraging process, calculated by the following formula: ; ; The formula corresponding to predator avoidance is: ; ; ; wherein u and v are random values in the interval of 0-1, and β is a constant equal to 1.5; The formula corresponding to elite guidance is: ; wherein, is a learning step factor, and Zc(t) represents the position of an elite individual from the elite individual after the tth optimization.
[0008] As an optional solution of the technical scheme of the present application, in S4, when the energy factor is greater than 1, the individual position is optimized by long-distance migration, hole digging or elite guidance, and when the energy factor is less than or equal to 1, the individual position is optimized by foraging or predator avoidance; the energy factor is calculated in the following manner: ; wherein, is the maximum number of iterations.
[0009] As an optional solution of the technical scheme of the present application, in S4, when the energy factor is less than 1, the sum of the probabilities of optimizing the individual position by long-distance migration, hole digging and elite guidance is 1, and the probability of optimizing the individual position by long-distance migration is 0.3; the probability of optimizing the individual position by elite guidance is calculated in the following formula: P = 0.7 × (0.3 + 0.4 × (t / Tmax)); wherein t is the current number of iterations, and Tmax is the maximum number of iterations.
[0010] As an optional solution of the technical scheme of the present application, in S4, when the energy factor is less than 1, the sum of the probabilities of optimizing the individual position by foraging or predator avoidance is 1, and the probability of optimizing the individual position by long-distance migration is 0.5.
[0011] As an optional solution of the technical scheme of the present application, the S2 comprises the following sub-steps: S201, adjusting the coordinate points in the individual position so as to orthogonalize the line segments in the corresponding path; S202, removing the coordinate points at non-corn points in the individual position; S203, keeping the longitudinal coordinate in the individual position fixed, and optimizing the transverse coordinate in the following formula: i = 2, 3…dim; wherein, Xi is the horizontal coordinate value of the i-th path point position after optimization, Xi-1 is the horizontal coordinate value of the i-1-th path point position before optimization, Xi+1 is the horizontal coordinate value of the i+1-th path point position before optimization, and Round represents rounding the result.
[0012] S204, keeping the horizontal coordinate fixed in the individual position, optimizing the vertical coordinate using the following formula: i = 2, 3…dim; wherein, Yi is the vertical coordinate value of the i-th path point position after optimization, Yi-1 is the vertical coordinate value of the i-1-th path point position before optimization, Yi+1 is the vertical coordinate value of the i+1-th path point position before optimization.
[0013] As an optional solution of the technical scheme of the present application file, in S3, the number of elite individuals accounts for 10% of the total number of individuals.
[0014] 3. Beneficial effects Compared with the prior art, the beneficial effects of the present application are: 1) The present application can quickly identify high-quality search areas and increase convergence speed by initializing individual positions with Latin hypercubes.
[0015] 2) The present application can expand the exploration range of FLNG cabin pipeline layout while realizing local fine-tuning optimization of path details by setting five individual position optimization modes, and can balance efficiency and accuracy by combining dynamic switching of energy factors. BRIEF DESCRIPTION OF DRAWINGS
[0016] Figure 1 is a logic block diagram of a FLNG pipeline layout method based on multi-strategy optimization; Figure 2 is a schematic diagram of the initial individual position of a FLNG pipeline layout method based on multi-strategy optimization; Figure 3 is a schematic diagram of the individual position after orthogonalization of a FLNG pipeline layout method based on multi-strategy optimization; Figure 4 is a schematic diagram of the elite individual position of a FLNG pipeline layout method based on multi-strategy optimization; Figure 5 is a schematic diagram of the individual position after the first iteration of a FLNG pipeline layout method based on multi-strategy optimization; Figure 6It is a schematic diagram of optimal individual position of a FLNG pipeline layout method based on multi-strategy optimization. Figure 7 It is a schematic diagram of fitness variation of a FLNG pipeline layout method based on multi-strategy optimization. DETAILED DESCRIPTION
[0017] The technical solutions of the present application will be described clearly and completely below in combination with the drawings.
[0018] Please refer to Figure 1 , the present application provides a FLNG pipeline layout method based on multi-strategy optimization, comprising the following steps: S1, initializing the number of individuals and the individual position in the population; S2, orthogonalizing the individual position in the population; S3, calculating the fitness of the individual according to the fitness, and determining the specified number of elite individual positions according to the fitness; if the individual position interferes with the obstacle, the fitness is set to infinity; otherwise, the fitness of the individual is calculated using the following formula: f=W1×L+W2×re+W2×EL; wherein W1, W2, W2 are weight parameters, respectively taking 4, 2, 3; L is the path length; re is the path point repetition number; EL is the number of elbows; S4, using any one of the five ways of long-distance migration, digging, foraging, avoiding natural enemies and elite guidance to optimize the individual position: S5, repeating S2-S4 until the maximum iteration number is reached; S6, performing pipeline layout according to the individual position with optimal fitness.
[0019] In step S1, the individual position is usually initialized in a random manner, and the random initialization manner may cause population aggregation phenomenon, affecting the convergence speed of the algorithm; therefore, in the present scheme, the individual position is initialized using the following formula: , k=1, 2…N; i=1, 2…dim; wherein Aik is the value of the kth individual in the i-dimensional space, Uik is a random number ranging from 0 to 1, N is the population size, LBj is the lower limit of the jth dimension, UBj is the upper limit of the jth dimension, and dim is the total number of dimensions. When the Latin hypercube initialization is used to define the individual position, the boundary conditions are considered from two dimensions of “spatial range constraint” and “obstacle interference constraint”: the physical boundary of the cabin is used to clearly define the upper and lower bounds of the search space, and the dimension value generated by the initialization formula is mapped to the corresponding physical coordinate interval to ensure that the position is within the layout space; at the same time, the initial position is detected for obstacle collision, and the invalid points coinciding with the obstacles are removed to provide initial samples for subsequent optimization.
[0020] Because the pipeline path needs to strictly adhere to orthogonality constraints, the position of each individual needs to be adjusted in step S2, which includes the following sub-steps: S201. Adjust the coordinates of the individual positions to make the line segments in their corresponding paths orthogonal; Specifically, this includes: by inserting an inflection point between two adjacent coordinates, the algorithm selects the one that makes the total path shorter from two deterministic schemes: "horizontal first, then vertical" and "vertical first, then horizontal," ensuring that the two newly generated line segments satisfy the orthogonal condition of equal horizontal coordinates (vertical) or equal vertical coordinates (horizontal).
[0021] S202. Remove the coordinates of non-corner points in the individual location; S203. Keep the vertical coordinate of the individual position fixed, and optimize the horizontal coordinate using the following formula: ,i=2,3…dim; in, To optimize the x-coordinate value of the i-th path point, To optimize the x-coordinate value of the (i-1)th path point, To optimize the x-coordinate value of the (i+1)th path point, Round indicates rounding the result.
[0022] S204. Keeping the horizontal coordinate of the individual position fixed, optimize the vertical coordinate using the following formula: ,i=2,3…dim; in, To optimize the ordinate value of the i-th path point, To optimize the ordinate value of the (i-1)th path point, The ordinate value of the position of the (i+1)th path point before optimization.
[0023] In order to select elite individuals as the direction for subsequent optimization and increase the convergence speed of the algorithm, the fitness of individuals needs to be calculated in step S3. The method for calculating the number of repetitions of path points is to traverse each pair of path points in the sequence of path points in the given individual position and count them by judging whether their coordinates are completely equal.
[0024] The lower the fitness, the more reasonable the path planning; according to the rule of minimum fitness, 10% of the individuals in the population are selected as elite individuals to guide the optimization of individual positions.
[0025] In step S4, five optimization methods are provided: long-distance migration, burrowing, foraging, avoiding predators, and elite guidance, which are used to balance the search range and convergence speed. The optimization formula for long-distance migration is: ; Where Zi (t+1) represents the position of the i-th individual after the (t+1)-th optimization; Zbest (t) represents the position of the elite individual; F serves as a flag for changing the search direction, taking a value of 1 or -1; R represents a random number vector characterizing Brownian motion, with values ranging from [0,1]; R is a vector of size 1×Dim, whose elements are uniformly distributed in the interval [0,1]. A random number within the range [1, 1]; Zi(t) represents the position of the i-th individual after the t-th optimization; Za(t) represents the position of the individual randomly selected from the population after the t-th optimization; The formula for digging a hole is: ; Where Zb(t) represents the position of an individual randomly selected from the population after the t-th optimization; L is a random number related to the current iteration number, calculated using the following formula: ; The formula for foraging is: ; Where, *spiral* represents the shape of a random spiking pattern during foraging, and is calculated using the following formula: ; ; The formula for avoiding predators is: ; ; ; Where u and ν are random values in the interval 0-1, and β is a constant equal to 1.5; The formula for elite guidance is: ; in, To learn the step size factor, we take 0.4, and Zc(t) represents the position of the elite individual from the elite individuals after the t-th optimization.
[0026] In S4, when the energy factor is greater than 1, the individual's location is optimized through long-distance migration, burrowing, or elite guidance; when the energy factor is less than or equal to 1, the individual's location is optimized through foraging or avoiding predators. The energy factor is calculated in the following way: ; in, This represents the maximum number of iterations.
[0027] When the energy factor is less than 1, the sum of the probabilities of optimizing an individual's location through long-distance migration, burrowing, and elite guidance is 1, and the probability of optimizing an individual's location through long-distance migration is 0.3. The probability of optimizing an individual's location through elite guidance is calculated using the following formula: P=0.7 ×(0.3 + 0.4 × (t / Tmax)); Where t is the current iteration number and Tmax is the maximum iteration number.
[0028] The use of an elite-guided approach with increasing probability for position updates avoids the reliance on random behavior in traditional algorithms, which can lead to ineffective exploration in complex spaces. Elite guidance focuses on high-quality elite individuals, reducing blind spots and improving optimization efficiency. Initially, the probability of elite guidance is 21%, preserving global exploration capabilities, and later increases to 49% to enhance the utilization of high-quality solutions, adapting to the needs of FLNG pipeline layout from wide-area search to precise optimization.
[0029] When the energy factor is less than 1, the sum of the probabilities of optimizing an individual's location by seeking out or avoiding predators is 1, and the probability of optimizing an individual's location by long-distance migration is 0.5.
[0030] The following will introduce this method with specific examples: In this case, the population size was set to 30 and the number of path points was set to 10.
[0031] The initial individual positions generated in step S1, along with layout space and obstacle information, are as follows: Figure 2 As shown: The individual positions after orthogonalization processing in step S2 are as follows: Figure 3 As shown: The location information of elite individuals determined in step S3, such as Figure 4 As shown: The location information after optimizing the individual location through step S4 is as follows: Figure 5 As shown: By repeating steps S2 to S4 until the maximum number of iterations reaches 1000, the location of the individual with the best fitness is selected for pipeline layout, such as... Figure 6 As shown.
[0032] By comparing this scheme with the traditional ALA algorithm, the results are as follows: Figure 7 As shown; in Figure 7 In the graph, the horizontal axis represents fitness, and the vertical axis represents the number of iterations.
Claims
1. A multi-strategy optimization-based FLNG pipeline layout method, characterized in that: Includes the following steps: S1. Initialize the number and location of individuals in the population; S2. Orthogonalize the positions of individuals in the population; S3. Calculate the fitness of each individual and determine the positions of a specified number of elite individuals based on their fitness. If an individual's position interferes with an obstacle, set the fitness to infinity; otherwise, calculate the individual's fitness using the following formula: f = W1 × L + W2 × re + W2 × EL; Where W1, W2, and W2 are weight parameters; L is the path length; re is the number of times a path point is repeated; and EL is the number of bends. S4. Optimize the individual's location using any one of the following five methods: long-distance migration, burrowing, foraging, avoiding predators, and elite guidance: S5. Repeat S2-S4 until the maximum number of iterations is reached; S6. Layout the pipeline according to the location of the individual with the best fitness.
2. The FLNG pipeline layout method based on multi-strategy optimization according to claim 1, characterized in that: In S1, the position of an individual is determined using Latin hypercube initialization, as shown in the following formula: ,k=1,2…N;i=1、2…dim; Where Aik is the value of the k-th individual in dimension i, Uik is a random number in the range of 0 to 1, N is the population size, LBj is the lower bound of the j-th dimension, UBj is the upper bound of the j-th dimension, and dim is the total number of dimensions.
3. The FLNG pipeline layout method based on multi-strategy optimization according to claim 1, characterized in that: In S4, the optimization formula for long-distance migration is: ； Where Zi (t+1) represents the position of the i-th individual after the (t+1)-th optimization; Zbest (t) represents the position of the elite individual; F serves as a flag for changing the search direction, taking a value of 1 or -1; R represents a random number vector characterizing Brownian motion, with values ranging from [0,1]; R is a vector of size 1×Dim, whose elements are uniformly distributed in the interval [0,1]. A random number within the range [1, 1]; Zi(t) represents the position of the i-th individual after the t-th optimization; Za(t) represents the position of the individual randomly selected from the population after the t-th optimization; The formula for digging a hole is: ； Where Zb(t) represents the position of an individual randomly selected from the population after the t-th optimization; L is a random number related to the current iteration number, calculated using the following formula: ； The formula for foraging is: ； Where, *spiral* represents the shape of a random spiking pattern during foraging, and is calculated using the following formula: ； ； The formula for avoiding predators is: ； ； ； Where u and ν are random values in the interval 0-1, and β is a constant equal to 1.5; The formula for elite guidance is: ； in, To learn the step size factor, Zc(t) represents the position of the elite individual from the elite individuals after the t-th optimization.
4. The FLNG pipeline layout method based on multi-strategy optimization according to claim 3, characterized in that: In S4, when the energy factor is greater than 1, the individual's location is optimized through long-distance migration, burrowing, or elite guidance; when the energy factor is less than or equal to 1, the individual's location is optimized through foraging or avoiding predators. The energy factor is calculated in the following way: ； in, This represents the maximum number of iterations.
5. The FLNG pipeline layout method based on multi-strategy optimization according to claim 1, characterized in that: In S4, when the energy factor is less than 1, the sum of the probabilities of optimizing an individual's position using long-distance migration, burrowing, and elite guidance is 1, and the probability of optimizing an individual's position using long-distance migration is 0.
3. The probability of optimizing an individual's position using elite guidance is calculated using the following formula: P=0.7 ×(0.3 + 0.4 × (t / Tmax)); Where t is the current iteration number and Tmax is the maximum iteration number.
6. The FLNG pipeline layout method based on multi-strategy optimization according to claim 1, characterized in that: In S4, when the energy factor is less than 1, the sum of the probabilities of optimizing an individual's location by seeking out or avoiding predators is 1, and the probability of optimizing an individual's location by long-distance migration is 0.
5.
7. The FLNG pipeline layout method based on multi-strategy optimization according to claim 1, characterized in that: S2 includes the following sub-steps: S201. Adjust the coordinates of the individual positions to make the line segments in their corresponding paths orthogonal; S202. Remove the coordinates of non-corner points in the individual location; S203. Keep the vertical coordinate of the individual position fixed, and optimize the horizontal coordinate using the following formula: ,i=2,3…dim; in, To optimize the x-coordinate value of the i-th path point, To optimize the x-coordinate value of the (i-1)th path point, To optimize the x-coordinate value of the (i+1)th path point, Round indicates rounding the result. S204. Keeping the horizontal coordinate of the individual position fixed, optimize the vertical coordinate using the following formula: ,i=2,3…dim; in, To optimize the ordinate value of the i-th path point, To optimize the ordinate value of the (i-1)th path point, The ordinate value of the position of the (i+1)th path point before optimization.
8. The FLNG pipeline layout method based on multi-strategy optimization according to claim 1, characterized in that: In S3, the number of elite individuals accounts for 10% of the total number of individuals.