Amphibious vehicle layout area utilization maximization method based on multi-strategy dynamic adjustment
A dynamic adjustment and multi-strategy technology, applied in the field of amphibious assault ship vehicle compartment vehicle layout, can solve the problems of waste of free area, slow algorithm convergence speed, and reduced algorithm space search ability
Active Publication Date: 2021-02-26
HARBIN UNIV OF SCI & TECH
12 Cites 0 Cited by
AI-Extracted Technical Summary
Problems solved by technology
In the process of placing rectangular parts, when the rectangular parts cannot be discharged into the lowest horizontal line and whether there are rectangular parts that can be discharged into the current lowest horizontal line among the subsequent undischarged rectangular parts, it will lead to waste of free space that can be used, affecting the final layout effect
[0005] (2) Vehicle arrangement optimization algorithm, the genetic algorithm adopting adaptive selection strategy in the early stage of evolution, increases the probability of individuals with high fitness value being selected, and reduces the probability of individuals with low fitness value being selected, making the selection method excessive Preferring to individuals with strong adaptability, the population may converge to a local optimal solution too quickly to fully explore the search space, resulting in premature algorithm
[0006] (3) In the late stage of population evolution of genetic algorithm, the difference in individual fitness value is very small, and the use of sequential crossover method may lead to invalid crossover and repeated individuals
[0007] (4) The mutation probability of the genetic algorithm adopts a fixed value, which reduces the ...
Method used
The self-adaptive genetic algorithm of contrasting vehicle layout and multi-strategy dynamic adjustment genetic algorithm fitness curve, traditional genetic algorithm initial parameter is set, iteration number of times 200, initial population size is 100, crossover probability is 0.6, and variation probability is 0.1, The initial parameters of genetic algorithm with multi-strategy dynamic adjustment are set as iteration times 200, initial population size is 100, subpopulation selection probability α0=β0=θ0=0.4, mutation probability p0=0.1. Comparing the simulation results of the two genetic algorithms for solving the vehicle layout problem, the change curves of the fitness values of the two genetic algorithms are shown in Figure 5 and Figure 6. The deck utilization rates of the adaptive genetic algorithm and the multi-strategy dynamic adjustment genetic algorithm are 88.74% and 94.09% respectively, and the calculation time is 219.37s and 167.91s respectively. and F-type amphibious vehicles are (12, 10, 11, 15, 14, 11) and (5, 18, 15, 9, 15, 14)...
Abstract
The invention relates to an amphibious vehicle layout area utilization maximization method based on multi-strategy dynamic adjustment, and the method comprises the steps: obtaining the related information of a vehicle cabin deck and a vehicle, determining a constraint condition, determining an objective function, employing decimal coding to code the vehicle, initializing parameters, randomly generating a vehicle layout sequence to form an initial population, calculating fitness values of individuals, storing the optimal individuals, judging whether the maximum number of iterations is reached or not, and dynamically adjusting the scales of the three sub-populations according to three different evaluation strategies and the selection probability of dynamic adjustment; and performing constrained crossover or annular crossover on the optimal individual and all individuals in the sub-populations, performing mutation operation by adopting a dynamically adjusted mutation probability, selecting effective evolution individuals in the three sub-populations to form a new population, and decoding the optimal individual of the last generation to obtain an optimal layout diagram. The method hasthe advantages that the optimal layout drawing can be quickly solved, and the layout area of the amphibious vehicle can be utilized to the maximum extent.
Application Domain
ForecastingGenetic algorithms
Technology Topic
AlgorithmIndustrial engineering +1
Image
Examples
- Experimental program(1)
Example Embodiment
[0076]In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings.
[0077]figure 1 It is the overall structure diagram of the bottom line algorithm of the present invention based on the matching degree and the multi-strategy dynamic adjustment genetic algorithm for the layout of the vehicle cabin deck vehicle. It includes: obtaining the amphibious vehicle and the space size unit (1) that can be nested; clear the layout of the amphibious vehicle Constraint condition unit (2); determine the objective function unit of vehicle layout (3); vehicle coding and initial population generation unit (4); vehicle layout multi-strategy genetic algorithm fitness value calculation and storage of the optimal individual unit (5) ; Vehicle Layout Optimization Algorithm Stop Judgment Unit (6); Vehicle Layout Multi-strategy Genetic Algorithm Population Segmentation Unit (7); Vehicle Layout Multi-strategy Genetic Algorithm Population Individual Selection Unit (8); Vehicle Layout Multi-strategy Genetic Algorithm Individual Elite cross unit (9); mutation unit (10) that dynamically adjusts the mutation probability of vehicle nesting multi-strategy genetic algorithm; new population generation unit of vehicle nesting multi-strategy genetic algorithm (11); dynamic adjustment of vehicle nesting multi-strategy genetic algorithm Select the probability unit (12); the optimal layout drawing generating unit (13) of the vehicle layout based on the lowest horizontal line algorithm.
[0078]The specific connections between the units are as follows: vehicle coding and initial population generation unit (4) encodes the vehicles in the unit for obtaining amphibious vehicles and nestable space size (1), initializes the parameters, and randomly generates m individuals Ai, That is, m packing sequences form the initial population Q(0). When the constraint conditions and objective functions in the amphibious vehicle layout unit (2) and the vehicle layout objective function unit (3) are satisfied, the vehicle layout multi-strategy genetic algorithm fitness value calculation and storage of the optimal individual unit ( 5) Calculate the fitness value of the individual and save the best individual. The vehicle layout optimization algorithm stop judging unit (6) judges whether the algorithm has reached the maximum number of iterations, and if the maximum number of iterations is reached, the optimal individual is returned to the vehicle layout optimization algorithm based on the lowest horizontal line algorithm to generate the optimal layout drawing unit (13) The optimal individual is decoded, and the optimal layout map of the vehicle layout on the vehicle deck is obtained. If it is not reached, it enters the next unit vehicle layout multi-strategy genetic algorithm population segmentation unit (7). Vehicle nesting multi-strategy genetic algorithm individual elite crossover unit (9) adopts contemporary optimal individuals and vehicle nesting multi-strategy genetic algorithm population individual selection unit (8) to cross all individuals in the three subpopulations. Vehicle nesting multi-strategy genetic algorithm dynamically adjusts the mutation unit of the mutation probability (10) uses the vehicle nesting multi-strategy genetic algorithm population individual selection unit (8) in the third individual evaluation strategy and dynamic mutation probability selection vehicle nesting multi-strategy genetic The newly generated individuals in the algorithm individual elite cross unit (9) complete the mutation operation. The new population generation unit (11) of the vehicle nesting multi-strategy genetic algorithm will effectively evolve individuals to form a new population, and the dynamic adjustment selection probability unit (12) of the vehicle nesting multi-strategy genetic algorithm will generate a new population based on the vehicle nesting multi-strategy genetic algorithm The number of effective evolutionary individuals in each group in unit (11) is adjusted to the size of each subpopulation. Multi-strategy genetic algorithm for vehicle layout calculation and storage of the optimal individual unit (5) calculates the individual fitness value of the new population, updates the optimal individual, and enters the next unit of vehicle layout optimization algorithm stop judgment unit (6), Form a closed loop of the vehicle layout algorithm.
[0079]The military vehicles transported by the amphibious assault ship are mainly military vehicles of various types, such as tanks and armored vehicles. The enveloping rectangle is used in the layout to simplify them into rectangles of different sizes. Because there is a gap between the vehicles, at the same time Personnel and fire-fighting passages are required to form an empty area, and the length and width of the vehicle can be appropriately extended during layout. Therefore, the present invention uses an envelope rectangle to represent the vehicle.
[0080]The specific implementation of each unit is described as follows:
[0081]Unit for obtaining amphibious vehicle and nestable space size (1) Collect the nestable space on the vehicle deck and the length and width of the vehicle, extend the length and width of the vehicle appropriately, use the envelope rectangle to represent the vehicle, and collect the length of the envelope rectangle , Wide information.
[0082]The realization of the constraint condition unit (2) for the layout of amphibious assault ships is as follows:
[0083]The layout method of the vehicle deck vehicle in the form of enveloping rectangular parts has the following constraints:
[0084](a) Any two enveloping rectangular vehicles do not overlap and each part cannot exceed the boundary of the vehicle deck;
[0085](b) The side of the rectangular vehicle to be discharged shall be parallel to the side of the vehicle cabin;
[0086](c) The enclosed rectangular vehicle cannot rotate, and the front of the vehicle faces one end of the vehicle door;
[0087](d) Each type of layout vehicle has a maximum number of restrictions.
[0088]The present invention first establishes the corresponding mathematical model of the layout of the amphibious assault ship vehicle deck vehicle for the above constraints. For example, the horizontal axis is the X axis and the right direction is positive, and the vertical axis is the Y axis. And the upward direction is positive, thereby establishing a coordinate system, such asfigure 2 Shown. In this coordinate system, take the lower left corner of the rectangular deck as the zero point of the coordinate system. The length of the rectangular deck coincides with the Y axis, the width of the rectangular deck coincides with the X axis, and the length of the rectangular deck is H and the width is V. Now Express w types of vehicles in the form of enveloping rectangular parts {L1,L2,...,Lw} Discharged onto the rectangular deck of the vehicle cabin, where the number of i-th enveloped rectangular parts (i=1, 2,...w) is ni, Its length and width are hiAnd vi. Vehicle envelope rectangle RiThe vertex of the upper left corner is A(xi1,yi1), the coordinate B(xi2,yi2)Yes:
[0089]
[0090]Where: xi1Represents the vehicle envelope rectangle RiThe abscissa of the top left corner vertex, yi1Represents the vehicle envelope rectangle RiThe ordinate of the vertex of the upper left corner, xi2Represents the vehicle envelope rectangle RiThe abscissa of the vertex of the lower right corner, yi2Represents the vehicle envelope rectangle RiThe ordinate of the vertex of the lower right corner, i represents the envelope rectangle of the i-th nested vehicle.
[0091]Take another vehicle envelope rectangle Rj, The vertex of the upper left corner is A(xj1,yj1), the coordinate B(xj2,yj2), the constraint conditions for the enveloping rectangle layout of the vehicle cabin are as follows:
[0092]
[0093]Where: numrRepresents the number of r-th vehicle layouts, NrRepresents the maximum number of r-th vehicles.
[0094]Determine the vehicle layout objective function unit (3). The objective of the present invention is to park more military vehicles of different types on a limited matrix vehicle deck, so as to maximize the utilization of the amphibious vehicle layout area, and the vehicle layout objective function for:
[0095]
[0096]Among them: G represents the objective function of vehicle layout, H and V represent the length and width of the rectangular deck, hiAnd viIndicates the length and width of the envelope rectangle of the i-th vehicle.
[0097]Enveloping rectangle vehicle encoding and initial population generation unit (4) uses decimal encoding to genetically encode the n vehicle envelope rectangles in the unit (1) for obtaining amphibious assault vehicle and nestable space size unit (1), and a non-repeated decimal integer As the unique identification code of the enveloping rectangular vehicle, a set of emission sequence represents an individual, and the number sequence indicates the sequence of the layout of each rectangular part. A complete integer sequence corresponds to a feasible layout plan, namely Ai={R1,R2,...,Rn}, AiIndicates the i-th nesting sequence, which is the i-th individual. And initialize the parameters, randomly generate m individuals, that is, m nesting sequences to form the initial population Q(0)={A1,A2,...,Am}.
[0098]The fitness value of the genetic algorithm with multiple strategies for vehicle layout calculation and storage of the optimal individual unit (5) The individual fitness value function is:
[0099]
[0100]Among them: g represents the individual fitness value function, q≤n, q represents the last vehicle enveloping rectangle of the nesting sequence that can be arranged on the vehicle deck, calculates all individual fitness values, and saves the optimal individual.
[0101]The vehicle layout optimization algorithm stop judging unit (6) sets the maximum number of iterations, and sets the algorithm stop condition to whether the maximum number of iterations is reached. If the stopping condition is met, the optimal individual is returned to the vehicle layout based on the lowest level algorithm. The optimal layout generating unit (13) decodes the optimal individual to obtain the optimal layout of the vehicle on the deck of the vehicle; if the stopping conditions are not met, enter the next unit, vehicle layout, multi-strategy genetic algorithm population segmentation Unit (7).
[0102]Vehicle layout multi-strategy genetic algorithm population segmentation unit (7), defined level set, individual average fitness value will As a collection, Called the level set of g with respect to Q(0). Then the population segmentation strategy is performed, and the population of the tth generation of evolution is set as Q(t), and the population is arranged in descending order of fitness value to obtain the level set Indicates the average fitness value of the t-th generation individual, and write it down Is greater than or equal to The smallest individual position lt, With ltDivide the population into two sub-populations, located at ltAll previous individuals are recorded as excellent population HP(t), located at ltAll subsequent individuals are recorded as ordinary population LP(t).
[0103]The realization of the multi-strategy genetic algorithm population selection unit (8) for vehicle layout is as follows:
[0104]According to three different individual evaluation strategies and selection probabilities combined with roulette method, the population is divided into three sub-populations: TA0(t), TB0(t), TC0(t), where t represents the evolutionary algebra of the population, and vehicle layout The first individual evaluation strategy of multi-strategy genetic algorithm, that is, the individual fitness function is:
[0105]
[0106]The selection probability of the first individual evaluation strategy is α, which constitutes the sub-population TA0(t). The second individual evaluation strategy introduces the degree of difference and increases the diversity of the population. Individual AiAnd AjThe difference degree between E(i,j) is expressed as:
[0107]
[0108]among them: z represents the locus of the individual layout sequence, azi And azj Means AiAnd AjThe value of the individual's z position. So the second individual evaluation strategy is:
[0109]g′(Ai)=E(s,i)g(Ai)
[0110]Among them: s means removing individual A from the populationiOther individuals. The selection probability of the second individual evaluation strategy is β, which constitutes the subpopulation TB0(t). The third individual evaluation strategy of the multi-strategy genetic algorithm for vehicle layout is:
[0111]
[0112]among them: Indicates the average fitness value of an individual, Represents the standard deviation of population fitness, and ε represents the smallest non-negative integer. The selection probability of the third individual evaluation strategy is θ, which constitutes the sub-population TC0(t).
[0113]The realization of the individual elite crossover unit (9) of the multi-strategy genetic algorithm for vehicle layout is as follows:
[0114]Use the contemporary optimal individual to complete the crossover operation with all the individuals in the three subpopulations TA0(t), TB0(t), and TC0(t). If the selected individual is from the excellent population HP(t), quote the difference degree E(i,j) ), calculate the degree of individual difference between the two parents involved in the crossover, and compare it with u, a number of u∈[0.5,1], if E(i,j)
[0115]The implementation of the mutation unit (10) that dynamically adjusts the mutation probability of the multi-strategy genetic algorithm for vehicle layout is as follows:
[0116]Determine the mutation probability p of the genetic algorithm with multiple strategies for vehicle layoutt, The specific formula is:
[0117]
[0118]Where: gt max Represents the optimal fitness value of the individual in the t-generation population, Represents the average fitness value of individuals in the excellent population HP(t), k represents the coefficient, k>0, ptThe value range is [0,0.05].
[0119]According to the vehicle layout multi-strategy genetic algorithm, the third individual evaluation strategy and mutation probability PtSelect three new populations TA1(t), TB1(t), and TC1(t) to participate in the mutation, and perform block shift mutation. The specific process is: generate two independently and randomly between [1,n] Locus c and d, and another random number r, r∈[1,nd], the shift moves the r gene sequences starting from locus d to the back of locus c. After the mutation is completed, new populations are obtained: TA2(t), TB2(t), TC2(t). And judge whether there are duplicate individuals in the new subpopulation, and if there are duplicate individuals, then mutate until no duplicate individuals are generated.
[0120]The realization of the new population generation unit (11) of the multi-strategy genetic algorithm for vehicle layout is as follows:
[0121]The crossover process produces individuals with higher fitness values than the selected individuals in the parent and the mutation process produces individuals with higher fitness values than the original individual, called effective evolutionary individuals. All effective evolutionary individuals in the three populations are merged into The new population Q(t).
[0122]The realization of the dynamic adjustment selection probability unit (12) of the multi-strategy genetic algorithm for vehicle layout is as follows:
[0123]Suppose the number of effective evolutionary individuals produced in the three subpopulations TA2(t), TB2(t), and TC2(t) are respectively at, Bt, Ct, Where t represents the evolutionary generation of the population, combined to form a new population Q(t), the new population size is mt, Mt=at+bt+ct, Represents the population size of the t generation. In the early stage of population evolution, a larger selection probability is used to expand the search space and increase the probability of outstanding individuals. With the increase of evolutionary generation, the selection probability of three subpopulations should be appropriately reduced to avoid large populations. , Increase the amount of calculation. The selection probabilities of the three subpopulations are:
[0124]
[0125]Where: α0, Β0, Θ0Respectively represent the selection probabilities of the initial three populations TA0(t), TB0(t), and TC0(t).
[0126]Suppose that when the size of the new population is smaller than m after evolving to the t+1 generation, m-m is randomly generatedtIndividuals form a new population, and the selection probabilities of three sub-populations are increased. The selection probabilities of the three sub-populations are:
[0127]
[0128]
[0129]
[0130]The optimal layout generation unit (13) of the vehicle layout based on the lowest horizontal line algorithm considers the ratio of the width of the enveloping rectangle of the vehicle to be arranged to the lowest horizontal line width and the alignment of the enveloping rectangle of the vehicle with the height of the enveloping rectangles of the vehicles on both sides. The concept of vehicle layout matching degree u is proposed, and the vehicle envelope rectangle with the largest matching degree is selected to be discharged into the lowest horizontal line, so as to maximize the utilization of the rectangular deck area. Introduce the influence factor η of vehicle layout, and its expression is:
[0131]
[0132]Where: Vpre Indicates the width of the enveloping rectangular vehicle to be arranged, VlowIndicates the minimum horizontal line width.
[0133]Secondly, combine the alignment of the rectangular part that can be discharged into the vehicle and the height on both sides, such asimage 3 As shown, the front of the vehicle is marked and shaded, and the vehicle is surrounded by an envelope rectangle. The rectangle marked with the front indicates the envelope rectangle of the vehicle that has been discharged on the deck, and the rectangle without the front of the mark is the envelope rectangle of the vehicle to be discharged.image 3 Middle (a) means that the enveloping rectangle of the vehicle cannot be discharged into the lowest horizontal line Vlow, The matching degree u=0 at this time. (b) and (c) indicate that the envelope rectangle of the vehicle can be discharged into Vlow, But not aligned with the height on both sides of the lowest horizontal line, matching degree (d) and (e) indicate that the envelope rectangle of the vehicle can be discharged into Vlow, But only aligned with the lowest horizontal line, matching degree (f) Indicates that the enveloping rectangle of the vehicle can be discharged into Vlow, And align with the heights on both sides of the lowest horizontal line, the matching degree is
[0134]Figure 4 The specific flow diagram of the optimal layout generation unit based on the lowest horizontal line algorithm for vehicle layout, mainly includes the following steps:
[0135]S1: Initialize the set of the highest contour line of the rectangular deck of the vehicle cabin, which only includes the bottom boundary of the rectangular deck, which is also the lowest horizontal line, and enter the individual layout sequence.
[0136]S2: Select the first vehicle envelope matrix in the sequence of vehicle envelope rectangles to be arranged, traverse the set of highest contour lines, and select the horizontal line with the lowest height. If there are more than one that meets the requirements, select the horizontal line with the leftmost vertical position. If the current enveloping rectangle vehicle can be discharged into the lowest horizontal line, the current enveloping rectangle of the vehicle is discharged close to the leftmost end of the horizontal line, and the highest contour line set is updated at the same time; otherwise, go to S3.
[0137]S3: Search the remaining vehicle envelope rectangle sequence, calculate the matching degree u of the envelope rectangles in the remaining vehicle envelope rectangle sequence, if u≠0, select the vehicle envelope rectangle with the largest matching degree value and rank it in the lowest Horizontal line, swap the positions of the two sorts, and update the highest contour line set. If u=0, it means that there is no rowable vehicle envelope rectangle on the current lowest horizontal line, merge with the adjacent lowest contour line, and update the highest contour line set at the same time, and go to S2.
[0138]S4: Repeat the above process until the envelope rectangle of all platable vehicles is discharged, and the optimal layout of amphibious vehicles is obtained.
[0139]Take the US "Wasp-class" amphibious assault ship as an example. The vehicle deck layout space is 100 meters long and 20 meters wide. The vehicles involved in the layout are Type A, Type B, Type C, Type D, Type E and There are six types of F-type models, considering the need to leave a gap between the vehicles, and the need for personnel and fire passages to form an empty area. Therefore, the length and width of the vehicle can be appropriately extended during layout. The extended size is: A Type B is 10.80m long and 3.20m wide, Type B is 9.80m long and 3.50m wide, Type C is 8.90m long and 2.90m wide, Type D is 7.50m long and 2.40m wide, Type E is 6.70m long and 2.50m wide, F The length of the model is 5.20m and the width of 4m. The maximum number of layouts for the six models is 20, and the decimal code is used to number in order, that is, the code for type A is 1-20, the code for type B is 21-40, and the code for type C is 21-40. The code is 41-60, the code for the D model is 61-80, the code for the E model is 81-100, and the code for the F model is 101-120.
[0140]Compare the adaptive genetic algorithm of vehicle layout and multi-strategy dynamic adjustment of genetic algorithm fitness curve, set the initial parameters of traditional genetic algorithm, the number of iterations is 200, the initial population size is 100, the crossover probability is 0.6, the mutation probability is 0.1, and the multi-strategy dynamic The adjusted initial parameters of the genetic algorithm set the number of iterations to 200, the initial population size is 100, and the subpopulation selection probability α0= Β0=θ0=0.4, mutation probability p0= 0.1. Comparing the simulation results of the two genetic algorithms to solve the vehicle layout problem, the fitness value change curve of the two genetic algorithms is as followsFigure 5 withFigure 6 Shown. The deck utilization rate of the adaptive genetic algorithm and the multi-strategy dynamically adjusted genetic algorithm are 88.74% and 94.09%, respectively, and the calculation time is 219.37s and 167.91s, respectively. The layout is A, B, C, D, E The number of amphibious vehicles of type F and F are (12,10,11,15,14,11) and (5,18,15,9,15,14) respectively. The deck utilization rate is increased by 5.35% compared with the adaptive genetic algorithm. Compared with the adaptive genetic algorithm, the time is shortened by 51.46s. The multi-strategy dynamic adjustment genetic algorithm is used to solve the vehicle layout problem, and the global optimal solution can be obtained, and the algorithm converges quickly.
PUM


Description & Claims & Application Information
We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.