CN117216937A

CN117216937A - Three-dimensional boxing method based on mixed ant colony simulated annealing algorithm

Info

Publication number: CN117216937A
Application number: CN202310943089.4A
Authority: CN
Inventors: 李想; 袁锐波; 胡启明; 施涛; 陈坤
Original assignee: Kunming University of Science and Technology
Current assignee: Kunming University of Science and Technology
Priority date: 2023-07-30
Filing date: 2023-07-30
Publication date: 2023-12-12

Abstract

The invention discloses a three-dimensional boxing method based on a mixed ant colony simulated annealing algorithm. Firstly, packing three-dimensional goods to be loaded into a tower set by a tower heuristic method considering complete cutting constraint, namely reducing the three-dimensional boxing problem to a two-dimensional boxing problem; then searching an optimal packing scheme of two-dimensional goods to be loaded by combining an ant colony simulated annealing algorithm with two pheromones and three dilution coefficients with a two-dimensional projection point algorithm; and secondly, finding out a tower of each two-dimensional goods to be loaded in the tower set in an optimal boxing scheme of the two-dimensional goods to be loaded, correspondingly adjusting the placing posture and the placing position of the goods in the tower respectively, and finally outputting the adjusted tower set. The invention considers the complete cutting constraint, ensures the stability of goods during transportation and the convenience during unloading, combines the global searching capability and the local searching capability by the algorithm, and has higher boxing rate in the face of complex multi-specification goods.

Description

Three-dimensional boxing method based on mixed ant colony simulated annealing algorithm

Technical Field

The invention relates to a three-dimensional boxing method, in particular to a three-dimensional boxing method based on a mixed ant colony simulated annealing algorithm, which can meet the complete cutting constraint and has global searching capability and local searching capability on the premise of meeting various basic constraints, and belongs to the technical field of three-dimensional boxing.

Background

The loading and transportation of goods is a critical step in the logistics industry. At present, the packing process of goods is mostly completed by manpower, and the goods can be piled up according to experience when the types and the quantity of the goods are small, but under the condition of coping with the large quantity of the goods, a reasonable packing scheme is difficult to give, so that the space utilization rate of the container is low, and the transportation cost is increased. Moreover, under the large trend of 'automation' in the logistics industry, the intellectualization of the boxing scheme is urgently realized, so that the problem that the space utilization rate of the container is maximized by researching a reliable, stable and efficient boxing algorithm to solve the near-optimal boxing scheme is a critical problem.

The three-dimensional packing problem belongs to a classical combination optimization problem, and is also an NP-hard problem, namely, an optimal packing scheme cannot be solved within polynomial time complexity, if an accurate algorithm is used for solving the problem, only the three-dimensional packing problem with small types and numbers of cargoes can be solved, if the types and numbers of cargoes become large, the number of feasible solutions combined in a solution space can be exponentially increased, the phenomenon is called combination explosion, and thus the accurate algorithm is difficult to obtain the optimal packing scheme within reasonable time. Therefore, an optimization algorithm needs to be designed to solve the problem of three-dimensional boxing in a reasonable time.

At present, genetic algorithm, simulated annealing algorithm or ant colony algorithm are combined with heuristic algorithm in the literature at home and abroad to solve the problems to obtain a certain result, but the genetic algorithm, simulated annealing algorithm or ant colony algorithm is adopted to have strong global searching capability or weak local searching capability, so that the loading rate of the container is not high when the boxing algorithm solves the complex multi-specification three-dimensional boxing problem; in addition, when the algorithms are designed at present, actual constraint (complete cutting constraint) is rarely considered, and convenience in cargo unloading and stability in transportation cannot be guaranteed, so that it is necessary to design a boxing algorithm which considers complete cutting constraint and has global searching capability and global searching capability.

Disclosure of Invention

The invention provides a three-dimensional boxing method of a mixed ant colony simulated annealing algorithm considering complete cutting constraint, which solves the problems of low loading efficiency, instability in cargo transportation and inconvenience in unloading caused by the fact that the complete cutting constraint is not considered, global searching capability and local searching capability cannot be considered in the existing algorithm. When the complete cutting constraint is processed, solving by adopting a heuristic algorithm of a 'tower' considering the constraint; for the problem that the existing optimization algorithm cannot achieve the global and local searching capability, an ant colony simulated annealing algorithm containing two pheromones and three pheromone dilution coefficients is adopted for solving the problem.

The technical scheme adopted for solving the technical problems is as follows: a three-dimensional boxing method based on a mixed ant colony simulated annealing algorithm comprises the following specific steps:

step 1: basic information of the container is read and a coordinate system is established using the basic information.

Step 2: and reading the basic information of the three-dimensional goods to be loaded.

Step 3: and stacking and packing the three-dimensional goods to be loaded up and down into a tower set by adopting a heuristic algorithm considering complete cutting constraint 'tower', namely reducing the three-dimensional boxing problem to a two-dimensional boxing problem, taking the contact surface of each 'tower foundation' and the bottom surface of the container as the two-dimensional goods to be loaded, and classifying the two-dimensional goods to be loaded according to the same length and width.

Step 4: and searching an optimal boxing scheme of the two-dimensional goods to be packaged by adopting an ant colony simulated annealing algorithm containing two pheromones and three pheromone dilution coefficients and a two-dimensional loading point boxing algorithm.

Step 5: and correspondingly outputting the obtained optimal boxing scheme of the two-dimensional goods to be packaged to the final boxing scheme of the three-dimensional goods to be packaged.

Specifically, the basic information of the container to be read in the step 1 includes length, width, height, volume and left rear lower corner point, which are respectively recorded as length L, width W, height H, volume V and left rear lower corner point as O, wherein length > width; and a coordinate system is established by taking the length as the X axis, the width as the Y axis, the height as the Z axis and the point O as the origin.

Specifically, the basic information of the goods to be read in step 2 includes the number of types of the goods, the number of each type of the goods, the length, width, height and volume of each type of the goods and the allowable placement posture of each type of the goods, and the number of the goods is recorded as m and the number of the goods is recorded as n _i I=1, 2,3, …, m, each cargo length l _i Width w _i Height is h _i V of volume _i Wherein the length is>The width and the allowable pose of each cargo are designated as P _i And P is _i Some and only the following formulae are satisfied:

specifically, for the heuristic algorithm of "tower" described in step 3, which considers the complete cut constraint, there are the following specific steps:

step 3.1: and sequencing the remaining three-dimensional cargoes to be loaded according to the largest surface area which can be provided by each cargo.

Step 3.2: taking the first ordered cargo in the step 3.1 as a tower foundation, and taking the maximum surface of the first ordered cargo as a bearing surface of the subsequent cargo.

Step 3.3: the subsequent cargoes are loaded according to the arrangement sequence of the step 3.1, in order to meet the complete cutting constraint, the cargoes can only be placed on a single cargoes each time, a weight function is selected according to the three-dimensional posture placing space, and the placing posture with the largest weight and the remaining space are selected for placing.

And 3.4, after the current goods are put in, the residual space selected in the step 3.3 is adaptively selected to be divided by one of two space dividing modes, each dividing mode divides the residual space into three subspaces (an upper subspace, a front subspace and a right subspace), and the three subspaces are incorporated into a residual space set.

Step 3.5: the space in the remaining space set is tried to be combined under the condition that the complete cutting constraint is met. And repeating the steps 3.3-3.5 until any goods cannot be put into the tower, and executing the step 3.6.

Step 3.6: and (3) rejecting the goods loaded into the tower, updating the residual goods, loading the tower into the tower set data set, and repeatedly executing the steps 3.2-3.6 until the residual goods are not found, and executing the step 3.7.

Step 3.7: extracting the contact surface between the tower foundation in the tower set and the bottom surface of the container, classifying the two-dimensional goods to be loaded according to the same length and width, and preparing for the subsequent algorithm.

Further, the selection of the current attitude and the remaining space of the goods to be loaded in the step 3.3 comprises the following specific steps:

And 3.3.1, respectively attempting to load the current loaded cargoes into all the residual spaces in the tower in six postures, and recording the residual spaces which can be put in and the states of each residual space which can be put in which the cargoes to be loaded can be put in.

And 3.3.2, selecting a weight function according to the three-dimensional gesture placement space to calculate the weight value of each placement condition.

Three-dimensional gesture placement space selects a weight function:

TASLWF _kj ＝-(RL _j -a _k +α)×(RW _j -b _k +α)×(RH _j -c _k +α) (2)

wherein: TASLWF _kj Selecting a weight value for a three-dimensional gesture placement space when the current loaded goods are placed in the j-th placeable residual space in the k-th placeable gesture; RL (RL) _j ，RW _j ，RH _j The j-th length of the rest space along the X axis, the Y axis and the Z axis; a, a _k ，b _k ，c _k The k-th placeable gesture in the six gestures for loading cargoes at present is respectively along the lengths of an X axis, a Y axis and a Z axis; alpha is a correction parameter.

And 3.3.3, selecting the placing gesture of the current loaded goods and the remaining space under the condition of highest weight value for placing.

Specifically, the following specific steps are selected for the two division modes of the residual space in the step 3.4:

step 3.4.1, dividing the loaded residual space in two ways, wherein each way divides the residual space into an upper side subspace, a front side subspace and a right side subspace, and the first way of dividing: the coordinates of the left rear lower corner point of the upper subspace are (x _R ，y _R ，z _R +c _R ) And a spatial length a along the X, Y and Z axes, respectively _R 、b _R And RH (relative humidity) _R -z _R -c _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the front subspace are (x _R +a _R ，y _R ，z _R ) And a spatial length RL along the X-axis, the Y-axis, and the Z-axis, respectively _R -a _R 、RW _R And RH (relative humidity) _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the right subspace are (x _R ，y _R +b _R ，z _R ) And a spatial length a along the X, Y and Z axes, respectively _R 、RW _R -b _R And RH (relative humidity) _R The method comprises the steps of carrying out a first treatment on the surface of the The second way of segmentation: the coordinates of the left rear lower corner point of the upper subspace are (x _R ，y _R ，z _R +c _R ) And a spatial length a along the X, Y and Z axes, respectively _R 、b _R And RH (relative humidity) _R -z _R -c _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the front subspace are (x _R +a _R ，y _R ，z _R ) And a spatial length RL along the X-axis, the Y-axis, and the Z-axis, respectively _R -a _R 、b _R And RH (relative humidity) _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the right subspace are (x _R ，y _R +b _R ， _R z) And a spatial length RL along the X-axis, the Y-axis, and the Z-axis, respectively _R 、RW _R -b _R And RH (relative humidity) _R Whereinx _R 、y _R And z _R For the coordinates of the lower left-rear point of the residual space selected in step 3.3, RL _R 、RW _R And RH (relative humidity) _R The remaining spaces selected in step 3.3 are along the spatial lengths of the X-axis, Y-axis and Z-axis, a, respectively _R 、b _R And c _R The spatial lengths along the X-axis, Y-axis and Z-axis, respectively, in the placement attitude selected in step 3.3 for the current load.

Step 3.4.2, comparing the bottom area of the front side subspace with the bottom area of the right side subspace of the first division method to obtain a larger sub-area S _maxone 。

S _maxone ＝max(((RL _R -a _R )×RW _R ),(a _R ×(RW _R -b _R ))) (3)

Step 3.4.3 comparing the bottom area of the front subspace with the bottom area of the right subspace in the first division to select a larger sub-area S _maxtwo 。

S _maxtwo ＝max(((RL _R -a _R )×b _R ),(RL _R ×(RW _R -b _R ))) (4)

Step 3.4.4, comparison S _maxone And S is equal to _maxtwo If S is of the size of _maxone The first residual spatial division mode is selected for large and vice versa.

And step 3.4.5, the three segmented subspaces are incorporated into a residual space set.

Specifically, on the premise that the space in the residual space set in the step 3.5 meets the complete cutting constraint, the merging is attempted under two conditions, namely, under the complete cutting constraint condition, the bottom surface of the merged residual space cannot be composed of the top surfaces of two cargoes, and the following specific steps are provided:

step 3.5.1, traversing each of the remaining spaces by six postures of the remaining unloaded goods, and dividing the remaining spaces into an unloading space and an unloading space.

Step 3.5.2, traversing all the remaining spaces by the non-loadable spaces, and trying to find the remaining spaces adjacent to the non-loadable spaces and flush with the bottom surface for merging when each non-loadable space traverses all the current remaining spaces; if found, try first kind to have the combination of the unnecessary abandoned space: checking whether two adjacent wide and high sides or long and high sides are equal, and if so, directly merging; if not, a second merging with redundant waste space is attempted: attempting to merge the spaces that are mutually strung together will traverse the six poses of the remaining unloaded goods to attempt to fill the strung space, if any remaining unloaded goods can be filled, then merge is performed, otherwise no merge is performed, and the unloaded space is rejected.

Specifically, the method for searching the two-dimensional packing scheme of the goods to be packed for the ant colony simulated annealing algorithm and the two-dimensional loading point packing algorithm which contain two pheromones and three pheromone dilution coefficients in the step 4 comprises the following specific steps:

and 4.1, taking the long and wide surfaces of the container as a loading plane of the two-dimensional goods to be loaded, wherein the coordinate system at the moment only has an X axis, a Y axis and an origin point O.

And 4.2, regarding each two-dimensional goods to be loaded as a node in the ant colony simulated annealing algorithm, and regarding the node where each ant walks as a boxing scheme.

Step 4.3, initializing and selecting two kinds of information, namely a follow-up cargo type pheromone and a preferred cargo type pheromone, wherein the cargo type pheromone is the pheromone concentration between each node, and the preferred cargo type pheromone selects the pheromone of the first passing node for each ant; initializing three different sizes of pheromone dilution coefficients: ρ ₁ ＝0,3ρ ₂ ＝0.35,ρ ₃ =0.4; calculating the similarity H between different types of two-dimensional cargoes to be loaded through a similarity function _ij The method comprises the steps of carrying out a first treatment on the surface of the Initializing the iteration times of the ant colony algorithm.

Similarity function:

wherein: h _ij For the ith two-dimensional goods to be loaded and the jth two-dimensional goods to be loadedSimilarity between goods; sl (S.L) _i And Sl _j The long sides of the ith two-dimensional goods to be loaded and the jth two-dimensional goods to be loaded are respectively; sw, sw _i And Sw _j The short sides of the ith two-dimensional goods to be loaded and the jth two-dimensional goods to be loaded are respectively;

and 4.4, determining a first-choice node and a subsequent node, namely a packing scheme, of each ant by using the probability function of the first-choice cargo type and the probability function of the selected subsequent cargo type according to the current two informativeness concentrations.

First choice cargo type probability function:

in the method, in the process of the invention,representing the probability that the kth ant takes the ith two-dimensional goods to be loaded as the preferred goods type; />The first-choice cargo type pheromone concentration of the i-th two-dimensional cargo to be loaded is the t-th generation.

Selecting a subsequent cargo type probability function:

wherein:representing the probability of selecting the j-th two-dimensional to-be-boxed after selecting the i-th two-dimensional to-be-loaded goods for loading by the ant k at the t-th generation; />Representing the t generation, and after the ith two-dimensional goods to be loaded are placed on ants k, remaining two-dimensional goods to be loaded are not placed; τ _ij And (t) represents the concentration of pheromone between the ith two-dimensional goods to be loaded and the jth two-dimensional boxes to be loaded.

And 4.5, decoding and solving each two-dimensional goods to be packaged placement point, placement gesture and area utilization rate in the boxing scheme represented by each ant by using a two-dimensional loading point boxing algorithm, wherein the two-dimensional goods to be packaged have only two placement gestures: the first is long along the X axis and short along the Y axis; the second is that the long side is along the Y axis and the short side is along the X axis; the packing scheme represented by each ant represents the packing sequence of the two-dimensional goods to be packed, namely the two-dimensional goods to be packed of the same type represented by the current node are completely loaded, and the loading of the two-dimensional goods to be packed of the same type represented by the next node is carried out.

And 4.6, sequencing each ant according to the area utilization rate from large to small, and selecting the first 30% of ants to perform simulated annealing operation respectively so as to avoid sinking into local optimum.

And 4.7, mixing the new ant population obtained after simulated annealing with the original ant population selected to be simulated annealed in the step 4.6, sequencing according to the area utilization rate from large to small, and selecting the first 50% of ants in the mixed ant population to replace the first 30% of ants in the original ant population, thereby ensuring that the number of ants in the ant population is unchanged.

And 4.8, dividing ants in the current ant group into three categories according to the area utilization rate (30% of the ants in front of the area utilization rate is a first category, 31% -70% of the ants in front of the area utilization rate is a second category, and 71% -100% of the ants in front of the area utilization rate is a third category), and respectively updating two pheromones.

First box type pheromone update:

wherein:for the t generation, the area utilization rate of the ant k (if the first node of the ant is i, the pheromone of the node is increased); sigma is a proportion parameter, M (t) is the first 30% of ant population of the t th generation of filling rate row; an ant population with a filling rate of 31% -70% in the t th generation of K (t); n (t) th generation of paddingThe filling rate is 71-100% of ant population;

Selecting a subsequent bin type pheromone update:

wherein:for the t generation, the area utilization of ant k (if the ant passes through node i and then passes through node j, the pheromone between the two nodes is increased).

And 4.9, comparing the current-generation optimal ant individual with the optimal historical area utilization, and if the current-generation optimal ant is larger than the ant individual with the optimal historical area utilization, updating the current-generation optimal ant individual as the optimal historical ant individual, and keeping the boxing scheme of the current-generation optimal ant individual and the ant individual, and vice versa.

And 4.10, repeatedly executing the steps 4.4-4.9 until the iteration termination times are reached, and outputting the boxing scheme of the historic optimal ant individual as an optimal boxing scheme.

Specifically, the two-dimensional loading point boxing algorithm in the step 4.5 comprises the following specific steps:

step 4.5.1, taking a point O at the left rear lower corner of the original container as an initial two-dimensional loading point, wherein the loading points all have corresponding loading areas; the loading area of the initial two-dimensional loading point is the bottom surface of the container, namely the long and wide surfaces of the container.

Step 4.5.2, selecting a placement scheme with a large weight value according to a weight function selected by a two-dimensional loading point when the two-dimensional goods to be loaded are placed, wherein the placement scheme comprises an optimal boxing gesture of the goods to be loaded and an optimal loading point to be placed; and when the goods are placed, the left rear corner point of the goods in the optimal placing posture is attached to the loading point for placing.

The two-dimensional gesture loading point selects a weight function:

SASLWF _kj ＝-(ZL _j -d _k +α)*(ZW _j -e _k +α) (10)

wherein: SASLWF _kj Selecting a weight value for a three-dimensional attitude space when the current two-dimensional loaded cargo is put in the jth placeable loading point in the kth placeable attitude; ZL (ZL) _j ，ZW _j The j-th loading point can be respectively arranged along the lengths of the X axis and the Y axis; d, d _k ，e _k The kth placeable pose of the two poses of the current two-dimensional load is along the length of the X-axis and the Y-axis, respectively.

Step 4.5.3, after the optimal placement point is placed into the cargo, two new loading points are generated in the X-axis direction and the Y-axis direction, namely, the X-axis loading point and the Y-axis loading point.

Step 4.5.4, judging whether the newly placed goods are shielded for the loading area corresponding to the unselected loading point; if so, modifications are made and vice versa.

Step 4.5.5, trying to expand the loading areas corresponding to the newly generated two loading points, namely, trying to project the expansion areas along the X-axis negative direction by the new loading points along the X-axis direction, and expanding the areas along the X-axis positive direction by the corresponding loading areas; the new loading point in the Y-axis direction is projected along the negative direction of the Y-axis to expand the area, the corresponding loading area is expanded along the positive direction of the Y-axis, and if the new loading point after the area expansion is repeated with the existing loading point, the new loading point is deleted, otherwise, the new loading point is reserved.

Step 4.5.6, updating the new loading points reserved after the area expansion is tried into a loading point set; and (3) removing the loaded cargoes, updating the remaining unloaded cargoes, and repeating the steps 4.5.2-4.5.6 until the bottom surface of the container is not placed or no cargoes are to be loaded.

Further, the simulated annealing operation for the first 30% of the ant population in step 4.6 includes the following steps:

step 4.6.1, initializing related parameters of a simulated annealing algorithm: the number of iterations of the inner and outer loops, the initialization temperature parameter, and the selection probabilities of the three sequence transformation structures (the probabilities of the single-point cross structure, the double-point cross structure, and the inverse transformation structure being selected are 40%, and 20%, respectively).

Single point cross structure: a point is randomly inserted into the original node sequence, and then the node sequences before and after the point are interchanged.

Double-point cross structure: two points are randomly inserted into the original node sequence, and then the front sequence from the front insertion point to the beginning of the node sequence and the rear sequence from the rear insertion point to the end of the node sequence are interchanged.

The reverse structure is as follows: two points are randomly inserted into the original sequence, and the node sequence between the two points is reversed.

And 4.6.2, replacing the node arrangement sequence of the ants according to three sequence transformation structures by the ants which are subjected to the simulated annealing algorithm currently.

And 4.6.3, decoding the ants with the replaced node sequences by a two-dimensional loading and boxing algorithm, and calculating the loading scheme and the area utilization rate of each two-dimensional to-be-boxed.

Step 4.6.4, comparing whether the area utilization rate of the ant before and after the node sequence transformation is increased, if so, reserving the ant, if so, attempting to reserve the node sequence represented by the ant with low area utilization rate after the sequence transformation by using a Metropolis criterion according to the current temperature.

And 4.6.5, comparing the area utilization rate of the preserved ant node sequence with that of the ant historical optimal node sequence, and taking the preserved ant node sequence as a new historical optimal node sequence of the ant if the area utilization rate of the ant node sequence is larger than that of the historical optimal node sequence, and vice versa.

And 4.6.6, taking the current ant node sequence as the current ant individual, repeating the steps 4.6.2-4.6.6 until the internal iteration is finished, updating the temperature according to a temperature updating formula, and executing the step 4.6.7.

Temperature update formula:

T＝T ₀ ×σ (11)

wherein: t is the updated temperature, T ₀ For the current temperature, σ is the proportional parameter (0<σ<1)。

Step 4.6.7, continuously executing steps 4.6.2-4.6.7 until the external circulation is finished, outputting the current historical optimal node sequence of the ant individual, storing the current historical optimal node sequence into a new ant population, and executing step 4.6.8.

And 4.6.8, eliminating the ant individuals subjected to simulated annealing in the original input ant population, selecting new ant individuals from the original input ant population as current ant individuals, and repeating the steps 4.6.1-4.6.8 until the original ant population has no ant individuals, and outputting the new ant population in the step 4.6.7.

Specifically, outputting the final three-dimensional to-be-loaded cargo packing scheme corresponding to the obtained two-dimensional to-be-loaded cargo optimal packing scheme in the step 5, wherein the method comprises the following steps of:

step 5.1, for each cargo in the optimal packing scheme of the two-dimensional cargo to be packed generated in step 4, finding a tower foundation and a tower corresponding to each cargo in the tower set generated in step 3;

and 5.2, according to the placing gesture and loading point of each cargo in the two-dimensional optimal cargo loading scheme, integrally adjusting the placing gesture and the placing position of the corresponding tower foundation and the cargo in the tower, and storing the placing gesture and the placing position of the adjusted tower foundation and the cargo in the tower.

And 5.3, calculating the overall space utilization rate of the container, and outputting a three-dimensional packing scheme.

Compared with the prior art, the invention has the beneficial effects that: packing three-dimensional cargoes to be packed into a tower set by taking a heuristic method of a tower with complete cutting constraint into consideration, extracting the tower bases of the towers from the bottom surface contacted with the bottom surface of a container to be packed into two-dimensional cargoes, searching an optimal packing scheme of the two-dimensional cargoes by an ant colony simulated annealing algorithm and a two-dimensional packing point algorithm which are designed with two pheromones and three pheromone dilution coefficients, finding a corresponding tower in the tower base of each two-dimensional cargoes in the obtained optimal packing scheme of the two-dimensional cargoes, adjusting the position and the posture of the corresponding tower according to the position and the posture of each two-dimensional cargoes, and packing the adjusted tower so as to realize the loading of the whole three-dimensional container; therefore, the invention considers the complete cutting constraint and ensures the convenience of cargo unloading and the stability of cargo transportation; the ant colony algorithm with two pheromones and three pheromone dilution coefficients is designed to enhance the local searching capability, then the ant colony algorithm with the two pheromones and the three pheromone dilution coefficients is prevented from being wholly sunk into the local optimum condition due to too fast convergence by the simulated annealing algorithm, namely the global searching capability is enhanced, so that the ant colony simulated annealing optimization algorithm with the two pheromones and the three pheromone dilution coefficients is designed to give consideration to the global searching capability and the local searching capability, and the container is ensured to have higher space utilization rate when facing complex multi-specification goods loading.

Drawings

FIG. 1 is an overall flow chart of an algorithm of the overall invention;

FIG. 2 is a schematic diagram of a three-dimensional coordinate system of a container;

FIG. 3 is a schematic view of six pose of each three-dimensional cargo;

FIG. 4 is a schematic illustration of a single "tower" that takes into account the full cut constraint;

FIG. 5 (a) is a schematic diagram of a first cargo boxing mode;

FIG. 5 (b) is a schematic diagram of a second cargo boxing mode;

FIG. 5 (c) is a schematic diagram of a third cargo boxing mode;

FIG. 5 (d) is a schematic diagram of a fourth cargo boxing mode;

FIG. 6 (a) is a schematic diagram of a first residual space division mode;

FIG. 6 (b) is a diagram illustrating a second residual space division scheme;

FIG. 7 (a) is a schematic diagram of merging without redundant waste space;

FIG. 7 (b) is a schematic diagram showing the combination of redundant waste space;

FIG. 8 is a schematic diagram of a new two-dimensional loading point and its loading area;

FIG. 9 (a) is a schematic view of a loading area for modifying an occluded X loading point;

FIG. 9 (b) is a schematic view of a modified loading area of an occluded Y loading point;

FIG. 10 (a) is a schematic view showing the expansion of the loading area of the X loading point to the negative X direction;

FIG. 10 (b) is a schematic view showing the expansion of the loading area of the X loading point to the positive X direction;

FIG. 10 (c) is a schematic view showing the expansion of the loading area of the Y loading point to the negative Y direction;

FIG. 10 (d) is a schematic view showing the expansion of the loading area of the Y loading point to the Y-axis positive direction;

FIG. 11 (a) is a schematic diagram of a delete duplicate X-axis load point;

FIG. 11 (b) is a schematic diagram of a delete duplicate Y-axis load point;

fig. 12 (a) is a schematic view of a single-point crossing structure;

FIG. 12 (b) is a schematic view of a two-point cross structure;

FIG. 12 (c) is a schematic diagram of the reverse structure;

FIG. 13 is a comparison of three algorithm iterations;

fig. 14 is a three-dimensional boxing effect graph of the HACSA algorithm.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the following description is taken in conjunction with the accompanying drawings and embodiments. The present invention will be described in further detail. It should be understood that the particular embodiments described herein are illustrative only and are not intended to limit the invention.

Example 1: as shown in fig. 1 to 14, a three-dimensional boxing method based on a mixed ant colony simulated annealing algorithm comprises the following steps (the whole flow chart is shown in fig. 1):

The specific implementation process is as follows: in order to solve the problems that the existing three-dimensional boxing algorithm does not consider actual constraint (complete cutting constraint) and cannot consider local searching capability and global searching capability, the three-dimensional boxing method is used for converting the three-dimensional boxing problem into a two-dimensional boxing problem through a heuristic method of a 'tower' considering complete cutting constraint, then an optimal boxing scheme of two-dimensional cargos to be packaged is searched through an ant colony simulated annealing algorithm combining two pheromones and three dilution coefficients with a two-dimensional projection point algorithm, each two-dimensional cargos to be packaged is found in the optimal boxing scheme of the two-dimensional cargos to be packaged, a 'tower' corresponding to the 'tower container' is found, the goods placing posture and the placing position in the 'tower' are correspondingly adjusted, and finally the adjusted 'tower set' is output, namely the boxing scheme of the three-dimensional cargos to be packaged.

Further, the basic information of the container to be read in the step 1 includes length, width, height, volume and left rear lower corner point, which are respectively recorded as length L, width W, height H, volume V and left rear lower corner point as O, wherein length > width; and a coordinate system is established by taking the length as the X axis, the width as the Y axis, the height as the Z axis and the point O as the origin.

Further, the basic information of the goods to be read in step 2 includes the number of types of the goods, the number of each type of the goods, the length, width, height and volume of each type of the goods and the allowable placement posture of each type of the goods, and the number of the types of the goods is respectively recorded as m and the number of the goods is respectively recorded as n _i I=1, 2,3, …, m, each cargo length l _i Width w _i Height is h _i V of volume _i Wherein the length is>The width and the allowable pose of each cargo are designated as P _i And P is _i Some and only the following formulae are satisfied:

further, for the "tower" described in step 3 that considers the full cut constraint (a single "tower" that considers the full cut constraint is shown in FIG. 4), there are the following specific steps:

full cut constraint: for each box inside the container, the bottom surface must have the bottom surface of the container or the bottom surface of other boxes as a support and cannot be suspended, and it is also required that the blocks formed by stacking the boxes on each other can be cut into individual boxes by several horizontal and vertical planes, and these planes for cutting cannot intersect the boxes in the blocks. This constraint mainly ensures that the cargo is stable enough in practice and facilitates the unloading of the forklift, as in fig. 5 (a) and (b) are allowed boxes, while fig. 5 (c) and (d) are not allowed boxes.

Step 3.3: the subsequent cargoes are loaded according to the arrangement sequence of the step 3.1, in order to meet the complete cutting constraint, each loading can be only carried out on a single cargoes, and the placement gesture with the largest weight and the residual space are selected to be put according to the three-dimensional gesture space loading weight function.

Step 3.7: extracting the contact surface between the tower foundation in the tower set and the bottom surface of the container, classifying the two-dimensional goods to be loaded into a subsequent algorithm for preparation according to the same length and width.

Three-dimensional gesture placement space selects a weight function:

TASLWF _kj ＝-(RL _j -a _k +α)×(RW _j -b _k +α)×(RH _j -c _k +α) (2)

wherein: TASLWF _kj Selecting a weight value for a three-dimensional attitude space when the current loaded goods are put in the j-th placeable residual space in the k-th placeable attitude; RL (RL) _j ，RW _j ，RH _j The j-th length of the rest space along the X axis, the Y axis and the Z axis; a, a _k ，b _k ，c _k The k-th placeable gesture in the six gestures for loading cargoes at present is respectively along the lengths of an X axis, a Y axis and a Z axis; alpha is a correction parameter.

And 3.3.3, selecting the placing gesture and the residual space of the current loaded goods under the condition of highest weight value.

Further, the following specific steps are selected for the two division modes of the residual space in the step 3.4:

step 3.4.1, dividing the loaded residual space in two ways, wherein each way divides the residual space into an upper side subspace, a front side subspace and a right side subspace, and the first way of dividing: the coordinates of the left rear lower corner point of the upper subspace are (x _R ，y _R ，z _R +c _R ) And a spatial length a along the X, Y and Z axes, respectively _R 、b _R And RH (relative humidity) _R -z _R -c _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the front subspace are (x _R +a _R ，y _R ，z _R ) And a spatial length RL along the X-axis, the Y-axis, and the Z-axis, respectively _R -a _R 、RW _R And RH (relative humidity) _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the right subspace are (x _R ，y _R +b _R ，z _R ) And a spatial length a along the X, Y and Z axes, respectively _R 、RW _R -b _R And RH (relative humidity) _R The method comprises the steps of carrying out a first treatment on the surface of the The second way of segmentation: the coordinates of the left rear lower corner point of the upper subspace are (x _R ，y _R ，z _R +c _R ) And a spatial length a along the X, Y and Z axes, respectively _R 、b _R And RH (relative humidity) _R -z _R -c _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the front subspace are (x _R +a _R ，y _R ，z _R ) And a spatial length RL along the X-axis, the Y-axis, and the Z-axis, respectively _R -a _R 、b _R And RH (relative humidity) _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the right subspace are (x _R ，y _R +b _R ， _R z) And a spatial length RL along the X-axis, the Y-axis, and the Z-axis, respectively _R 、RW _R -b _R And RH (relative humidity) _R Wherein x is _R 、y _R And z _R For the coordinates of the lower left-rear point of the residual space selected in step 3.3, RL _R 、RW _R And RH (relative humidity) _R The remaining spaces selected in step 3.3 are along the spatial lengths of the X-axis, Y-axis and Z-axis, a, respectively _R 、b _R And c _R The selection in step 3.3 for the currently loaded cargoThe spatial lengths along the X-axis, Y-axis and Z-axis, respectively, in the resting position.

S _maxone ＝max(((RL _R -a _R )×RW _R )，(a _R ×(RW _R -b _R ))) (3)

S _maxtwo ＝max(((RL _R -a _R )×b _R )，(RL _R ×(RW _R -b _R ))) (4)

Further, on the premise that the space in the residual space set in the step 3.5 meets the complete cutting constraint, the merging is attempted under two conditions, namely, under the complete cutting constraint condition, the bottom surface of the merged residual space cannot be composed of the top surfaces of two cargoes, and the following specific steps are provided:

Further, the method for searching the two-dimensional packing scheme of the goods to be packed for the ant colony simulated annealing algorithm and the two-dimensional loading point packing algorithm which contain two pheromones and three dilution coefficients in the step 4 comprises the following specific steps:

Similarity function:

wherein: h _ij The similarity between the ith two-dimensional goods to be loaded and the jth two-dimensional goods to be loaded is the similarity; sl (S.L) _i And Sl _j The long sides of the ith two-dimensional goods to be loaded and the jth two-dimensional goods to be loaded are respectively; sw, sw _i And Sw _j The short sides of the ith two-dimensional goods to be loaded and the jth two-dimensional goods to be loaded are respectively;

First choice cargo type probability function:

Selecting a subsequent cargo type probability function:

First box type pheromone update:

wherein:for the t generation, the area utilization rate of the ant k (if the first node of the ant is i, the pheromone of the node is increased); sigma is a proportion parameter, M (t) is the first 30% of ant population of the t th generation of filling rate row; an ant population with a filling rate of 31% -70% in the t th generation of K (t); an ant population with the filling rate of 71-100% in the t th generation of N (t);

Selecting a subsequent bin type pheromone update:

Further, the two-dimensional loading point boxing algorithm in the step 4.5 comprises the following specific steps:

The two-dimensional gesture loading point selects a weight function:

SASLWF _kj ＝-(ZL _j -d _k +α)*(ZW _j -e _k +α) (10)

After the optimal placement point is placed into the cargo, two new loading points, i.e., an X-axis loading point and a Y-axis loading point, are generated in the X-axis direction and the Y-axis direction, as shown in fig. 8.

Step 4.5.4, judging whether the newly placed goods are shielded for the loading area corresponding to the unselected loading point; if so, modifications are made and vice versa, as shown in FIGS. 9 (a) and (b).

Step 4.5.5, attempting to expand the loading area corresponding to the newly generated two loading points, i.e. attempting to expand the loading area by projection along the negative X-axis direction by the new loading point along the X-axis direction, as shown in fig. 10 (a); its corresponding loading area expands the loading area along the positive X-axis direction as shown in fig. 10 (b). The new loading point in the Y-axis direction is projected along the negative direction of the Y-axis to expand the loading area, as shown in FIG. 10 (c); the corresponding loading area expands the loading area along the positive Y-axis direction as shown in fig. 10 (d); if the new loading point after the area expansion is repeated with the existing loading point, it is deleted, otherwise, it is reserved, as shown in fig. 11 (a) and (b).

Single point cross structure: a point is randomly inserted into the original node sequence, and then the node sequences before and after the point are interchanged, as shown in fig. 12 (a).

Double-point cross structure: two points are randomly inserted into the original node sequence, and then the front sequence from the front insertion point to the beginning of the node sequence and the rear sequence from the rear insertion point to the end of the node sequence are interchanged, as shown in fig. 12 (b).

The reverse structure is as follows: two points are randomly inserted into the original sequence, and the node sequence between the two points is reversed, as shown in fig. 12 (c).

Temperature update formula:

T＝T ₀ x sigma (11) formula: t is the updated temperature, T ₀ For the current temperature, σ is the proportional parameter (0<σ<1)。

Step 4.6.7, continuously executing steps 4.6.2-4.6.7 until the external circulation is finished, outputting the current historical optimal node sequence of the ant individual, storing the current historical optimal node sequence into a new ant group, and executing step 4.6.8.

Further, outputting the final three-dimensional to-be-loaded cargo boxing scheme corresponding to the two-dimensional to-be-loaded cargo optimal boxing scheme obtained in the step 5, wherein the method comprises the following steps of:

In order to further illustrate the accuracy and reliability of the method, the invention proves the superiority in three-dimensional boxing by carrying out simulation analysis on two groups of cases by Matlab.

(1) Case one:

the method for randomly generating data provided by A genetic algorithm for the two-dimensional strip packing problem with rectangular pieces published by Bischoff in European Journal of Operational Research is used as a data source of the case, wherein 10 types of data including BR 1-BR 10 are provided, 25 groups of test data are concentrated in each type of data, the types of boxes in the 10 types of data are from weak to strong types, the performance of the algorithm can be reflected, and the method compares and analyzes the 3D-RSO algorithm, the mixed traditional simulated annealing algorithm (HSA algorithm) and the mixed traditional ant colony algorithm (HAC algorithm) in consideration of the algorithm of complete cutting constraint (the mixed ant colony simulated annealing algorithm of the invention is called as HACSA algorithm for short) in the residual space optimization algorithm which is known as efficient solution of three-dimensional boxing problem and is published in computer engineering and application; the test comparison results are shown in table 1:

TABLE 1

As can be seen from table 1:

1) Because the 3D-RSO is a heuristic algorithm for ordering the to-be-packaged boxes according to a certain rule, the direct boxing is not optimized by combining the meta-heuristic algorithm, and therefore, the space utilization rate is lower than that of the HSA, HAC and HACSA mixed heuristic algorithm.

2) With the increase of the types of cargoes to be loaded, the waste space of the container is increased, the loading rate of all algorithms is reduced, and the problem solving difficulty is increased.

3) From the average space utilization, the loading rate of the HACSA algorithm provided by the invention is higher than that of other algorithms, which shows that the algorithm provided by the invention has feasibility and is suitable for solving the problem of three-dimensional boxing.

(2) Case two:

carrying out example verification by adopting order information of goods provided by a certain company in Yunnan, wherein 15 kinds of goods are in total in the order, as shown in table 2; and on the basis, the HACSA algorithm, the HAC algorithm and the HSA algorithm of the present invention are compared and analyzed, and an iterative loading comparison chart of the three algorithms is provided, as shown in fig. 13. Test results of three algorithms: HACSA volume utilization of 94.05%, HAC volume utilization of 93.18%, HAS volume utilization of 92.32%; and gives a three-dimensional loading effect map of the HACSA of the present invention for this order, as shown in fig. 14.

TABLE 2

As shown in fig. 13, the HACSA algorithm of the present invention obtains the optimal loading rate in about 10 iterations, and the loading rates of the HSA algorithm and the HAC algorithm are not changed in about 60 generations and 140 generations, respectively, which indicates that the HACSA of the present invention has a higher convergence rate; the final loading rate of the HACSA algorithm is higher than that of the HAC algorithm and the HAS algorithm, so that the convergence accuracy of the HACSA algorithm is kept leading, and the HACSA algorithm is opposite to the other two algorithms; the HACSA algorithm ensures convergence accuracy and convergence speed by setting two pheromones and three pheromone dilution coefficients and adding a simulated annealing algorithm, and is more suitable for solving the problem of multi-specification three-dimensional boxing.

The boxing algorithm of the tower meeting the complete cutting constraint is adopted, so that the stability of cargo transportation and the convenience of cargo unloading are ensured, and the three-dimensional boxing problem is reduced to be a two-dimensional boxing problem; the two-dimensional boxing problem is solved by combining an ant colony simulated annealing algorithm provided with two types of pheromones and three dilution concentrations with a two-dimensional projection point algorithm, so that the overall searching capability and the local searching capability of the algorithm are considered, and good loading rate for the complex multi-specification three-dimensional boxing problem is ensured.

While the present invention has been described in detail with reference to the drawings, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention within the knowledge of those skilled in the art.

Claims

1. A three-dimensional boxing method based on a mixed ant colony simulated annealing algorithm is characterized by comprising the following steps of: the method comprises the following specific steps:

step 1: reading basic information of the container and establishing a coordinate system by using the basic information;

step 2: reading basic information of three-dimensional goods to be loaded;

step 3: stacking and packing three-dimensional goods to be loaded up and down into a tower set by adopting a heuristic algorithm considering complete cutting constraint 'tower', taking the contact surface of each 'tower foundation' and the bottom surface of a container as two-dimensional goods to be loaded, and classifying the two-dimensional goods to be loaded according to the same length and width;

step 4: adopting an ant colony simulated annealing algorithm containing two pheromones and three pheromone dilution coefficients and a two-dimensional loading point boxing algorithm to find an optimal boxing scheme of two-dimensional goods to be packaged;

2. The three-dimensional boxing method based on the mixed ant colony simulated annealing algorithm as claimed in claim 1, wherein the method comprises the following steps: the basic information of the container to be read in the step 1 comprises length, width, height, volume and left rear lower corner points, which are respectively marked as L, W, H, V and O, wherein the length is greater than the width; and a coordinate system is established by taking the length as the X axis, the width as the Y axis, the height as the Z axis and the point O as the origin.

3. The three-dimensional boxing method based on the mixed ant colony simulated annealing algorithm as claimed in claim 2, wherein the method comprises the following steps: the basic information of the goods to be read in the step 2 comprises the types of the goods, the quantity of each goods, the length, width, height and volume of each goods and the allowable placing posture of each goods, wherein the types of the goods are respectively recorded as m, and the quantity of each goods is n _i I=1, 2,3, …, m, each cargo length l _i Width w _i Height is h _i V of volume _i Wherein the length is>The width and the allowable pose of each cargo are designated as P _i And P is _i Some and only the following formulae are satisfied:

4. the three-dimensional boxing method based on the mixed ant colony simulated annealing algorithm as claimed in claim 1, wherein the method comprises the following steps: the heuristic algorithm of the "tower" in step 3, which considers the complete cut constraint, comprises the following specific steps:

Step 3.1: sequencing the remaining three-dimensional cargoes to be loaded according to the largest surface area which can be provided by each cargo;

step 3.2: taking the first ordered cargo in the step 3.1 as a tower foundation, and taking the maximum surface of the first ordered cargo as a bearing surface of the subsequent cargo;

step 3.3: the subsequent cargoes are loaded according to the arrangement sequence of the step 3.1, and can be only placed on a single cargoes each time, a weight function is selected according to the three-dimensional posture placing space, and the placing posture with the largest weight and the rest space are selected for placing;

step 3.4, after the current goods are put in, the residual space selected in the step 3.3 is adaptively selected to be divided into one of two space dividing modes, and each dividing mode divides the residual space into three subspaces: an upper side subspace, a front side subspace and a right side subspace, and incorporating the three subspaces into a set of remaining spaces;

step 3.5: under the premise of meeting the complete cutting constraint, the space in the residual space set tries to combine under the two conditions, and the steps 3.3-3.5 are repeatedly executed until any goods cannot be put in the tower, and the step 3.6 is executed;

step 3.6: rejecting the goods loaded into the tower, updating the residual goods, loading the tower into the tower set data set, repeatedly executing the steps 3.2-3.6 until the residual goods are not available, and executing the step 3.7;

5. The three-dimensional boxing method based on the mixed ant colony simulated annealing algorithm, according to claim 4, is characterized in that: the selection of the current attitude and the residual space of the goods to be loaded in the step 3.3 comprises the following specific steps:

step 3.3.1, respectively attempting to load the current loaded cargoes into all the residual spaces in the tower in six postures, and recording the residual spaces which can be put in and the states of each residual space which can be put in which the cargoes to be loaded can be put in;

step 3.3.2, selecting a weight function according to the three-dimensional gesture placement space to calculate a weight value under each placement condition;

three-dimensional gesture placement space selects a weight function:

TASLWF _kj ＝-(RL _j -a _k +α)×(RW _j -b _k +α)×(RH _j -c _k +α) (2)

wherein: TASLWF _kj Selecting a weight value for a three-dimensional gesture placement space when the current loaded goods are placed in the j-th placeable residual space in the k-th placeable gesture; RL (RL) _j ，RW _j ，RH _j The j-th length of the rest space along the X axis, the Y axis and the Z axis; a, a _k ，b _k ，c _k The k-th placeable gesture in the six gestures for loading cargoes at present is respectively along the lengths of an X axis, a Y axis and a Z axis; alpha is a correction parameter;

6. The three-dimensional boxing method based on the mixed ant colony simulated annealing algorithm, according to claim 4, is characterized in that: the two dividing modes of the residual space in the step 3.4 are selected by the following specific steps:

step 3.4.1, dividing the loaded residual space in two ways, wherein each way divides the residual space into an upper side subspace, a front side subspace and a right side subspace, and the first way of dividing: the coordinates of the left rear lower corner point of the upper subspace are (x _R ，y _R ，z _R +c _R ) And spatial lengths along the X, Y and Z axes, respectively, ofa _R 、b _R And RH (relative humidity) _R -z _R -c _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the front subspace are (x _R +a _R ，y _R ，z _R ) And a spatial length RL along the X-axis, the Y-axis, and the Z-axis, respectively _R -a _R 、RW _R And RH (relative humidity) _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the right subspace are (x _R ，y _R +b _R ，z _R ) And a spatial length a along the X, Y and Z axes, respectively _R 、RW _R -b _R And RH (relative humidity) _R The method comprises the steps of carrying out a first treatment on the surface of the The second way of segmentation: the coordinates of the left rear lower corner point of the upper subspace are (x _R ，y _R ，z _R +c _R ) And a spatial length a along the X, Y and Z axes, respectively _R 、b _R And RH (relative humidity) _R -z _R -c _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the front subspace are (x _R +a _R ，y _R ，z _R ) And a spatial length RL along the X-axis, the Y-axis, and the Z-axis, respectively _R -a _R 、b _R And RH (relative humidity) _R The method comprises the steps of carrying out a first treatment on the surface of the The coordinates of the lower left-rear point of the right subspace are (x _R ，y _R +b _R ， _R z) And a spatial length RL along the X-axis, the Y-axis, and the Z-axis, respectively _R 、RW _R -b _R And RH (relative humidity) _R Wherein x is _R 、y _R And z _R For the coordinates of the lower left-rear point of the residual space selected in step 3.3, RL _R 、RW _R And RH (relative humidity) _R The remaining spaces selected in step 3.3 are along the spatial lengths of the X-axis, Y-axis and Z-axis, a, respectively _R 、b _R And c _R The space lengths along the X axis, the Y axis and the Z axis respectively in the placement posture selected in the step 3.3 for the current loaded goods;

step 3.4.2, comparing the bottom area of the front side subspace with the bottom area of the right side subspace of the first division method to obtain a larger sub-area S _maxone：

S _maxone ＝max(((RL _R -a _R )×RW _R )，(a _R ×(RW _R -b _R ))) (3)

Step 3.4.3, dividing the first type of divisionThe bottom area of the front side subspace is compared with the bottom area of the right side subspace to select a larger sub-area S _maxtwo ：

S _maxone ＝max(((RL _R -a _R )×RW _R )，(a _R ×(RW _R -b _R ))) (4)

Step 3.4.4, comparison S _maxone And S is equal to _maxtwo If S is of the size of _maxone The first residual space division mode is selected by the big rule, and vice versa;

7. The three-dimensional boxing method based on the mixed ant colony simulated annealing algorithm, according to claim 4, is characterized in that: under the premise of meeting the complete cutting constraint, the space in the residual space set in the step 3.5 tries to combine under two conditions, namely under the complete cutting constraint condition, the bottom surface of the combined residual space cannot be composed of the top surfaces of two cargoes, and the method comprises the following specific steps:

Step 3.5.1, traversing each residual space by six postures of the residual unloaded cargoes, and dividing the residual spaces into a non-loadable space and a loadable space;

8. The three-dimensional boxing method based on the mixed ant colony simulated annealing algorithm as claimed in claim 1, wherein the method comprises the following steps: the ant colony simulated annealing algorithm and the two-dimensional loading point boxing algorithm which contain two pheromones and three pheromone dilution coefficients in the step 4 find a boxing scheme of two-dimensional goods to be packaged, and the method comprises the following specific steps of:

Step 4.1, taking the long and wide surfaces of the container as a loading plane of the two-dimensional goods to be loaded, wherein a coordinate system at the moment only comprises an X axis, a Y axis and an origin point O;

step 4.2, regarding each two-dimensional goods to be loaded as a node in an ant colony simulated annealing algorithm, and regarding the node where each ant walks as a boxing scheme;

step 4.3, initializing and selecting two kinds of information, namely a follow-up cargo type pheromone and a preferred cargo type pheromone, wherein the cargo type pheromone is the pheromone concentration between each node, and the preferred cargo type pheromone selects the pheromone of the first passing node for each ant; initializing three different sizes of pheromone dilution coefficients: ρ ₁ ＝0,3ρ ₂ ＝0.35,ρ ₃ =0.4; calculating the similarity H between different types of two-dimensional cargoes to be loaded through a similarity function _ij The method comprises the steps of carrying out a first treatment on the surface of the Initializing iteration times of an ant colony algorithm;

similarity function:

step 4.4, determining a first-choice node and a subsequent node, namely a boxing scheme, of each ant by using a first-choice cargo type probability function and a subsequent cargo type probability function according to the current two informativeness concentrations;

First choice cargo type probability function:

in the method, in the process of the invention,representing the probability that the kth ant takes the ith two-dimensional goods to be loaded as the preferred goods type;the concentration of the first-choice cargo type pheromone of the ith two-dimensional cargo to be loaded is t generation;

selecting a subsequent cargo type probability function:

wherein:representing the probability of selecting the j-th two-dimensional to-be-boxed after selecting the i-th two-dimensional to-be-loaded goods for loading by the ant k at the t-th generation; />Representing the t generation, and after the ith two-dimensional goods to be loaded are placed on ants k, remaining two-dimensional goods to be loaded are not placed; τ _ij (t) represents the concentration of pheromone between the ith two-dimensional goods to be loaded and the jth two-dimensional boxes to be loaded;

and 4.5, decoding and solving each two-dimensional goods to be packaged placement point, placement gesture and area utilization rate in the boxing scheme represented by each ant by using a two-dimensional loading point boxing algorithm, wherein the two-dimensional goods to be packaged have only two placement gestures: the first is long along the X axis and short along the Y axis; the second is that the long side is along the Y axis and the short side is along the X axis; the packing scheme represented by each ant represents the packing sequence of two-dimensional goods to be packed, namely, the two-dimensional goods to be packed of the same class represented by the current node are completely loaded, and the loading of the two-dimensional goods to be packed of the same class represented by the next node is carried out;

Step 4.6, sequencing each ant according to the area utilization rate from large to small, and selecting the first 30% of ants to perform simulated annealing operation respectively so as to avoid sinking into local optimum;

step 4.7, mixing the new ant population obtained after simulated annealing with the original ant population selected to be simulated annealed in step 4.6, and sorting according to the area utilization rate, wherein the first 50% of ants in the mixed ant population replace the first 30% of ants in the original ant population, so that the number of ants in the ant population is ensured to be unchanged;

step 4.8, dividing ants in the current ant group into three types according to area utilization rate: the first 30 percent of the area utilization rate is the first class, the 31 percent to 70 percent is the second class, and the 71 percent to 100 percent is the third class, and the updating of two kinds of pheromones is respectively carried out:

first box type pheromone update:

wherein:for the t generation, increasing the area utilization rate of the ant k, if the first node of the ant is i, increasing the pheromone of the node; sigma is a proportion parameter, M (t) is the first 30% of ant population of the t th generation of filling rate row; an ant population with a filling rate of 31% -70% in the t th generation of K (t); an ant population with the filling rate of 71-100% in the t th generation of N (t);

Selecting a subsequent bin type pheromone update:

wherein:for the t generation, if the ant passes through the node i and then passes through the node j, the area utilization rate of the ant k is increased, and the pheromone between the two nodes is increased;

step 4.9, comparing the current generation of the optimal ant individuals with the optimal historical area utilization rate, if the current generation of the optimal ant is larger than the historical optimal ant individuals, updating the current generation of the optimal ant individuals as the historical optimal ant individuals, and keeping the boxing scheme of the current generation of the optimal ant individuals, and vice versa;

9. The three-dimensional boxing method based on the mixed ant colony simulated annealing algorithm as claimed in claim 8, wherein the method comprises the following steps: the two-dimensional loading point boxing algorithm in the step 4.5 and the simulation annealing operation of the first 30% of ant population in the step 4.6 are respectively carried out, and the method comprises the following specific steps:

step 4.5.1, taking a point O at the left rear lower corner of the original container as an initial two-dimensional loading point, wherein the loading points all have corresponding loading areas; the loading area of the initial two-dimensional loading point is the bottom surface of the container, namely the long and wide surfaces of the container;

Step 4.5.2, selecting a placement scheme with a large weight value according to a weight function selected by a two-dimensional loading point when the two-dimensional goods to be loaded are placed, wherein the placement scheme comprises an optimal boxing gesture of the goods to be loaded and an optimal loading point to be placed; when the goods are placed, the left rear corner point of the goods in the optimal placing posture is attached to the loading point for placing;

the two-dimensional gesture loading point selects a weight function:

SASLWF _kj ＝-(ZL _j -d _k +α)*(ZW _j -e _k +α) (10)

wherein: SASLWF _kj For the current two-dimensional load, the load can be placed at the j-th loading point to be k-thA weight value is selected in a three-dimensional gesture space when the gesture is placed in; ZL (ZL) _j ，ZW _j The j-th loading point can be respectively arranged along the lengths of the X axis and the Y axis; d, d _k ，e _k The length of the kth placeable gesture along the X axis and the Y axis respectively in the two gestures of the current two-dimensional cargo loading;

step 4.5.3, after the optimal placement point is placed into the goods, two new loading points are generated in the X-axis direction and the Y-axis direction, namely an X-axis loading point and a Y-axis loading point;

step 4.5.4, judging whether the newly placed goods are shielded for the loading area corresponding to the unselected loading point; if so, then making modifications and vice versa;

step 4.5.5, trying to expand the loading areas corresponding to the newly generated two loading points, namely, trying to project the expansion areas along the X-axis negative direction by the new loading points along the X-axis direction, and expanding the areas along the X-axis positive direction by the corresponding loading areas; the new loading point in the Y-axis direction is projected to expand the area along the negative direction of the Y-axis, the corresponding loading area is expanded along the positive direction of the Y-axis, if the new loading point after the area expansion is repeated with the existing loading point, the new loading point is deleted, otherwise, the new loading point is reserved;

Step 4.5.6, updating the new loading points reserved after the area expansion is tried into a loading point set; removing the loaded cargoes, updating the remaining unloaded cargoes, and repeating the steps 4.5.2-4.5.6 until the bottom surface of the container is not placed or no cargoes are to be loaded;

in step 4.6, the simulated annealing operation is carried out on the first 30% of ant population respectively, and the steps are as follows:

step 4.6.1, initializing related parameters of a simulated annealing algorithm: the iteration times of the internal and external loops, the initialization temperature parameters and the selection probability of three sequence transformation structures: the probability of selecting the single-point cross structure, the double-point cross structure and the inverse structure is 40%,40% and 20% respectively;

single point cross structure: randomly inserting a point into the original node sequence, and exchanging the node sequences before and after the point;

double-point cross structure: randomly inserting two points into the original node sequence, and then exchanging a front sequence from the insertion point at the front position to the beginning of the node sequence with a rear sequence from the insertion point at the rear position to the end of the node sequence;

the reverse structure is as follows: randomly inserting two points in the original sequence, and reversing the node sequence between the two points;

Step 4.6.2, replacing the node arrangement sequence of the ants according to three sequence transformation structures by the ants which are subjected to the simulated annealing algorithm at present;

step 4.6.3, decoding the ants with the replaced node sequences by a two-dimensional loading and boxing algorithm, and calculating the loading scheme and the area utilization rate of each two-dimensional to-be-boxed;

step 4.6.4, comparing whether the area utilization rate of the ant before and after the node sequence transformation is increased, if so, reserving the ant, if so, attempting to reserve the node sequence represented by the ant with low area utilization rate after the sequence transformation by utilizing a Metropolis criterion according to the current temperature;

step 4.6.5, comparing the area utilization rate of the preserved ant node sequence with that of the ant historical optimal node sequence, and taking the preserved ant node sequence as a new historical optimal node sequence of the ant if the area utilization rate of the ant node sequence is larger than that of the historical optimal node sequence, and vice versa;

step 4.6.6, taking the current ant node sequence as the current ant individual, repeating the steps 4.6.2-4.6.6 until the internal iteration is finished, updating the temperature according to a temperature updating formula, and executing step 4.6.7;

temperature update formula:

T＝T ₀ ×σ (11)

wherein: t is the updated temperature, T ₀ Sigma is a proportional parameter and has a value greater than 0 and less than 1 for the current temperature;

step 4.6.7, continuously executing the steps 4.6.2-4.6.7 until the external circulation is finished, outputting the current historical optimal node sequence of the ant individual, storing the current historical optimal node sequence in a new ant group, and executing the step 4.6.8;

10. The three-dimensional boxing method based on the mixed ant colony simulated annealing algorithm as claimed in claim 1, wherein the method comprises the following steps: in the step 5, outputting the final three-dimensional to-be-loaded cargo boxing scheme correspondingly to the obtained two-dimensional to-be-loaded cargo optimal boxing scheme, wherein the method comprises the following steps of:

step 5.2, according to the placing gesture and loading point of each cargo in the two-dimensional optimal cargo loading scheme, adjusting the placing gesture and the placing position of the corresponding tower foundation and the cargo in the tower integrally, and storing the placing gesture and the placing position of the adjusted tower foundation and the cargo in the tower;