Bridge crane operation path optimization method based on improved genetic simulated annealing algorithm
By improving the genetic simulation annealing algorithm, combining segmented roulette strategy and cross-repair operation, the bridge crane operation path is optimized, and the problems of low efficiency and long no-load time in the existing technology are solved, and the problem of low bridge crane operation path optimization in the existing technology is achieved. Efficient steel plate inlet and outgoing operations.
Patent Information
- Application Number
- CN202411044359.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-11-15
AI Technical Summary
The existing bridge crane operation path optimization method has problems of low efficiency and long no-load time when dealing with in-store and out-of-store tasks. Especially when genetic algorithms are prone to local optimization and unstable solution efficiency, it is difficult to meet actual needs. .
Using an improved genetic simulation annealing algorithm, the bridge crane operation path optimization model is constructed by constructing steel plate overturning strategy and bridge crane operation path optimization model, combined with segmented roulette strategy, cross repair and mutation operations, the bridge crane operation path is optimized to avoid individual transitional reproduction and ineffective solutions. , improve diversity and solve quality.
It significantly improves the smoothness and efficiency of the bridge crane operation path, reduces no-load time, improves the overall efficiency of outgoing/input, avoids local optimal traps, and improves the stability and solution quality of the algorithm.
Smart Images

Figure CN118966499A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of logistics and warehousing, and relates to the optimization of a bridge crane operation path, and specifically to a bridge crane operation path optimization method based on an improved genetic simulated annealing algorithm. Background Art
[0002] In their previous management and operation, traditional steel structure processing and manufacturing enterprises did not pay much attention to the management of raw material warehouses. Generally, they only maintained normal operation. As a result, the management of raw material warehouses had problems such as messy production tasks, large differences in equipment basic conditions, complex sources of operators, and low efficiency due to active operations. With the gradual development of relevant technologies and research on automated warehouses, the automated warehousing management of the manufacturing industry has made great progress. As the starting point for manufacturing enterprises to carry out production activities, the automated steel plate warehouse is not only the main site for the production line to store steel plate raw materials, but also an important cargo transfer site. It undertakes the functions of raw material classification, warehousing, stacking, storage, picking, and supplying the required raw materials according to the production plan. Among them, the bridge crane path directly affects the time allocation of a single operation task, and thus affects the transportation efficiency.
[0003] A reasonable path planning scheme can reduce the time when the bridge crane is empty and improve transportation efficiency. According to the in / out operation process, when the bridge crane is not in operation, it is idle with the in-warehouse feeding position as the initial position; when performing in-warehouse operation, first lift the incoming steel plate from the feeding position, and then transport it to the designated stacking position to complete the stacking, repeating until the current truck is unloaded, and the in-warehouse task is completed; when performing out-warehouse operation, the bridge crane will move from the initial position to the stacking position where the steel plate to be out-warehouse is located. If there is a non-out-warehouse steel plate above the target stacking position, the stacking operation will be completed first, and then the target steel plate will be lifted to the RGV, and all the steel plates to be out-warehouse in the current warehouse area will be transported in turn to complete an out-warehouse task. At present, most of the crane scheduling problems are aimed at terminal container scheduling, and there is little research on the optimization scheduling of bridge cranes in automated warehouses.
[0004] Pang Long et al. optimized the picking path of warehouse stackers by integrating genetic algorithm with ant colony algorithm [see Pang Long, Lu Jingui. Optimization of picking path in automated stereoscopic warehouse based on ant colony genetic algorithm [J]. Computer Engineering and Science, 2012, 34(3): 148-151]; however, the optimization only considers a single task. When the warehousing and outbound tasks are carried out at the same time, the number of empty operations of the stacker increases. In addition, due to the problems of genetic algorithm being easy to fall into local optimality, unstable solution efficiency, and low robustness, it is necessary to make improvement plans for specific problems to make them fit the actual problem and improve the solution quality and efficiency. Summary of the invention
[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a bridge crane operation path optimization method based on an improved genetic simulated annealing algorithm, which effectively meets the bridge crane path planning problem and improves the efficiency of steel plate entry / exit.
[0006] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions to achieve it.
[0007] The bridge crane operation path optimization method based on the improved genetic simulated annealing algorithm provided by the present invention comprises the following steps:
[0008] S1 builds a steel plate stacking strategy;
[0009] S2 builds a bridge crane operation path optimization model; divides the bridge crane task into three parts: SC task, DC task and stacking task, and obtains the objective functions of the three parts respectively;
[0010] S3 solves the bridge crane operation path optimization model based on the improved genetic annealing algorithm to obtain the optimal operation path of the bridge crane; this step includes the following sub-steps:
[0011] S31 determines the initial parent population popm and the fitness of each individual in the initial parent population popm, and arranges them in descending order of fitness; each individual includes the bridge crane operation path, its operation time and the stacking plan;
[0012] S32 selects the parent population through a segmented roulette strategy to determine a number of individuals as chromosomes;
[0013] S33 calculates the crossover probability according to the Sigmoid function, and then randomly groups the selected chromosomes in pairs according to the crossover probability, and then performs a crossover operation on the two grouped chromosomes and repairs them;
[0014] S34 calculates the mutation probability based on the Sigmoid function, and then randomly selects the gene position on the chromosome for mutation operation based on the mutation probability to obtain the offspring population;
[0015] S35 calculates the fitness of each individual in the offspring population and compares it with the fitness of the corresponding individual in the parent population. According to the probability of accepting the solution, the individuals in the offspring population replace the corresponding individuals in the parent population.
[0016] S36 determines whether the maximum number of iterations has been reached. If so, the optimal individual is output; otherwise, the probability of accepting the solution is cooled down, and the process returns to step S32 to perform the next iteration until the maximum number of iterations is reached.
[0017] In the above step S1, the steel plate stacking strategy is constructed as follows:
[0018]
[0019] Among them, l j’max , l j’min and They represent the maximum, minimum and average lengths of the steel plates in position j' respectively; l jr represents the length of the rth blocking plate in stack position j; Y i′jr is the stacking decision variable, representing the upper r-th layer of blocking plate in the stack position j where the outgoing steel plate i' is located. i’jr =1, otherwise Y i’jr =0; t r is the unit blocking plate thickness; h j’ is the target stack height for stacking; (x j ,y j ) and (x j′ ,y j′ ) respectively represent the coordinates of the stack position j where the steel plate to be shipped is located and its stack position j' where it can be turned over;
[0020] The stacking time is determined based on the above strategy.
[0021] The stacking time is calculated according to the following formula:
[0022]
[0023] When R j >0:
[0024]
[0025] Among them, t(b j ,b j′ ) is the time taken by the bridge crane to move from stack position j to the target stack position j'; R j is the set of blocking plates on stack position j; t r is the unit blocking plate thickness; T up,j and T up,j′ are the operating time of the bridge crane for lifting and loading and unloading operations on stack positions j and j' respectively; H is the maximum height of the stack position; h j and h j’ V is the target stack height for outbound delivery and the target stack height for reverse delivery; up It is the vertical operating speed of the bridge crane.
[0026] In the above step S2, the bridge crane operation path optimization model constructed is as follows:
[0027]
[0028] I max =max(I,I′),I min =min(I,I′),b k ∈{bI ,b O};
[0029]
[0030] Among them, Z m / Z n is the no-load decision variable, which means that after the m / nth SC / DC task is completed, if the next step is to execute the outbound task, the value is 1, otherwise it is 0; T SC,m is the time required for the crane to perform the mth SC task; T DC,n The time required for the crane to perform the nth DC task; T Sj The time required to complete a stacking operation; Y i′jr is the stacking decision variable; R j is the set of blocking plates on stack position j; b m is the target stack position of the SC task; b 2n-1 and b 2n are the target stack positions for the DC task; I and I′ are the sets of steel plates for inbound and outbound respectively; t(b1, b2) is the time required for the crane to move from stack position 1 to stack position 2; t(b i ,b k ) is the crane running from stack position i to the entry / exit platform b k time; b I 、b O They are the parking space for incoming trucks and the unloading space for material in the steel plate storage area.
[0031] Then, the mathematical model for optimizing the operation path of the steel plate material warehouse bridge crane is as follows:
[0032]
[0033] In the above step S31, initially, M individuals are randomly selected to form an initial population, and the operation sequence corresponding to each individual is encoded using a Boolean matrix; at the same time, according to the steel plate stacking strategy and the bridge crane operation path optimization model, the bridge crane operation path, operation time and stacking plan for each individual are determined.
[0034] In the above step S32, according to the set selection rate p s , using the segmented roulette strategy to select the initial parent population, and select M×p s Individual.
[0035] In the specific implementation, the parent population is divided into k segments (the number of individuals in each segment is ), randomly select one of the segments and choose from them using roulette. The probability of selecting the first segment is p l for:
[0036]
[0037] Among them, f ξ is the fitness value of the ξth individual in the lth segment;
[0038] For the lth segment, use the random function rand(·) to generate a random number for each individual in the segment, and the random number is between (0,1); then select the corresponding individual according to the cumulative probability of selection; l = 1, 2, ..., k;
[0039] Then, choose another segment and repeat the above operation until the number of individuals selected is M×p s .
[0040] In the above step S33, the crossover probability p c Calculated according to the following formula:
[0041]
[0042] Among them, p c,max and p c,min is the preset maximum and minimum crossover probability; f i is the fitness value of the selected individual; f max 、f min 、f avg are the maximum, minimum and average fitness values of the population respectively;
[0043] Before performing the crossover operation, the target crossover point of the chromosome and one or more repeated individuals in the other sequence are determined, and set as the index target point. After the crossover is completed, the index target point is also exchanged.
[0044] In the above step S34, the mutation probability p m Calculated according to the following formula:
[0045]
[0046] Among them, p m,max and p m,min is the pre-set maximum and minimum mutation probability; f i is the fitness value of the selected individual; f max 、f min 、f avg are the maximum, minimum and average fitness values of the population, respectively.
[0047] In the above step S35, the probability of accepting a solution is calculated as:
[0048]
[0049] Among them, f irepresents the fitness of the ith individual in the parent population; f′ i is the fitness of the corresponding individual in the offspring population; T is the current temperature.
[0050] In the above step S36, it is determined whether the current number of iterations has reached the maximum number of iterations. If so, the individual with the largest fitness in the updated parent population is taken as the optimal individual and output, including the bridge crane operation path, operation time and stacking plan; if the maximum number of iterations has not been reached, the temperature is lowered by T←aT as the temperature of the next iteration, a is the set cooling coefficient, and then returns to step S32 to loop the operation until the output conditions are met.
[0051] Compared with the prior art, the steel plate storage stack allocation method based on the improved genetic simulated annealing algorithm provided by the present invention has the following beneficial effects:
[0052] 1) The present invention integrates the genetic algorithm and the simulated degradation algorithm and improves the method. By designing different adaptive operators (including segmented roulette strategy, cross repair operation, mutation operation, etc.), it can effectively meet the bridge crane path planning problem. The solved bridge crane operation path is relatively more streamlined, and the idle time and operation efficiency are greatly improved;
[0053] 2) The segmented roulette strategy designed in the present invention can avoid the over-reproduction of individuals, which is conducive to the maintenance of diversity;
[0054] 3) The present invention can avoid the problem of invalid solutions caused by sequence missing and duplication in the execution of digital sequence crossover operations through repair after crossover. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] Figure 1 This is a schematic diagram of the overall layout of the automated steel sheet warehouse;
[0056] Figure 2 Schematic diagram of the bridge crane operation path under different operation modes;
[0057] Figure 3 It is a flow chart of the method for allocating stacking positions for steel plates based on the improved genetic simulated annealing algorithm;
[0058] Figure 4 This is a schematic diagram of the steel plate stacking status;
[0059] Figure 5 This is a schematic diagram of the distribution of reversible stacking positions from a bird's-eye view;
[0060] Figure 6 This is a schematic diagram of the SC / DC task combination method;
[0061] Figure 7 It is a flowchart based on the improved genetic simulated annealing algorithm;
[0062] Figure 8 It is a schematic diagram of the coding method of the outbound / inbound operation sequence;
[0063] Fig. 9 This is a schematic diagram of the segmented roulette strategy;
[0064] Fig.10 Schematic diagram of chromosome crossover operation; (a) is the crossover operation, and (b) is the repair operation;
[0065] Fig.11 This is a schematic diagram of chromosome mutation operation;
[0066] Fig.12 It is a schematic diagram of the layout of a single reservoir area;
[0067] Fig.13 It is a schematic diagram of the target stack position information of the task;
[0068] Fig.14 Schematic diagram of the global operation path after the ISAGA algorithm optimization scheme; (a) corresponds to before optimization, and (b) corresponds to after optimization;
[0069] Fig.15 Comparison diagram of the iterative process obtained by different algorithms; among them, (a) corresponds to the global operation time optimization iteration diagram, and (b) corresponds to the optimization iteration diagram of the no-load operation. DETAILED DESCRIPTION
[0070] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.
[0071] Example 1
[0072] In order to explain in detail the steel plate storage stack allocation method based on the improved genetic simulated annealing algorithm provided by the present invention, the steel plate material warehouse is set as follows in this embodiment:
[0073] (1) The basic layout of the steel plate warehouse stacking area can be described as a rectangular area consisting of y rows and x columns of stacking positions and evenly divided into z equal blocks. The stacking position division, overall layout and coordinate setting are as follows: Figure 1 As shown, assuming that the first stack in the upper left corner of the entire warehouse is the coordinate origin (0,0), then the coordinates of stack j in the yth row and xth column located in the z storage area can be expressed as The geometric center distance between adjacent stacks in the x direction is ls , in the y direction is w s Since there are gaps between each stack, for the convenience of subsequent research, the warehouse can be abstracted as a (x×y) stack with a size of (l s × s ) grid area composed of rectangles.
[0074] In the process of making production plans, steel plate raw materials are often managed based on "project number". Therefore, in the order information list, the specifications and project numbers of incoming materials are the key attributes of stack allocation: Let the set of steel plates to be stored be I, then the length, width, thickness and project number of the i-th steel plate are (l i ,w i ,t i ,p i ). For the steel plates in stock, the average length of the steel plates on stack j is specified as l j , the average width is w j , the initial stacking height is h j The RGV track is set on the central axis of the material warehouse, and a rail car station is set on it every time it passes through a warehouse area, recorded as
[0075] (2) There is no overlap of tasks, i.e. one batch of tasks must be completed before the next batch can be started;
[0076] (4) Each storage area is equipped with a bridge crane, which can only lift one steel plate at a time and cannot carry out cross-area operations;
[0077] (5) The operating speed of the bridge crane and RGV is constant and slow, and their starting and braking time are ignored;
[0078] (3) The upper left point is set as the coordinate origin (0,0), and the incoming truck parking space b I =(0,y / 2) as the feed position of the storage area, RGV station b O =(x+1,y / 2) as the discharge position of the storage area;
[0079] (6)b j =(x j ,y j ) represents the coordinates of stack position j. The time t(b1,b2) taken by the bridge crane to move from b1 to b2 is:
[0080]
[0081] Where V x and V y are the running speeds of the bridge crane along the x and y directions respectively.
[0082] The operation mode of the bridge crane in the automated steel plate warehouse can also be divided into two types: SC and DC. The SC operation mode of the bridge crane refers to the individual handling of the steel plates in a sequence in a set of task sequences, that is, the bridge crane performs the task of taking a steel plate out of the warehouse or into the warehouse one by one each time, and its movement path is as follows: Figure 2 (a) shows; the DC operation mode of the bridge crane means that after lifting a steel plate to the designated stack, it will move to the stack where the next steel plate to be shipped out is located and perform a shipping task. Its movement path is as follows Figure 2 (b) as shown.
[0083] Based on the above steel plate warehouse setting, the steel plate storage stack allocation method based on the improved genetic simulated annealing algorithm provided in this embodiment is as follows: Figure 3 As shown, the following steps are included:
[0084] S1 constructs a steel plate stacking strategy.
[0085] Due to the special nature of the steel plate storage objects, high-level shelves cannot be set up in the steel plate warehouse. Instead, the steel plates are stored by stacking them layer by layer. When the target steel plate to be shipped is not at the top of the stack, that is, there is a blocking plate above it that is not to be shipped, it is necessary to reverse the stacking operation first. At this time, the stacking status of the target steel plate in the stack position is as follows: Figure 4 Therefore, the stacking principle for outbound steel plates is first proposed, and then the path optimization model of the bridge crane inbound and outbound operation is established based on this principle as a part of the inbound and outbound operation.
[0086] The stacking problem only occurs during the outbound operation, which means that the blocking plate on the target steel plate to be outbound will reselect the stacking position. Only after the blocking plate is moved to other stacking positions can the bridge crane lift the target steel plate out of the warehouse. Therefore, the stacking operation is actually an operation decision-making sub-process accompanying a single outbound operation. The time required to complete a stacking operation is T Sj The objective function is:
[0087]
[0088] When R j >0:
[0089]
[0090]
[0091] Among them, t(b j ,b j′ ) is the time taken by the bridge crane to move from stack position j to the target stack position j'; R j is the set of blocking plates on stack position j; t r is the unit blocking plate thickness; T up,j and T up,j′are the operating time of the bridge crane for lifting and loading and unloading operations on stack positions j and j' respectively; H is the maximum height of the stack position; h j and h j’ V is the target stack height for outbound delivery and the target stack height for reverse delivery; up T is the vertical running speed of the bridge crane. Sj The physical meaning of is the total time it takes for the bridge crane to lift a blocking plate, move it to the top of the target stack, lower it to place the blocking plate, and the reverse process of the above process. Through this stacking operation model, the basic principles of the stacking plan can be formulated:
[0092] (1) The selection of the target stack position should meet the basic constraints of the stack position allocation strategy and the basic operating principles of the material warehouse;
[0093] (2) When formulating a stacking plan, the nearest stacking location should be selected to improve work efficiency;
[0094] (3) In order to prevent the stacking operation from interfering with the subsequent outbound / inbound tasks, the stack containing the steel plates to be outbound should be avoided as the target stack for stacking, so as to reduce the complexity of the problem.
[0095] The constraint model is established based on the basic principle of stacking operation mentioned above. The condition that the blocking plate on the original stack j where the steel plate to be shipped is located can be turned over to the j' stack is as follows:
[0096]
[0097] Formula (5) indicates that the length of the blocking plate should match the target stacking position; j’max , l j’min and They represent the maximum, minimum and average lengths of the steel plates in position j' respectively; l jr represents the length of the rth blocking plate in stack position j; Y i′jr is the stacking decision variable, representing the upper r-th layer of blocking plate in the stack position j where the outgoing steel plate i' is located. i’jr =1, otherwise Y i’jr = 0; Formula (6) indicates that the stack height after stacking is completed The stack height limit H must not be exceeded.
[0098] The stacking operation should select the target stacking position nearby and design the stacking strategy such as Figure 5 The top view shown. The middle stack is the stack where the steel plate to be shipped is located, and the remaining numbers 1 to 8 represent the positions of the stacks that can be turned over and their selection priorities (the smaller the number, the higher the priority). The stack selection strategy of "column first, then row" is adopted, and the following constraints should be met:
[0099]
[0100] Equation (7) restricts the unit distance of the gantry crane movement not to exceed the limit value, (x j , y j ) and (x j′ , y j′ ) represent the coordinates of the stack position j where the steel plate to be stored out is located and its possible restacking position j', respectively.
[0101] S2 constructs an optimization model for the gantry crane operation path.
[0102] Assume that the optimal sequence for the gantry crane to execute a batch of inbound / outbound tasks is W = {I, I'}, where I = {i1, i2, i3,..., i r1} is the set of inbound tasks, and r1 is the number of inbound tasks; I' = {i1', i2', i3', …, i' r2} is the set of outbound tasks, and r2 is the number of outbound tasks. The i and i' in this sequence are arranged alternately and can be re-specified in pairs as W = (N DC , M SC ), N DC is the DC task queue N DC = {(i k1 , i' k1 ), (i k2 , i' k2 ), …, (i n , i' n )}, and at the same time, due to the different numbers of outbound and inbound tasks, the remaining I max - I min tasks form the SC task M SC . For example Figure 6 in, if I < I', then the first r1 items of I' are combined with I to form N DC , and the remaining r2 - r1 items of I' automatically form M SC . It is stipulated that the total time required for the gantry crane to complete a batch of inbound and outbound tasks in this sequence is:
[0103] T = T SC + T DC + T S (8);
[0104] The above equation divides the tasks of the gantry crane into three parts: SC tasks, DC tasks, and restacking tasks, and calculates their execution times respectively. Their objective functions are:
[0105]
[0106] Among them, Z m / Z nIt is an empty-load decision variable, indicating that after the m / nth SC / DC task is completed, if the next step is to execute the outbound task, the value is 1, otherwise it is 0; its connotation is that since the material flow path of the steel plate warehouse is unidirectional, after completing a DC task, if there are subsequent incoming steel plates, the bridge crane needs to move empty-load from the outflow position to the feed position to perform the incoming operation, and this part cannot be ignored when calculating T.
[0107] In summary, under the above constraints, the mathematical model for optimizing the operation path of the bridge crane in the automated steel plate warehouse is as follows:
[0108]
[0109] S3 solves the bridge crane operation path optimization model based on the improved genetic annealing algorithm and obtains the optimal operation path of the bridge crane.
[0110] This embodiment is based on the traditional genetic annealing algorithm, and improves the coding design method, chromosome crossover method, and chromosome mutation method to obtain an improved genetic annealing algorithm suitable for optimizing the operation path of the bridge crane. The bridge crane operation path optimization model is solved to obtain the optimal operation path of the bridge crane.
[0111] Improve the genetic annealing algorithm and first set the input and output.
[0112] The input includes: bridge crane parameters (such as the running speed of the bridge crane in the three coordinate directions), stack coordinates (that is, the stack position coordinates set in the entire warehouse area), initial stacking situation (that is, the steel plate stacking situation in the warehouse area before the task is carried out) and in / out steel plate information (including the target stack information of the in / out steel plates, etc.); the number of iterations Max_C.
[0113] The output includes: the optimal task sequence w in the population best Its operation time and stacking plan.
[0114] like Figure 7 As shown, this step includes the following sub-steps:
[0115] S31 determines the initial parent population popm and the fitness of each individual in the initial parent population popm, and arranges them in descending order of fitness; each individual contains the bridge crane operation path, its operation time and stacking plan.
[0116] In order to facilitate decoding and algorithm writing, for the SC / DC combined warehousing operation mode of the bridge crane, combined with the mathematical model and stacking strategy established in this embodiment, the bridge crane operation coding adopts an integer coding method based on the task sequence, and its sequence number is mapped to the coordinates of the target stack position of the steel plate corresponding to the task in the warehouse. Let the integer vector W = {w1, w2, w3, ..., w k ,…w r1+r2} is a bridge crane operation plan, W is the task sequence composed of I and I', w k is the mapping after the warehouse entry and exit tasks are rearranged, representing the sequence number of steel plate k in the current task sequence. Figure 8 As shown, popm represents a sequence of tasks after encoding, that is, an individual in the population.
[0117] In this embodiment, a double-layer 0-1 coded Boolean matrix is used to determine the in / out task and the stacking scheme, including two layers: the first layer is to encode the in / out task, and the second layer is to encode the stacking target position. In the first layer of in / out task determination, 1 represents storage and 0 represents out; according to the steel plate stacking strategy given above, when w * ∈I′, for the outgoing steel plate set I′, let the set of reversible stacking positions of the blocking plates on each outgoing plate be R j ; As mentioned above, the blocking plate has 8 optional stacking positions, R j = {r j1 ,r j2 ,r j3 ,…,r j8}; The elements in the collection correspond to Figure 5 There are 8 optional stacking positions in the stacking table, so the 0-1 coding mode can also be implemented for the second layer of stacking, 1 represents that the stacking position is selected as the stacking target position, otherwise the stacking position is not selected and 0. The Boolean matrix of the above coding method is shown in Table 1.
[0118] Table 1. Boolean matrix of stacking operation coding
[0119]
[0120] At the beginning of the algorithm, M task sequences are randomly generated as the initial parent population according to the initial stacking situation and the information of steel plates entering / exiting the warehouse. At this time, the Boolean matrix encoding of the M task sequences is also determined.
[0121] For the outgoing steel plates in the operation sequence corresponding to the selected individuals, a set of stackable positions is selected, and the stacking time is calculated by combining the stacking strategy given above (Formulas (2)-(7)), and the stacking target position is selected to obtain the stacking plan; the stacking plan for each outgoing steel plate includes the stacking target position selected from the set of stackable positions and its stacking time.
[0122] Then, based on the given stack coordinates, initial stacking conditions, in / out information of steel plates, and the operation sequence corresponding to each individual, the bridge crane operation path and operation time of each individual are calculated in combination with the formulas (9)-(13) given above.
[0123] In a genetic algorithm, the fitness of an individual can measure the degree of survival advantage of the individual in the population, and the higher the fitness, the better the quality of the solution. In this embodiment, the fitness f(T) of each individual in the initial parent population is calculated according to the following formula.
[0124]
[0125] Among them, T max , T min are the maximum and minimum execution time in the current sequence set respectively; α is set to a value much smaller than T to avoid the denominator being zero. min The number of.
[0126] Then the M individuals in the initial parent population are arranged in descending order of fitness.
[0127] S32 selects the parent population through a segmented roulette strategy to determine several individuals as chromosomes.
[0128] In this embodiment, according to the set selection rate p s , using the segmented roulette strategy to select the initial parent population, and select M×p s Individual.
[0129] In the specific implementation, a segmented roulette strategy is designed to select the parent population. Fig. 9 As shown, the parent population is divided into k segments (the number of individuals in each segment is ), randomly select one of the segments and choose from them using roulette. The probability of selecting the first segment is p l for:
[0130]
[0131] Among them, f ξ is the fitness value of the ξth individual in the lth segment.
[0132] For the lth segment, a random function rand(·) is used to generate a random number for each individual in the segment, and the random number is between (0,1); then the corresponding individual is selected according to the cumulative probability of the selection probability; l = 1, 2, ..., k; if the cumulative probability is greater than the random number at the corresponding position, the individual is selected, otherwise it is not selected.
[0133] Then, choose another segment and repeat the above operation until the number of individuals selected is M×p s .
[0134] Then, according to the following steps S33-S34, crossover and mutation operations are performed on the chromosome selected in step S32 to obtain a progeny population.
[0135] S33 calculates the crossover probability based on the Sigmoid function, and then randomly groups the selected chromosomes into pairs based on the crossover probability, and then performs a crossover operation on the two grouped chromosomes and repairs them.
[0136] In this embodiment, the crossover probability p c Calculated according to the following formula:
[0137]
[0138] Among them, p c,max and p c,min is the preset maximum and minimum crossover probability; f i is the fitness value of the selected individual; f max 、f min 、f avg are the maximum, minimum and average fitness values of the population respectively. According to the Sigmoid function The function value approaches 1 when x>9.903 and approaches 0 when x<-9.903, so we set A=9.903.
[0139] The crossover operation is the core step in the iteration of the genetic algorithm that can generate a new generation of populations. Its principle is similar to the process of chromosome crossing and exchanging gene fragments in genetics. The digital string of the solution after encoding plays the role of the chromosome. A random segment of the two substrings is intercepted and exchanged and recombined to generate two sets of new solutions for its offspring. Since the bridge crane operation plan is encoded as a string of digital sequences, only performing traditional operations will result in invalid solutions with missing or repeated sequences, and the sequence needs to be repaired. In this embodiment, before performing the crossover operation, the target intersection point of the chromosome and one or more repeated genes (or points) in the other sequence are determined, and set as the index target point. After the crossover is completed, the index target point is also exchanged.
[0140] In this embodiment, a random segment matching crossover strategy is adopted. First, two chromosomes are randomly selected, and a random number between (0, 1) is generated using the random function rand(·). If the crossover probability is greater than the random number, a crossover operation is performed on the two chromosomes; a crossover point and one or more genes (or points) at which the target crossover point is repeated in the other sequence are randomly determined on the two chromosomes, and are set as index target points. The equal-length genes of the two chromosomes at the target crossover point are exchanged. The process is as follows: Fig.10 (a); then Fig.10 As shown in (b), the index target points of the two chromosomes are exchanged to complete the crossover and repair operations on the chromosomes.
[0141] S34 calculates the mutation probability based on the Sigmoid function method, and then randomly selects the gene position on the chromosome after the crossover operation according to the mutation probability to perform a mutation operation to obtain the offspring population.
[0142] In this embodiment, the mutation probability p m Calculated according to the following formula:
[0143]
[0144] Among them, p m,max and p m,min is the pre-set maximum and minimum mutation probability; f i is the fitness value of the selected individual; f max 、f min 、f avg are the maximum, minimum and average fitness values of the population respectively. According to the Sigmoid function The function value approaches 1 when x>9.903, and approaches 0 when x<-9.903, so we set A′=9.903.
[0145] The mutation operation adopts the inversion method, such as Fig.11 As shown, first randomly select a chromosome after the crossover operation, and use the random function rand(·) to generate a random number between (0,1). If the mutation probability p m If it is greater than the random number, the chromosome is mutated; two points in the chromosome are randomly selected to exchange their positions to form a new bridge crane operation sequence plan.
[0146] Through the above steps S33-S34, a progeny population can be generated according to the chromosome selected in step S32.
[0147] S35 calculates the fitness of each individual in the offspring population and compares it with the fitness of the corresponding individual in the parent population. Based on the probability of accepting the solution, the individuals in the offspring population replace the corresponding individuals in the parent population.
[0148] Since the information of the steel plates leaving the warehouse is certain, the stackable set selected based on the information of the steel plates leaving the warehouse is set unchanged, so the stacking time determined previously remains unchanged. Here, the stackable set can also be re-determined according to the operation sequence of each individual in the offspring population, and then the stacking time can be re-determined.
[0149] Based on the given stack coordinates, initial stacking conditions, in / out information of steel plates, and the operation sequence corresponding to each individual, the bridge crane operation path and operation time of each individual in the offspring population are calculated in combination with the formulas (9)-(13) given above.
[0150] According to the fitness calculation formula (14) given above, calculate the fitness of each individual in the offspring population.
[0151] The genetic algorithm has a strong global search capability in the early stage of iteration, but in the later stage of iteration, as the convergence continues to increase, it is easy to fall into the local optimum. In order to make it have a stronger ability to jump out of the local optimum, this embodiment designs the solution acceptance probability calculation formula as follows:
[0152]
[0153] Among them, f i represents the fitness of the ith individual in the parent population; f′ i is the fitness of the corresponding individual in the offspring population; T is the current temperature. Since each individual in the offspring population is obtained by crossover and mutation operations of individuals in the parent population, its position does not change.
[0154] The random function rand(·) is used to generate a random number between (0,1). If P>the random number, the individuals in the offspring population are used to replace the corresponding individuals in the parent population to update the parent population.
[0155] S36 determines whether the maximum number of iterations has been reached. If so, the optimal individual is output; otherwise, the probability of accepting the solution is cooled down, and the process returns to step S32 to perform the next iteration until the maximum number of iterations is reached.
[0156] Determine whether the current number of iterations has reached the maximum number of iterations. If so, take the individual with the largest fitness in the updated parent population as the optimal individual and output it, including the bridge crane operation path, operation time and stacking plan. If the maximum number of iterations has not been reached, cool down by T←aT as the temperature of the next iteration, a is the set cooling coefficient, and then return to step S32 to loop until the output condition is met.
[0157] When the annealing temperature reaches the maximum number of iterations, the temperature is kept constant.
[0158] Application Examples
[0159] In this application example, taking a single reservoir area as an example, the effect of the bridge crane operation path optimization method based on the improved genetic simulated annealing algorithm (referred to as ISAGA algorithm) provided in the above-mentioned embodiment 1 is verified.
[0160] The stacking area in a single warehouse is set up with 6 rows and 5 columns of 30 stacks, which are classified by item number and size. According to the principle of stack allocation, the order of steel plate size can be ignored when stacking in the same stack. Fig.12As shown, the two ends of the stacking area are the incoming truck parking space and the RGV parking space. The truck parking space is the feeding space, and the positions on both sides are set as the spreader replacement area and are not used for stacking. The upper left corner of this point is set as the coordinate origin (0,0), the feed position coordinates are (0,3.5), and the RGV parking space on the right is the discharge position with coordinates (6,3.5).
[0161] Table 2 Number of incoming and outgoing steel plates and their target stack coordinates
[0162]
[0163] According to the actual working conditions, a set of inbound and outbound task data is generated as shown in Table 2, including 14 incoming steel plates with completed stacking position allocation decisions and 16 outbound steel plates with confirmed positions. Because the material warehouse is involved in production, production outbound operations are the first priority among all operation types, but since there is only one bridge crane in a single warehouse area, inbound and outbound operations are allowed to be carried out in coordination.
[0164] The target stack position information of this batch of tasks is as follows: Fig.13 As shown, the points indicated by red numbers represent warehousing tasks, and the points indicated by blue numbers represent outbound tasks. There are situations where there are multiple outbound steel plates in the same stack (overlapping blue numbers) or a certain stack is both an inbound and outbound target stack (overlapping red numbers), which increases the difficulty of scheduling and makes the problem more in line with the actual situation.
[0165] For the above-mentioned inbound / outbound tasks, the ISAGA algorithm provided in Example 1 is used to optimize and solve the operation path of the bridge crane, and the parameter settings of the algorithm are shown in Table 3. In order to verify the applicability of the algorithm, a comparative test is performed using the GA algorithm at the same time, and the common parameters of the two are kept consistent.
[0166] Table 3 ISAGA algorithm parameter settings
[0167]
[0168] The ISAGA algorithm is used to solve the bridge crane operation path optimization model to optimize the above-mentioned inbound / outbound tasks and output the operation paths before and after optimization. Fig.14 shown.
[0169] exist Fig.14In the figure, the “*” on the left indicates the feed position of the warehouse area. The incoming steel plates are transported from here to the target stack position by the bridge crane, and the outgoing steel plates are all at the outgoing position “*” on the right. The dotted line represents the empty travel of the bridge crane. As can be seen from Figure (a), the crane operation path before optimization is relatively messy, and there are obviously some "backtracking" situations, that is, the bridge crane chooses to perform the storage task far from the feed position and then turns back to perform the task far from the outgoing position, and then returns to the feed position empty from the outgoing position. This kind of operation route is mixed with a large number of empty travels. Arranging too many of these routes is the main reason for the increase in the time consumption of the overall operation. In Figure (b), the optimized operation plan has greatly improved this situation. It can be seen that the operation path is relatively smoother, and the small number of "backtracking" situations that are difficult to avoid are also shorter.
[0170] The advantage of ISAGA is that the optimization rate of no-load operation time is greater, reaching nearly a quarter, so the overall solution is better. The specific implementation plan after optimization is shown in Table 4. The numbers with "#" in the bridge crane operation sequence are storage tasks, and those with "*" are outbound tasks.
[0171] Table 4 Comparison of optimized operation paths
[0172]
[0173]
[0174] From the comparison of specific solutions, it can be seen that the solution of ISAGA can be simply summarized as the round-trip execution of SC tasks, DC tasks and SC tasks for storage, while the solution obtained by GA is relatively messy. This also indirectly confirms that in the design of the operation path solution, blindly increasing the number of DC tasks does not necessarily lead to better operation efficiency, because after each set of DC tasks is executed, if the next instruction is to leave the warehouse, the bridge crane will cross the entire warehouse area and return to the feeding position, which greatly increases the idle time. Therefore, reducing the idle travel is the key to the path optimization problem.
[0175] In order to avoid special cases in a single experiment, the above experiment was repeated 20 times. The basic data and basic parameters of the algorithm used were the same. The mean, variance and average optimization rate of the results of the 20 experiments were recorded and statistically shown in Table 5.
[0176] Table 5 Statistics of algorithm stability test results
[0177]
[0178] From the data in the above table, we can see that the solution quality of the ISAGA algorithm is better in the 20 solutions. The mean of its global operation time and the mean of its idle time are not much different from the single test results in the previous section. The optimization rate is obviously superior to that of the GA algorithm. Its lower variance value also proves that the performance of this algorithm is more stable than that of GA. Fig.15 By comparing the iteration curves of each algorithm, it can be seen that ISAGA does not have an advantage in the early iteration effect due to its more complex adaptive operator and local search mechanism. In the later stage of iteration, ISAGA's feedback mechanism adaptive adjustment capability, combined with the acceptance criterion of simulated annealing, enables GA to continue searching deeper into the solution space when it has converged after less than 50 generations, and continuously finds better solutions. It did not converge to a better bridge crane scheduling operation plan until 250 generations later, and the optimization rate was greatly improved.
[0179] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0180] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0181] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0182] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.
[0183] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.
Claims
1. A bridge crane operation path optimization method based on an improved genetic simulated annealing algorithm, characterized in that: The following steps are involved: S1 builds a steel plate stacking strategy; S2 builds a bridge crane operation path optimization model; divides the bridge crane task into three parts: SC task, DC task and stacking task, and obtains the objective functions of the three parts respectively; S3 solves the bridge crane operation path optimization model based on the improved genetic annealing algorithm to obtain the optimal operation path of the bridge crane; this step includes the following sub-steps: S31 determines the initial parent population popm and the fitness of each individual in the initial parent population popm, and arranges them in descending order of fitness; each individual includes the bridge crane operation path, its operation time and the stacking plan; S32 selects the parent population through a segmented roulette strategy to determine a number of individuals as chromosomes; S33 calculates the crossover probability according to the Sigmoid function, and then randomly groups the selected chromosomes in pairs according to the crossover probability, and then performs a crossover operation on the two grouped chromosomes and repairs them; S34 calculates the mutation probability based on the Sigmoid function, and then randomly selects the gene position on the chromosome for mutation operation based on the mutation probability to obtain the offspring population; S35 calculates the fitness of each individual in the offspring population and compares it with the fitness of the corresponding individual in the parent population. According to the probability of accepting the solution, the individuals in the offspring population replace the corresponding individuals in the parent population. S36 determines whether the maximum number of iterations has been reached. If so, the optimal individual is output; otherwise, the probability of accepting the solution is cooled down, and the process returns to step S32 to perform the next iteration until the maximum number of iterations is reached.
2. The bridge crane operation path optimization method based on the improved genetic simulated annealing algorithm according to claim 1 is characterized in that: In step S1, the steel plate stacking strategy is constructed as follows: Among them, l j’max , l j’min and They represent the maximum, minimum and average lengths of the steel plates in position j' respectively; l jr represents the length of the rth blocking plate in stack position j; Y i′jr is the stacking decision variable, representing the upper r-th layer of blocking plate in the stack position j where the outgoing steel plate i' is located. i’jr =1, otherwise Y i’jr =0; t r is the unit blocking plate thickness; h j’ is the target stack height for stacking; (x j ,y j ) and (x j′ ,y j′ ) respectively represent the coordinates of the stack position j where the steel plate to be shipped is located and its stack position j' where it can be turned over; The stacking time is calculated according to the following formula: When R j >0: Among them, t(b j ,b j′ ) is the time taken by the bridge crane to move from stack position j to the target stack position j'; R j is the set of blocking plates on stack position j; t r is the unit blocking plate thickness; T up,j and T up,j′ are the operating time of the bridge crane for lifting and loading and unloading operations on stack positions j and j' respectively; H is the maximum height of the stack position; h j and h j’ V is the target stack height for outbound delivery and the target stack height for reverse delivery; up It is the vertical running speed of the bridge crane.
3. The bridge crane operation path optimization method based on improved genetic simulated annealing algorithm according to claim 2 is characterized in that: In step S2, the bridge crane operation path optimization model constructed is as follows: I max =max(I,I′),I min =min(I,I′),b k ∈{b I ,b O }; Among them, Z m / Z n is the no-load decision variable, which means that after the m / nth SC / DC task is completed, if the next step is to execute the outbound task, the value is 1, otherwise it is 0; T SC,m is the time required for the crane to perform the mth SC task; T DC,n The time required for the crane to perform the nth DC task; T Sj The time required to complete a stacking operation; Y i′jr is the stacking decision variable; R j is the set of blocking plates on stack position j; b m is the target stack position of the SC task; b 2n-1 and b 2n are the target stack positions for the DC task; I and I′ are the sets of steel plates for inbound and outbound respectively; t(b1, b2) is the time required for the crane to move from stack position 1 to stack position 2; t(b i ,b k ) is the crane running from stack position i to the entry / exit platform b k time; b I 、b O They are the parking spaces for incoming trucks and the unloading spaces for the steel plate storage area; Then, the mathematical model for optimizing the operation path of the steel plate material warehouse bridge crane is as follows:
4. The bridge crane operation path optimization method based on the improved genetic simulated annealing algorithm according to claim 3 is characterized in that: In the above step S31, initially, M individuals are randomly selected to form an initial population, and the operation sequence corresponding to each individual is encoded using a Boolean matrix; at the same time, according to the steel plate stacking strategy and the bridge crane operation path optimization model, the bridge crane operation path, operation time and stacking plan for each individual are determined.
5. The bridge crane operation path optimization method based on improved genetic simulated annealing algorithm according to claim 1 is characterized in that: In step S32, according to the set selection rate p s , using the segmented roulette strategy to select the initial parent population, and select M×p s Individual.
6. The bridge crane operation path optimization method based on improved genetic simulated annealing algorithm according to claim 5 is characterized in that: The parent population is divided into k segments, and the number of individuals in each segment is Randomly select one of the segments and choose from it using roulette; the probability of selecting the first segment is p l for: Among them, f ξ is the fitness value of the ξth individual in the lth segment; For the lth segment, use the random function rand(·) to generate a random number for each individual in the segment, and the random number is between (0,1); then select the corresponding individual according to the cumulative probability of selection; l = 1, 2, ..., k; Then, choose another segment and repeat the above operation until the number of individuals selected is M×p s .
7. The bridge crane operation path optimization method based on improved genetic simulated annealing algorithm according to claim 1 is characterized in that: In step S33, the crossover probability p c Calculated according to the following formula: Among them, p c,max and p c,min is the preset maximum and minimum crossover probability; f i is the fitness value of the selected individual; f max 、f min 、f avg are the maximum, minimum and average fitness values of the population respectively; Before performing the crossover operation, determine the target crossover point of the chromosome and one or more points that are repeated in the other sequence, and set them as index target points. After the crossover is completed, the index target points are also exchanged.
8. The bridge crane operation path optimization method based on improved genetic simulated annealing algorithm according to claim 1 is characterized in that: In step S34, the mutation probability p m Calculated according to the following formula: Among them, p m,max and p m,min is the pre-set maximum and minimum mutation probability; f i is the fitness value of the selected individual; f max 、f min 、f avg are the maximum, minimum and average fitness values of the population, respectively.
9. The bridge crane operation path optimization method based on improved genetic simulated annealing algorithm according to claim 1 is characterized in that: In step S35, the probability of accepting a solution is calculated as: Among them, f i represents the fitness of the ith individual in the parent population; f′ i is the fitness of the corresponding individual in the offspring population; T is the current temperature.
10. The bridge crane operation path optimization method based on improved genetic simulated annealing algorithm according to claim 9 is characterized in that: In step S36, determine whether the current number of iterations has reached the maximum number of iterations. If so, the individual with the largest fitness in the updated parent population is taken as the optimal individual and output, including the bridge crane operation path, operation time and stacking plan; if the maximum number of iterations has not been reached, cool down by T←aT as the temperature of the next iteration, a is the set cooling coefficient, and then return to step S32 to loop the operation until the output conditions are met.
