A scheduling optimization method for AGV in docks

Abstract

The present invention discloses a scheduling optimization method applied to terminal AGV, and the implementation process is as follows: a ship transporting containers arrives at the terminal, and a single cantilever field bridge and an automatic guided vehicle (AGV) work together to transfer the container from the transport ship to the container storage area to complete the scheduling work. The scheduling optimization process of the present invention adopts a two-stage model. In the first stage, an optimization scheduling model of a single cantilever field bridge and AGV is established with the shortest time as the optimization target, and an improved simulated annealing algorithm (IGSAA) is proposed to solve the first stage model, so that low fitness solutions are accepted during the calculation process of the first stage, and the local optimal solution will not be trapped, thereby improving the search range; in the second stage, the AGV path planning model considers the conflict situation when multiple AGVs are running simultaneously, establishes a multi-AGV conflict-free path planning mathematical model with improved shortest running time, proposes an improved conflict-free path planning strategy, and realizes conflict-free path planning for multiple AGVs at the terminal. The present invention improves the working efficiency of the single cantilever field bridge and AGV at the terminal.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned vehicles, and in particular to a scheduling optimization method applied to AGVs at docks. Background Art

[0002] The equipment of automated container terminals is mainly divided into three types according to their operation mode: horizontal transport equipment AGV, vertical handling equipment quay crane and yard crane. Due to the limitations of equipment configuration, loading and unloading technology and working space of automated container terminals, optimizing the scheduling of each link has become the key to improving the efficiency of automated container terminals. As the three important basic equipment resources of automated container terminals, the scheduling and allocation of quay cranes, AGVs and yard cranes are crucial to improving the operating efficiency of automated container terminals.

[0003] In recent years, automated container terminals have been built and put into operation one after another. Most of these automated terminal yards adopt a vertical shore layout. In view of the high proportion of water-to-water transshipment, the Phase IV Terminal of Shanghai Yangshan Port adopts a mixed layout mode of spacing between non-cantilevered container areas and single-cantilevered container areas.

[0004] The AGV parking space selection and the driving path planning of the lanes in the single cantilever box area directly determine the time when the AGV enters the single cantilever box area and interacts with the field crane, which is related to the moving path and waiting time of the field crane, thus affecting the operating efficiency of the single cantilever field crane. Scientifically and effectively optimizing the operating sequence of the single cantilever field crane and the AGV driving path is an urgent problem to be solved in the current automated terminal.

[0005] The present invention takes the terminal container as the research object, proposes an improvement on the scheduling method between the terminal AGV and the terminal yard crane, and improves the collaborative working efficiency of the terminal single-cantilever yard crane and AGV. Summary of the invention

[0006] Purpose of the invention: In view of the optimization of the coordinated work between the single cantilever gantry crane and AGV at the terminal as pointed out in the background technology, the present invention proposes a scheduling optimization method applied to the AGV at the terminal to improve the working efficiency of the single cantilever gantry crane and AGV at the terminal.

[0007] Technical solution: The present invention provides a scheduling optimization method for terminal AGV, which is realized based on the coordinated operation of single cantilever field bridge and automatic guided vehicle AGV. The single cantilever field bridge is arranged in pairs, and the lanes in the field bridge include the working lane under the cantilever and the two transport lanes in the middle; the flow directions of the four lanes are two-way, one-way to the sea side, one-way to the land side, and two-way from top to bottom; the land side is arranged with a container truck operation interaction area as needed; the interactive operation of the single cantilever field bridge and the AGV occurs on the cantilever side of the field bridge, and the task of each AGV is determined by the starting point and end point of the container; the specific implementation includes the following steps:

[0008] Step 1: A two-stage algorithm optimization model is constructed by the way that the single cantilever field bridge and the automatic guided vehicle AGV work together; the first stage of the two-stage algorithm optimization model establishes an optimization scheduling model of the single cantilever field bridge and the AGV with the shortest time as the optimization target, and the second stage is an AGV path planning model;

[0009] Step 2: An improved simulated annealing algorithm IGSAA is proposed to solve the one-stage model in step 1, so that low fitness solutions are accepted during the one-stage calculation process, and local optimal solutions will not be trapped, thereby increasing the search range;

[0010] Step 3: For the multi-AGV conflict-free path planning mathematical model with improved shortest running time established in the second stage of step 1, an improved conflict-free path planning strategy is proposed to realize the conflict-free path planning of multiple AGVs in the terminal.

[0011] Furthermore, the two-stage algorithm optimization model of the single cantilever field bridge and AGV working in cooperation with each other in step 1 is specifically as follows:

[0012] The shortest time required to complete the work of all containers in the yard crane:

[0013]

[0014] Where T is the time period set, The time and time required to complete all containers to be operated for the container area H yard bridge;

[0015] The decision variables of the position when the field bridge starts working are:

[0016]

[0017] Among them, K h represents the set of h-yard bridges in the container area, k h =1 is the sea side bridge, k h =2 is the land side bridge; k h The location of the container when the work starts, For Bei, For the column; It indicates the position information of the seaside / landside crane during operation. When the crane is working, s=1; when the crane is not working, s=0.

[0018] Continuous movement process of the field crane during operation:

[0019]

[0020]

[0021] in, The location of the field bridge at time t; Bh is the set of all bays in container area h; b is any bay information, b∈B h ;

[0022] The dispatch time calculation formula for each container is as follows:

[0023]

[0024] Among them, n is the container currently operated by the yard crane; Indicates the field bridge k at time t h The decision variable for starting work, s = 1 when the field bridge is working, otherwise s = 0; Indicates the field bridge k at time t h The decision variable for the work has been completed, f = 1 when the work is completed, otherwise f = 0; b i For parking space i∈V ph The corresponding bay number of container area h; t1 is the time for the crane to lift or put down the container in the container area and AGV; t2 is the time required for the crane to move one bay; t3 is the time required for the crane to move one unit column; y ni represents the decision variable for the parking point of the AGV transporting the container, When the AGV is parked in the parking space, i = 1, otherwise i = 0; M is a random positive number; Indicates field bridge k h Decision variables for completing the virtual container task; Indicates field bridge k h The virtual container task of the operation starts; is a collection of containers to be stored in container area h; is the set of containers to be extracted from container zone h, n∈N h ; V ph It is a collection of parking spaces in the container area;

[0025] Field Bridge h The time relationship between completing the current task container n and switching to the next task container n' is as follows:

[0026]

[0027]

[0028] Where n' is the container under the crane; b i ' is the next parking space i'∈V ph The corresponding parking space number of container area h; t is the current operation time point; t' is the next operation time point; i is the current parking space; i' is the next parking space;

[0029] When each yard crane on the terminal is initially working, the container time constraint of the operation is:

[0030]

[0031] The time constraint of the parking spaces where two containers n and n' are located when they are continuously cross-operated on the yard bridge is:

[0032]

[0033] Among them, t4 represents the time when two AGVs share a parking space, the current AGV drives out, and the next AGV drives into the parking space;

[0034] The total time required for the crane to complete the work of this batch of containers:

[0035]

[0036] in, It indicates the total time of all containers to be operated in the current area in container zone h;

[0037] In the above formula, The mathematical relationship between them is:

[0038]

[0039] In the above formula, The decision variable types are all {0, 1}.

[0040] Furthermore, the specific method of the improved simulated annealing algorithm IGSAA in step 2 is:

[0041] Step 1: Encoding, encode the solution obtained by the first-stage model, and design three rows of chromosomes. The length of the chromosome represents the number of containers to be operated in the optimized container area. The first row represents the container number, the second row represents the single cantilever yard bridge for operation, and the third row represents the parking space of the AGV;

[0042] Step 2: Generate the initial solution. The containers in each row are sorted according to their task sequence. The vacant single cantilever bridge is detected and selected, thereby generating an initial solution individual to form the initial population.

[0043] Step 3: Calculate the fitness value. Use the evaluation index to screen the individuals in the initial population and calculate the fitness value of the selected individuals. The formula is as follows:

[0044]

[0045] Among them, F(S i) is the individual fitness value; σ, τ, υ are evaluation indexes; C1, C2, C3 are three individuals randomly selected from the population;

[0046] Step 4: Crossover mutation, using a smaller crossover rate and a larger mutation rate to prevent falling into the local optimum and maintain the current higher fitness population; crossover rate P c and mutation rate P m The calculation formula is as follows:

[0047]

[0048] Where t is the current iteration number; T is the maximum iteration number; c is a random number c∈[0,1], which makes the evolution random; P cmax is the maximum crossover probability, P cmin is the minimum crossover probability, P mmax is the maximum mutation probability, P mmin is the minimum mutation probability, k and y are the crossover rates P respectively. c and mutation rate P m The interpolation formula of k, y is as follows:

[0049]

[0050] Step 5: Simulate annealing neighborhood operation, randomly select two positions for the first two rows of genes respectively and independently, exchange the gene values ​​of the two positions, and randomly select the gene value of the third row to generate new genes;

[0051] Step 6: Adapt the Metropolis criterion, reselect the last solution, compare the current target value with the previous target value, and select the optimal solution using the probability calculation method of exp(-dC / T0), where dC is the difference of the objective function and T0 is the mean of the solution.

[0052] Furthermore, the mathematical model of multi-AGV conflict-free path planning with improved shortest running time in step 3 is specifically implemented as follows:

[0053] The shortest time for all AGVs to complete container transportation tasks:

[0054]

[0055] T A0ij When the jth AGV performs the i-th task, the no-load running time is:

[0056]

[0057] Where u represents the uth road section that the AGV travels during the working process, u∈[1,G ij]; U1 represents the road section set consisting of the container storage points except the temporary container stop where the yard bridge is located; L is the length of the AGV vehicle driving section; R abij is a decision variable, which determines whether the jth AGV passes through the road section h when performing the i-th task ab , R abij =1 means passing through section h ab , otherwise, it does not pass; q Z0ij The total number of empty turns for the jth AGV to perform the i-th task; q K0ij For F ij After the dynamic path planning, the total number of obstacle avoidances when the j-th AGV performs the i-th task without load; F ij represents the total number of dynamic path planning times when the jth AGV performs the ith task; s T is the total distance that the AGV travels during the turning, obstacle avoidance, and loading process; v is the speed at which the AGV travels;

[0058] T A1ij When the jth AGV performs the i-th task, the loading operation time is:

[0059]

[0060] Among them, q Z1ij The total number of turns for the jth AGV to perform the i-th task; q K1ij For F ij After the dynamic path planning, the jth AGV performs the total number of obstacle avoidances for the ith task; U2 represents the road section set consisting of the container storage points except the temporary container stop and container storage point where the yard bridge is located;

[0061] T W0ij When the jth AGV performs the i-th task, the idle waiting time; T W1ij When the jth AGV performs the i-th task, the loading waiting time is:

[0062]

[0063] Among them, t wkij represents the kth obstacle avoidance time when the jth AGV performs the i-th task; t p1 The time it takes for the AGV to accelerate and decelerate;

[0064] T J1ij The time it takes for the jth AGV to load a container when performing the i-th task; T Z0ij When the j-th AGV performs the i-th task, the no-load turning time is:

[0065] T Z0ij =q Z0ij ·(tp +t p1 )

[0066] T Z1ij =q Z1ij ·(t p +t p1 )

[0067] Among them, t p It takes time for the AGV to turn.

[0068] T Z1ij When the jth AGV performs the i-th task, the cargo turning time is:

[0069] T Z1ij =t qij

[0070] Among them, t qij It represents the waiting time for the jth AGV to wait for the loading container at the yard bridge when performing the i-th task;

[0071] T Q1ij When the jth AGV performs the i-th task, the waiting time in the queue is:

[0072] T J1ij =t 1d +t p1 +t 1i

[0073] Among them, t 1d The time required to load and place the container for the AGV; t p1 The time taken for AGV to accelerate and decelerate; t 1i The time it takes for a single cantilever crane to retrieve a container from the AGV.

[0074] Furthermore, the conflict-free path planning strategy is improved in step 3, and is specifically implemented as follows:

[0075] The improved dynamic obstacle avoidance strategy is to determine the AGV waiting for obstacle avoidance and its stop waiting time and spatial position, and realize conflict-free path planning through global obstacle avoidance. Considering that AGV has two states: loaded and empty, when two AGVs conflict, the load conditions of the two AGVs are judged, and the loaded AGV is given priority; if both AGVs are loaded, the AGV with the shortest working time is considered, and the AGV that takes the shortest time to complete the work has priority.

[0076] Beneficial effects:

[0077] 1. The present invention uses the shortest time as the optimization target to establish a model, and the AGV dispatches and cooperates with the single cantilever field bridge to perform operations. The AGV path and the single cantilever field bridge path are coordinated and optimized, which can achieve the optimal dispatch of terminal containers at the fastest speed;

[0078] 2. The present invention adopts an improved dynamic obstacle avoidance strategy to avoid conflicts, and the replanned path is a global dynamic conflict-free path. According to the global dynamic conflict-free path, the AGV completes the conflict prediction of transportation when carrying objects.

[0079] 3. The present invention proposes an improved simulated annealing algorithm (IGSAA) to solve the one-stage model, so that low fitness solutions are accepted during the one-stage calculation process, and local optimal solutions will not be trapped, thereby increasing the search range. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 A system flow chart of a scheduling optimization method for terminal AGV provided by an embodiment of the present invention;

[0081] Figure 2 The working process diagram of the single cantilever yard bridge applied to the wharf;

[0082] Figure 3 An improved simulated annealing algorithm (IGSAA) flow chart is proposed for the present invention. DETAILED DESCRIPTION

[0083] The present invention and the retrograde are further described below in conjunction with the drawings in the examples of the present invention. The following examples provide a detailed and complete description of the technical solution of the present invention, but do not limit the present invention in any form. On the basis of the present invention, a number of optimizations and improvements made by ordinary technicians in this field without making other creative ideas. The above all belong to the scope of protection of the present invention.

[0084] The present invention proposes a scheduling optimization method for AGVs at the wharf. The method adopts a cooperative operation mode between a single cantilever field bridge and an automatic guided vehicle (AGV) to construct a two-stage algorithm optimization model to improve the working efficiency of the single cantilever field bridge and AGV at the wharf. In the first stage, an optimization scheduling model of a single cantilever field bridge and AGV is established with the shortest time as the optimization target. An improved simulated annealing algorithm (IGSAA) is proposed to solve the first-stage model, so that low fitness solutions are accepted during the calculation process of the first stage, and the local optimal solution will not be trapped, thereby improving the search range. In the second stage, the AGV path planning model considers the conflict situation when multiple AGVs are running at the same time, establishes a multi-AGV conflict-free path planning mathematical model with improved shortest running time, proposes an improved conflict-free path planning strategy, and realizes conflict-free path planning for multiple AGVs at the wharf.

[0085] After the ship docks, there are two main work links: the loading and unloading of containers by the single cantilever of the terminal bridge and the horizontal transportation of containers by the AGV. In the embodiment of the present invention, the single cantilever bridges are arranged opposite to each other in pairs. The lanes in the bridge include the working lane under the cantilever and the two transport lanes in the middle; the flow directions of the four lanes are two-way, one-way to the sea side, one-way to the land side, and two-way from top to bottom; the land side is arranged with a container truck operation interaction area as needed; the interaction between the single cantilever bridge and the AGV occurs on the cantilever side of the bridge, and the task of each AGV is determined by the starting and end points of the container. The specific plan is as follows:

[0086] S1: During dock work, AGV is dispatched to cooperate with the single cantilever field bridge to perform operations. The AGV path and the single cantilever field bridge path are collaboratively optimized. The present invention uses the shortest time as the optimization target to establish a model.

[0087] The mathematical model is established based on the working relationship between the single cantilever field bridge and AGV as follows:

[0088] The shortest time required to complete the work of all containers in the yard crane:

[0089]

[0090] Where T is the time period set, Time and time to complete all containers to be operated for the container area h yard bridge.

[0091] The decision variables of the position when the field bridge starts working are:

[0092]

[0093] Among them, K h represents the set of h-yard bridges in the container area, k h =1 is the sea side bridge, k h =2 is the land side bridge; k h The location of the container when the work starts, For Bei, For the columns, It indicates the position information of the seaside / landside crane during operation. When the crane is working, s=1; when the crane is not working, s=0.

[0094] Continuous movement process of the field crane during operation:

[0095]

[0096] in, The location of the field bridge at time t; B h is the set of all bays in container area h; b is any bay information, b∈B h .

[0097] The dispatch time calculation formula for each container is as follows:

[0098]

[0099] Where n is the container currently operated by the crane, n∈N h ; Indicates the field bridge k at time t h The decision variable for starting work, s = 1 when the field bridge is working, otherwise s = 0; Indicates the field bridge k at time t h The decision variable for the work has been completed, f = 1 when the work is completed, otherwise f = 0; b i For parking space i∈V ph The corresponding bay number of container area h; t1 is the time for the crane to lift or put down the container in the container area and AGV; t2 is the time required for the crane to move one bay; t3 is the time required for the crane to move one unit column; y ni represents the decision variable for the parking point of the AGV transporting the container, When the AGV is parked in the parking space, i = 1, otherwise i = 0; M is a random positive number; Indicates field bridge k h Decision variables for completing the virtual container task; Indicates field bridge k h The virtual container task of the job starts, is a collection of containers to be stored in container area h; V is the container set to be extracted from container area h; ph It is a collection of parking spaces in the container area.

[0100] Field Bridge h The time relationship between completing the current task container n and switching to the next task container n' is as follows:

[0101]

[0102] Where n' is the container under the crane; b i ' is the next parking space i'∈V ph The corresponding parking space number of container area h; t is the current operation time point; t' is the next operation time point; i is the current parking space; i' is the next parking space.

[0103] When each yard crane on the terminal is initially working, the container time constraint of the operation is:

[0104]

[0105] The time constraint of the parking spaces where two containers n and n' are located when they are continuously cross-operated on the yard bridge is:

[0106]

[0107] Among them, t4 represents the time when two AGVs share a parking space and the current AGV drives out and the next AGV drives into the parking space.

[0108] The total time required for the crane to complete the work of this batch of containers:

[0109]

[0110] in, It indicates the total time of all containers to be operated in the current area in container zone h.

[0111] In the above formula, The mathematical relationship between them is:

[0112]

[0113] In the above formula, The decision variable types are all {0, 1}.

[0114] S2: The present invention proposes multi-AGV conflict-free path planning with the shortest time as the goal. The specific implementation steps are to initially plan the shortest AGV path and save it. When a conflict occurs between AGVs, an improved dynamic obstacle avoidance strategy is used to avoid the conflict, and the re-planned path is a global dynamic conflict-free path. According to the global dynamic conflict-free path, the AGV completes the conflict prediction of transportation when loading.

[0115] S6: The improved dynamic obstacle avoidance strategy proposed in the present invention is to determine the AGV waiting for obstacle avoidance and its time and spatial position for stopping and waiting, and realize conflict-free path planning through global obstacle avoidance. Considering that AGV has two states: loaded and unloaded, when two AGVs conflict, the load conditions of the two AGVs are judged, and the loaded AGV is given priority; if both AGVs are loaded, the AGV with the shortest working time is considered, and the AGV with the shortest time to complete the work has priority.

[0116] S7: Since the AGV needs to continuously receive container scheduling tasks, the AGV path needs to be planned in real time, and the location information of other AGVs needs to be continuously received. The improved dynamic obstacle avoidance strategy is used to update the obstacle avoidance path according to the time information to complete the current container scheduling task.

[0117] S8: In the second stage, the AGV path planning model considers the conflict situation when multiple AGVs are running simultaneously, and establishes a multi-AGV conflict-free path planning mathematical model with the shortest running time as follows:

[0118] The shortest time for all AGVs to complete container transportation tasks:

[0119]

[0120] The calculation formulas for each time of the total working time of AGV are as follows:

[0121] T A0ij When the jth AGV performs the i-th task, the no-load running time is:

[0122]

[0123] Where u represents the uth road section that the AGV travels during the working process, u∈[1,G ij ]; U1 represents the road section set consisting of the container storage points except the temporary container stop where the yard bridge is located; L is the length of the AGV vehicle driving section; R abij is a decision variable, which determines whether the jth AGV passes through the road section h when performing the i-th task ab , R abij =1 means passing through section h ab , otherwise, it does not pass; q Z0ij The total number of empty turns for the jth AGV to perform the i-th task; q K0ij For F ij After the dynamic path planning, the total number of obstacle avoidances when the j-th AGV performs the i-th task without load; F ij represents the total number of dynamic path planning times when the jth AGV performs the ith task; s T is the total distance that the AGV travels during the turning, obstacle avoidance, and loading process; v is the speed at which the AGV travels;

[0124] T A1ij When the jth AGV performs the i-th task, the loading operation time is:

[0125]

[0126] Among them, q Z1ij The total number of turns for the jth AGV to perform the i-th task; q K1ij For F ij After the dynamic path planning, the jth AGV performs the total number of obstacle avoidances for the ith task; U2 represents the road section set consisting of the container storage points except the temporary container stop and container storage point where the yard bridge is located;

[0127] T W0ij When the jth AGV performs the i-th task, the idle waiting time; T W1ij When the jth AGV performs the i-th task, the loading waiting time is:

[0128]

[0129] Among them, t wkij represents the kth obstacle avoidance time when the jth AGV performs the i-th task; t p1 The time it takes for the AGV to accelerate and decelerate;

[0130] T J1ij The time it takes for the jth AGV to load a container when performing the i-th task; T Z0ij When the j-th AGV performs the i-th task, the no-load turning time is:

[0131] T Z0ij =q Z0ij ·(t p +t p1 )

[0132] T Z1ij =q Z1ij ·(t p +t p1 )

[0133] Among them, t p It takes time for the AGV to turn.

[0134] T Z1ij When the jth AGV performs the i-th task, the cargo turning time is:

[0135] T Z1ij =t qij

[0136] Among them, t qij It represents the waiting time for the jth AGV to wait for the loading container at the yard bridge when performing the i-th task;

[0137] T Q1ij When the jth AGV performs the i-th task, the waiting time in the queue is:

[0138] T J1ij =t 1d +t p1 +t 1i

[0139] Among them, t 1d The time required to load and place the container for the AGV; t p1 The time taken for AGV to accelerate and decelerate; t 1i The time it takes for a single cantilever crane to transfer a container to the AGV;

[0140] S9: The yard bridge scheduling and AGV parking space selection of the single cantilever container area do not interfere with each other, so the single container working area is optimized and scheduled in the first stage. The present invention proposes an improved simulated annealing algorithm (IGSAA) to solve the one-stage model, so that low fitness solutions are accepted during the one-stage calculation process, and the local optimal solution will not be trapped, thereby improving the search range.

[0141] The specific implementation process of the IGSAA algorithm is as follows:

[0142] Step 1: Encoding. Encode the solution obtained by the first-stage model and design three rows of chromosomes. The length of the chromosome represents the number of containers to be operated in the optimized container area. The first row indicates the container number, the second row indicates the single cantilever yard bridge for operation, and the third row indicates the parking space for the AGV.

[0143] Step 2: Generate initial solution. The containers in each row are sorted according to their task sequence, and the empty single cantilever bridge is detected and selected, thereby generating an initial solution individual to form the initial population.

[0144] Step 3: Calculate the fitness value. Use the evaluation index to screen the individuals in the initial population and calculate the fitness value of the selected individuals. The formula is as follows:

[0145]

[0146] Among them, F(S i ) is the individual fitness value; σ, τ, υ are evaluation indexes; C1, C2, C3 are three individuals randomly selected from the population.

[0147] Step 4: Crossover mutation. Use a smaller crossover rate and a larger mutation rate to prevent falling into the local optimum and maintain the current high fitness population. Crossover rate P c and mutation rate P m The calculation formula is as follows:

[0148]

[0149] Where t is the current iteration number; T is the maximum iteration number; c is a random number c∈[0,1], which makes the evolution random; P cmax is the maximum crossover probability, P cmin is the minimum crossover probability, P mmax is the maximum mutation probability, P mmin is the minimum mutation probability, k and y are the crossover rates P respectively. c and mutation rate P m interpolation.

[0150] The calculation formula for k,y is as follows:

[0151]

[0152] Step 5: Simulated annealing neighborhood operation. Randomly select two positions for the first two rows of genes separately and independently, exchange the gene values ​​of the two positions, and randomly select the gene values ​​of the third row to generate new genes.

[0153] Step 6: Adaptive Metropolis criterion. Reselect the last solution, compare the current target value with the previous target value, and select the optimal solution using the probability calculation method of exp(-dC / T0).

[0154] Among them, dC is the difference of the objective function and T0 is the mean of the solution.

[0155] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent transformation or modification made according to the spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. A scheduling optimization method for terminal AGVs, which is based on the coordinated operation of single cantilever field cranes and automatic guided vehicles (AGVs). Single cantilever field cranes are arranged in pairs, and the lanes in the field cranes include the working lanes under the cantilever and the two transport lanes in the middle; the flow directions of the four lanes from top to bottom are bidirectional, one-way to the sea side, one-way to the land side, and bidirectional; the land side is arranged with a container truck operation interaction area as needed; the interactive operation of the single cantilever field crane and the AGV occurs on the cantilever side of the field crane, and the task of each AGV is determined by the starting point and end point of the container; it is characterized in that The specific implementation includes the following steps: Step 1: A two-stage algorithm optimization model is constructed by the way that the single cantilever field bridge and the automatic guided vehicle AGV work together; the first stage of the two-stage algorithm optimization model establishes an optimization scheduling model of the single cantilever field bridge and the AGV with the shortest time as the optimization target, and the second stage is an AGV path planning model; The two-stage algorithm optimization model for the coordinated operation of the single cantilever field bridge and the AGV in step 1 is as follows: The shortest time required to complete the work of all containers in the yard crane: Where T is the time period set, The time and time required to complete all containers to be operated for the container area H yard bridge; The decision variables of the position when the field bridge starts working are: Among them, K h represents the set of h-yard bridges in the container area, k h =1 is the sea side bridge, k h =2 is the land side bridge; k h The location of the container when the work starts, For Bei, For the column; Indicates the position information of the seaside / landside crane during operation. When the crane is working, s=1; when the crane is not working, s=0; Continuous movement process of the field crane during operation: in, The location of the field bridge at time t; B h is the set of all bays in container area h; b is any bay information, b∈B h ; The dispatch time calculation formula for each container is as follows: Where n is the container currently operated by the crane, n∈N h ; Indicates the field bridge k at time t h The decision variable for starting work, s = 1 when the field bridge is working, otherwise s = 0; Indicates the field bridge k at time t h The decision variable for the work has been completed, f = 1 when the work is completed, otherwise f = 0; b i For parking space i∈V ph The corresponding bay number of container area h; t1 is the time for the crane to lift or put down the container in the container area and AGV; t2 is the time required for the crane to move one bay; t3 is the time required for the crane to move one unit column; y ni represents the decision variable for the parking point of the AGV transporting the container, When the AGV is parked in the parking space, i = 1, otherwise i = 0; M is a random positive number; Indicates field bridge k h Decision variables for completing the virtual container task; Indicates field bridge k h The virtual container task of the operation starts; is a collection of containers to be stored in container area h; V is the container set to be extracted from container area h; ph It is a collection of parking spaces in the container area; Field Bridge h The time relationship between completing the current task container n and switching to the next task container n' is as follows: Where n' is the container under the crane; b i ' is the next parking space i'∈V ph The corresponding parking space number of container area h; t is the current operation time point; t' is the next operation time point; i is the current parking space; i' is the next parking space; When each yard crane on the terminal is initially working, the container time constraint of the operation is: The time constraint of the parking spaces where two containers n and n' are located when they are continuously cross-operated on the yard bridge is: Among them, t4 represents the time when two AGVs share a parking space, the current AGV drives out, and the next AGV drives into the parking space; The total time required for the crane to complete the work of this batch of containers: in, It indicates the total time of all containers to be operated in the current area in container zone h; In the above formula, The mathematical relationship between them is: In the above formula, y ni The decision variable types are all {0, 1}; Step 2: An improved simulated annealing algorithm IGSAA is proposed to solve the one-stage model in step 1, so that low fitness solutions are accepted during the one-stage calculation process, and local optimal solutions will not be trapped, thereby increasing the search range; The specific method of the improved simulated annealing algorithm IGSAA in step 2 is: Step 1: Encoding, encode the solution obtained by the first-stage model, and design three rows of chromosomes. The length of the chromosome represents the number of containers to be operated in the optimized container area. The first row represents the container number, the second row represents the single cantilever yard bridge for operation, and the third row represents the parking space of the AGV; Step 2: Generate the initial solution. The containers in each row are sorted according to their task sequence. The vacant single cantilever bridge is detected and selected, thereby generating an initial solution individual to form the initial population. Step 3: Calculate the fitness value. Use the evaluation index to screen the individuals in the initial population and calculate the fitness value of the selected individuals. The formula is as follows: Among them, F(S i ) is the individual fitness value; σ, τ, υ are evaluation indexes; C1, C2, C3 are three individuals randomly selected from the population; Step 4: Crossover mutation, using a smaller crossover rate and a larger mutation rate to prevent falling into the local optimum and maintain the current higher fitness population; crossover rate P c and mutation rate P m The calculation formula is as follows: Where t is the current iteration number; T is the maximum iteration number; c is a random number c∈[0,1], which makes the evolution random; P cmax is the maximum crossover probability, P cmin is the minimum crossover probability, P mmax is the maximum mutation probability, P mmin is the minimum mutation probability, k and y are the crossover rates P respectively. c and mutation rate P m The interpolation formula for k and y is as follows: Step 5: Simulate annealing neighborhood operation, randomly select two positions for the first two rows of genes respectively and independently, exchange the gene values ​​of the two positions, and randomly select the gene value of the third row to generate new genes; Step 6: Adaptive Metropolis criterion, reselect the last solution, compare the current target value with the previous target value, and select the optimal solution by the probability calculation method of exp(-dC / T0), where dC is the difference of the objective function and T0 is the mean of the solution; Step 3: For the multi-AGV conflict-free path planning mathematical model with improved shortest running time established in the second stage of step 1, an improved conflict-free path planning strategy is proposed to realize the conflict-free path planning of multiple AGVs at the terminal; Improve the mathematical model of multi-AGV conflict-free path planning with the shortest running time. The specific steps are as follows: The shortest time for all AGVs to complete container transportation tasks: T A0ij When the jth AGV performs the i-th task, the no-load running time is: Where u represents the uth road section that the AGV travels during the working process, u∈[1,G ij ]; U1 represents the road section set consisting of the container storage points except the temporary container stop where the yard bridge is located; L is the length of the AGV vehicle driving section; R abij is a decision variable, which determines whether the jth AGV passes through the road section h when performing the i-th task ab , R abij =1 means passing through section h ab , otherwise, it does not pass; q Z0ij The total number of empty turns for the jth AGV to perform the i-th task; q K0ij For F ij After the dynamic path planning, the total number of obstacle avoidances when the j-th AGV performs the i-th task without load; F ij represents the total number of dynamic path planning times when the jth AGV performs the ith task; s T is the total distance that the AGV travels during the turning, obstacle avoidance, and loading process; v is the speed at which the AGV travels; T A1ij When the jth AGV performs the i-th task, the loading operation time is: Among them, q Z1ij The total number of turns for the jth AGV to perform the i-th task; q K1ij For F ij After the dynamic path planning, the jth AGV performs the total number of obstacle avoidances for the ith task; U2 represents the road section set consisting of the container storage points except the temporary container stop and container storage point where the yard bridge is located; T W0ij When the jth AGV performs the i-th task, the idle waiting time; T W1ij When the jth AGV performs the i-th task, the loading waiting time is: Among them, t wkij represents the kth obstacle avoidance time when the jth AGV performs the i-th task; t p1 The time it takes for the AGV to accelerate and decelerate; T J1ij The time it takes for the jth AGV to load a container when performing the i-th task; T Z0ij When the j-th AGV performs the i-th task, the no-load turning time is: T Z0ij =q Z0ij ·(t p +t p1 ) T Z1ij =q Z1ij ·(t p +t p1 ) Among them, t p It takes time for the AGV to turn. T Z1ij When the jth AGV performs the i-th task, the cargo turning time is: T Z1ij =t qij Among them, t qij It represents the waiting time for the jth AGV to wait for the loading container at the yard bridge when performing the i-th task; T Q1ij When the jth AGV performs the i-th task, the waiting time in the queue is: T J1ij =t 1d +t p1 +t 1i Among them, t 1d The time required to load and place the container for the AGV; t p1 The time taken for AGV to accelerate and decelerate; t 1i The time it takes for a single cantilever crane to transfer a container to the AGV; Improve the conflict-free path planning strategy, which is implemented as follows: The improved dynamic obstacle avoidance strategy is to determine the AGV waiting for obstacle avoidance and its stop waiting time and spatial position, and realize conflict-free path planning through global obstacle avoidance. Considering that AGV has two states: loaded and empty, when two AGVs conflict, the load conditions of the two AGVs are judged, and the loaded AGV is given priority; if both AGVs are loaded, the AGV with the shortest working time is considered, and the AGV that takes the shortest time to complete the work has priority.

Images