Automatic positioning method for container stacking position based on Al

Abstract

The invention relates to the technical field of container stacking, in particular to an Al-based container stacking position automatic positioning method. Comprising the following steps: setting a plurality of monitoring sub-regions based on harbor storage yard parameters, and constructing a storage yard electronic map; an intelligent agent is constructed according to the historical stacking parameters, and the intelligent agent sets a first-level stacking plan according to a processing result of the to-be-stacked container parameters; according to a preset evaluation time node, generating an operation efficiency value of each monitoring sub-region, and according to all the operation efficiency values, judging whether an updating instruction of the intelligent agent is generated or not; the Al intelligent agent processes the characteristic parameters of the to-be-stacked containers, selects a corresponding location searching sub-strategy, rapidly selects a proper stacking point through the location searching sub-strategy and an electronic monitoring map, and intelligently distributes the stacking position through overall planning of all the to-be-stacked containers, thereby reducing the moving distance in a storage yard, avoiding repeated scheduling, and improving the stacking efficiency of the to-be-stacked containers. The stacking efficiency of containers at ports and piers is improved, and the space use of a storage yard is optimized.

Description

Technical Field

[0001] The present application relates to the technical field of container stacking, and in particular to an automatic positioning method for container stacking positions based on AI. Background Art

[0002] With the rapid development of global trade, container transportation has become one of the main modes of international trade. As important nodes in the logistics chain, ports and container yards undertake the tasks of loading, unloading, storage and transshipment of a large number of containers.

[0003] However, the traditional container stacking location management method mainly relies on manual operation and experience judgment, which has the following problems: Low efficiency: Manual allocation and search for container stacking locations takes a long time, especially in large-scale yards, it is difficult to quickly locate the target container. Yard congestion and unreasonable stacking location allocation lead to low equipment operation efficiency. Summary of the invention

[0004] The purpose of this application is: to solve the above technical problems, this application provides an automatic positioning method for container stacking positions based on AI, aiming to improve the stacking efficiency of containers at port terminals and optimize the use of yard space.

[0005] In some embodiments of the present application, the AI ​​intelligent agent processes the characteristic parameters of the containers to be stacked, selects the corresponding positioning sub-strategy, and quickly selects the appropriate stacking point through the positioning sub-strategy and the electronic monitoring map. It also intelligently allocates the stacking positions through the overall planning of all the containers to be stacked, thereby reducing the moving distance within the yard and avoiding repeated scheduling.

[0006] In some embodiments of the present application, multiple initial sub-strategies are generated by reasonably allocating the number series of the first-level stacking points of each container to be stacked, and each initial sub-strategy is simulated and analyzed according to the simulation sub-model to determine the overall space utilization and scheduling efficiency of the yard within the prediction period, and the corresponding stacking sub-strategy is set according to the optimization result to improve the stacking efficiency of containers at the port terminal and optimize the use of yard space.

[0007] In some embodiments of the present application, a method for automatically positioning a container stacking position based on AI is provided, comprising: Set multiple monitoring sub-areas based on port yard parameters and build an electronic map of the yard; An intelligent agent is constructed based on historical stacking parameters, and the intelligent agent sets a primary stacking plan based on the processing results of the stacking container parameters; Generate the operating efficiency value of each monitoring sub-area according to the preset evaluation time node, and determine whether to generate the update instruction of the intelligent agent according to all the operating efficiency values; When multiple monitoring sub-areas are set, they include: Establish a monitoring sub-area sequence A, A=(a1, a2…a i …a n ), where a i is the ith monitoring sub-area; n is the number of monitoring sub-areas.

[0008] In some embodiments of the present application, when constructing an intelligent agent, it includes: Set multiple characteristic evaluation indicators based on historical stacking parameters; Construct multiple box categories based on all feature evaluation indicators; Establish a box category sequence B, B=(b1,b2…b i …b m1 ), where b i is the i-th box category; m1 is the number of box categories; Sequentially set the i-th box category as the target box category; Generate a training data packet of the target box category based on historical stacking parameters; Generate a location-finding substrategy for the target box category based on the training data packet; Generate the location-finding sub-strategy for each box category in turn; Construct an agent based on all the location-finding sub-strategies.

[0009] In some embodiments of the present application, when the intelligent agent sets a primary stacking plan according to the processing result of the stacking container parameters, it includes: Establish a sequence C of containers to be stacked according to the parameters of the containers to be stacked; C=(c1,c2…c i …c m2 ), where c i is the i-th container to be stacked, m2 is the number of containers to be stacked; Sequentially set the i-th container to be stacked as the target container; The intelligent agent extracts the characteristic parameters of the target container and generates a first-level location strategy for the target container; Select multiple primary stacking points for target containers based on the electronic monitoring map and primary location strategy; Establish the first-level stacking point sequence P of the target container, P=(p1, p2…p i …p r ), where p i is the i-th primary stacking point of the target container; r is the number of primary stacking points of the target container; Sequentially establish a primary stacking point number list for each container to be stacked; Construct multiple initial sub-strategies based on all the first-level stacking point series; A first-level stacking strategy is set based on all initial sub-strategies.

[0010] In some embodiments of the present application, when the primary stacking strategy is set according to all the initial sub-strategies, it includes: Establish the initial sub-strategy sequence W, W = (w1, w2…w i …w n1 ), where w i is the i-th initial sub-strategy; n1 is the number of initial sub-strategies; Sequentially set the i-th initial sub-strategy as the target initial sub-strategy; Generate the scheduling evaluation value f of the target initial sub-strategy; Generate scheduling evaluation values ​​of each initial sub-strategy in turn; Establish a scheduling evaluation value series F, F=(f1, f2…f i …f n1 ), where f i is the scheduling evaluation value of the i-th initial sub-strategy; n1 is the number of initial sub-strategies; Set the maximum value f in the scheduling evaluation value sequence F max The corresponding initial sub-strategy is the first-level stacking strategy.

[0011] In some embodiments of the present application, when generating the scheduling evaluation value f of the target initial sub-strategy, it includes: Establish forecast cycle and simulation sub-models; Generate expected load parameters of the target initial sub-strategy within the forecast period according to the simulation sub-model; Generate the target initial sub-strategy load evaluation value U according to the expected load parameters; Generate a stacking evaluation value for each container to be stacked in the target initial sub-strategy; Establish a stacking evaluation value series D, D = (d1, d2…d i …d m2 ), where d i is the stacking evaluation value of the i-th container to be stacked generated based on the target initial sub-strategy; m2 is the number of containers to be stacked; Generate a scheduling evaluation value f; f=e1*Q1* d i ]+e2*Q2* Y1(i)*(d i -d')]+e3*Q3*U; Wherein, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; e3 is the preset third weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; Q3 is the preset third fixed coefficient; Y 1（i) is the selection coefficient; if (d i -d')>0,Y1(i)=1; if (d i -d')>0,Y1(i)=0; d' is the preset scheduling evaluation value threshold.

[0012] In some embodiments of the present application, when generating a bearer evaluation value U of a target initial sub-strategy according to an expected bearer parameter, the method includes: U=e4*Q4*[ β 1i *(j i -j' i) ]+e5*Q5*[ β 2i *t i ]; Wherein, e4 is the preset fourth weight coefficient; e5 is the preset fifth weight coefficient; Q4 is the preset fourth fixed coefficient; Q5 is the preset fifth fixed coefficient; β 1i is the impact factor of the i-th box category; j i The remaining load of the i-th box category is generated based on the expected load parameters; j' i is the load safety threshold of the i-th box category in the prediction period; θ1 is the number of load evaluation indicators; β 2i is the influencing factor of the i-th load evaluation index; t i is the reference value of the i-th load evaluation index generated based on the expected load parameters.

[0013] In some embodiments of the present application, when generating the stacking evaluation value of each container to be stacked in the target initial sub-strategy, it includes: According to the sequence C of containers to be stacked, set c i The container to be evaluated; Generate a stacking evaluation value d of the container to be evaluated according to the target initial sub-strategy; d= η 1i *g i; Among them, θ2 is the number of stacking evaluation indicators; η 1i is the influencing factor of the i-th stacking evaluation index; g i is the reference value of the i-th stacking evaluation index of the container to be evaluated in the target initial sub-strategy.

[0014] In some embodiments of the present application, when generating the operating efficiency value of each monitoring sub-area, it includes: According to the monitoring sub-area sequence A, set a i Monitor sub-areas for the target; Obtain the monitoring data packet of the target monitoring sub-area at the current evaluation time node; Generate the operation efficiency value h of the target monitoring sub-region according to the monitoring data packet; h = η 2i * v i ; Among them, θ3 is the number of efficiency evaluation indicators; η 2i is the influence factor of the i-th efficiency evaluation indicator; Ⅵ is the reference value of the i-th efficiency evaluation indicator in the target monitoring sub-region; Generate the operation efficiency values of each monitoring sub-region in sequence; Establish an operation efficiency value sequence H at the current evaluation time node, H = (h1, h2…h i …h n ), where h i is the operation efficiency value of the i-th monitoring sub-region at the current evaluation time node; n is the number of monitoring sub-regions.

[0015] In some embodiments of the present application, when determining whether to generate an update instruction for the intelligent agent, it includes: Generate an update evaluation value s at the current evaluation time node according to the operation efficiency value sequence H; s = e6 * Q6 * Y2(i) * (h i - h') 2 + e7 * Q7 * (h i - Δh) 2 Among them, e6 is a preset sixth weight coefficient; e7 is a preset seventh weight coefficient; Q6 is a preset sixth fixed coefficient; Q7 is a preset seventh fixed coefficient; Y 2（ i) is a selection coefficient; if (h i - h') > 0, Y2 (i) = 1; if (h i - h') < 0, Y2(i) = 0; h' is a preset operation efficiency value threshold; Δh is the average value of all data in the operation efficiency value sequence H; Determine whether to generate an update instruction according to the update evaluation value s.

[0016] In some embodiments of the present application, when determining whether to generate an update instruction according to the update evaluation value s, it includes: Preset a first update evaluation value threshold S1 and a second update evaluation value threshold S2, and S1 < S2; If s < S1, no update instruction is generated at the current evaluation time node; If S1 ≤ s < S2, a first-level update instruction is generated at the current evaluation time node; If s ≥ S2, a second-level update instruction is generated at the current evaluation time node.​

[0017] Compared with the prior art, the automatic positioning method for container stacking positions based on Al in the embodiment of the present application has the following beneficial effects: The AI ​​agent processes the characteristic parameters of the containers to be stacked, selects the corresponding location-finding sub-strategy, and quickly selects the appropriate stacking point through the location-finding sub-strategy and the electronic monitoring map. It also intelligently allocates the stacking positions through the overall planning of all containers to be stacked, reduces the moving distance within the yard, and avoids repeated scheduling.

[0018] By reasonably allocating the primary stacking points of each container to be stacked, multiple initial sub-strategies are generated, and each initial sub-strategy is simulated and analyzed according to the simulation sub-model to determine the overall space utilization and scheduling efficiency of the yard within the prediction period. The corresponding stacking sub-strategy is set according to the optimization results to improve the stacking efficiency of containers at the port terminal and optimize the use of yard space. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flow chart of an automatic positioning method for container stacking positions based on Al in a preferred embodiment of the present application. DETAILED DESCRIPTION

[0020] The specific implementation methods of the present application are further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present application but are not intended to limit the scope of the present application.

[0021] In the description of the present application, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0022] The terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0023] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0024] like Figure 1 As shown, a method for automatically positioning a container stacking position based on Al in a preferred embodiment of the present application includes: S101: setting multiple monitoring sub-areas based on port yard parameters and constructing an electronic map of the yard; S102: constructing an intelligent agent according to historical stacking parameters, and the intelligent agent sets a primary stacking plan according to the processing result of the stacking container parameters; S103: Generate an operating efficiency value for each monitoring sub-area according to a preset evaluation time node, and determine whether to generate an update instruction for the intelligent agent according to all operating efficiency values; When multiple monitoring sub-areas are set, they include: Establish a monitoring sub-area sequence A, A=(a1, a2…a i …a n ), where a i is the ith monitoring sub-area; n is the number of monitoring sub-areas.

[0025] Specifically, multiple monitoring sub-areas are set according to the yard parameters of the port terminal and the road parameters inside the yard, and corresponding monitoring devices are set in each monitoring sub-area to collect container stacking data in each monitoring sub-area in real time.

[0026] Specifically, an electronic monitoring map is constructed based on the collected container stacking data of each monitoring sub-area, so as to monitor the status of each container inside the yard and the space usage status of the yard.

[0027] Specifically, when building an agent, it includes: Set multiple characteristic evaluation indicators based on historical stacking parameters; Construct multiple box categories based on all feature evaluation indicators; Establish a box category sequence B, B=(b1,b2…b i …b m1 ), where b i is the i-th box category; m1 is the number of box categories; Sequentially set the i-th box category as the target box category; Generate a training data packet of the target box category based on historical stacking parameters; Generate a location-finding substrategy for the target box category based on the training data packet; Generate the location-finding sub-strategy for each box category in turn; Construct an agent based on all the location-finding sub-strategies.

[0028] Specifically, the characteristic evaluation indicators include, but are not limited to, the overall weight of the container (empty container plus cargo), the length of container stay (i.e. the difference between the expected arrival time and the expected departure time), the type of cargo inside the container, the size of the container and other parameters.

[0029] Specifically, by quantifying each feature evaluation index, multiple value ranges of each feature evaluation index are generated, and multiple box categories are constructed according to the random combination results.

[0030] Specifically, by collecting all container stacking parameters corresponding to a single container category in the historical stacking parameters, a corresponding training data packet is generated, and through iterative optimization analysis of the training data packet, the location-finding sub-strategy of the current container category is set.

[0031] Specifically, the focus of the location-finding sub-strategies within each container category and the evaluation criteria for whether the stacking points are qualified are different. For example, for containers with a shorter stay time, high stacking points are given priority to facilitate loading and unloading, avoiding other containers being stacked on the current container during the stay period, causing operation delays.

[0032] Specifically, the location-finding sub-strategy includes the optimization strategy for the current box category and the stacking point judgment criteria.

[0033] In a preferred embodiment of the present application, when the intelligent agent sets a primary stacking plan according to the processing result of the stacking container parameters, it includes: Establish a sequence C of containers to be stacked according to the parameters of the containers to be stacked; C=(c1, c2…c i …c m2 ), where c i is the i-th container to be stacked, m2 is the number of containers to be stacked; Sequentially set the i-th container to be stacked as the target container; The intelligent agent extracts the characteristic parameters of the target container and generates a first-level location strategy for the target container; Select multiple primary stacking points for target containers based on the electronic monitoring map and primary location strategy; Establish the first-level stacking point sequence P of the target container, P=(p1, p2…p i …p r ), where pi is the i-th primary stacking point of the target container; r is the number of primary stacking points of the target container; Sequentially establish a primary stacking point number list for each container to be stacked; Construct multiple initial sub-strategies based on all the first-level stacking point series; A first-level stacking strategy is set based on all initial sub-strategies.

[0034] Specifically, the container to be stacked with the least first-level stacking points is set as the first container, and any first-level stacking point in the first-level stacking point sequence corresponding to the first container is randomly selected as the second-level stacking point of the first container. After deleting the second-level stacking point in all the first-level stacking point sequences, the container to be stacked with the least remaining first-level stacking points is selected as the second container, and any remaining first-level stacking point in the first-level stacking point sequence corresponding to the second container is selected as the second-level stacking point of the second container. Similarly, a second-level stacking point is selected for each container to be stacked, and an initial sub-strategy is generated according to all the second-level stacking points. According to the above rules, multiple initial sub-strategies are generated by exhaustive method.

[0035] Specifically, when setting the first-level stacking strategy based on all initial sub-strategies, it includes: Establish the initial sub-strategy sequence W, W = (w1, w2…w i …w n1 ), where w i is the i-th initial sub-strategy; n1 is the number of initial sub-strategies; Sequentially set the i-th initial sub-strategy as the target initial sub-strategy; Generate the scheduling evaluation value f of the target initial sub-strategy; Generate scheduling evaluation values ​​of each initial sub-strategy in turn; Establish a scheduling evaluation value series F, F=(f1, f2…f i …f n1 ), where f i is the scheduling evaluation value of the i-th initial sub-strategy; n1 is the number of initial sub-strategies; Set the maximum value f in the scheduling evaluation value sequence F max The corresponding initial sub-strategy is the first-level stacking strategy.

[0036] Specifically, the larger the scheduling evaluation value, the higher the space utilization rate of the yard by the current initial sub-strategy, the lower the overall cost of the stacking process, and the higher the operation scheduling efficiency.

[0037] It can be understood that in the above embodiment, the AI ​​intelligent agent processes the characteristic parameters of the containers to be stacked, selects the corresponding location-finding sub-strategy, and quickly selects the appropriate stacking point through the location-finding sub-strategy and the electronic monitoring map, and through the overall planning of all containers to be stacked, intelligently allocates the stacking positions, reduces the moving distance in the yard, and avoids repeated scheduling.

[0038] In a preferred embodiment of the present application, when generating the scheduling evaluation value f of the target initial sub-strategy, it includes: Establish forecast cycle and simulation sub-models; Generate expected load parameters of the target initial sub-strategy within the forecast period according to the simulation sub-model; Generate the target initial sub-strategy load evaluation value U according to the expected load parameters; Generate a stacking evaluation value for each container to be stacked in the target initial sub-strategy; Establish a stacking evaluation value series D, D = (d1, d2…d i …d m2 ), where d i is the stacking evaluation value of the i-th container to be stacked generated based on the target initial sub-strategy; m2 is the number of containers to be stacked; Generate a scheduling evaluation value f; f=e1*Q1* d i ]+e2*Q2* Y1(i)*(d i -d')]+e3*Q3*U; Wherein, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; e3 is the preset third weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; Q3 is the preset third fixed coefficient; Y 1（ i) is the selection coefficient; if (d i -d')>0,Y1(i)=1; if (d i -d')>0,Y1(i)=0; d' is the preset scheduling evaluation value threshold.

[0039] Specifically, all parameters in the model are normalized by presetting the first fixed coefficient, the second fixed coefficient and the third fixed coefficient, so that each parameter in the model is within the same value range.

[0040] Specifically, a simulation sub-model is constructed based on the historical scheduling parameters. By establishing a prediction cycle, the simulation sub-model is used to complete the execution of the initial sub-strategy, and the expected operating parameters of the yard within the prediction cycle after the execution are simulated to generate the predicted load parameters.

[0041] Specifically, the predicted carrying parameters are the expected remaining carrying capacity of the same category and the expected remaining carrying capacity in each monitoring sub-area.

[0042] Specifically, when generating the bearer evaluation value U of the target initial sub-strategy according to the expected bearer parameters, it includes: U=e4*Q4*[ β 1i *(j i -j' i) ]+e5*Q5*[ β 2i *t i ]; Wherein, e4 is the preset fourth weight coefficient; e5 is the preset fifth weight coefficient; Q4 is the preset fourth fixed coefficient; Q5 is the preset fifth fixed coefficient; β 1i is the impact factor of the i-th box category; j i The remaining load of the i-th box category is generated based on the expected load parameters; j' i is the load safety threshold of the i-th box category in the prediction period; θ1 is the number of load evaluation indicators; β 2i is the influencing factor of the i-th load evaluation index; t i is the reference value of the i-th load evaluation index generated based on the expected load parameters.

[0043] Specifically, the remaining capacity refers to the number of containers of the current container type that can be received within the forecast period.

[0044] Specifically, the load evaluation index includes but is not limited to multiple parameters such as the average value and variance of the remaining load of each monitoring sub-area.

[0045] Specifically, all parameters in the model are normalized by presetting the fourth fixed coefficient and the fifth fixed coefficient, so that each parameter in the model is within the same value range.

[0046] Specifically, the larger the load evaluation value, the higher the space utilization efficiency of the yard after executing the target initial sub-strategy, and the smaller the possibility of subsequent repeated scheduling.

[0047] Specifically, when generating the stacking evaluation value of each container to be stacked in the target initial sub-strategy, it includes: According to the sequence C of containers to be stacked, set c i The container to be evaluated; Generate a stacking evaluation value d of the container to be evaluated according to the target initial sub-strategy; d= η 1i *g i; Among them, θ2 is the number of stacking evaluation indicators; η 1i is the influencing factor of the i-th stacking evaluation index; g i is the reference value of the i-th stacking evaluation index of the container to be evaluated in the target initial sub-strategy.

[0048] Specifically, the larger the stacking evaluation value is, the higher the adaptability of the stacking position of the current header to be stacked is.

[0049] It can be understood that in the above embodiment, multiple initial sub-strategies are generated by reasonably allocating the number series of the first-level stacking points of each container to be stacked, and each initial sub-strategy is simulated and analyzed according to the simulation sub-model to determine the overall space utilization and scheduling efficiency of the yard within the prediction period, and the corresponding stacking sub-strategy is set according to the optimization result to improve the stacking efficiency of containers at the port terminal and optimize the use of yard space.

[0050] In a preferred embodiment of the present application, when generating the operating efficiency value of each monitoring sub-area, it includes: According to the monitoring sub-area sequence A, set a i Monitor sub-areas for the target; Obtain the monitoring data packet of the target monitoring sub-area at the current evaluation time node; Generate the operating efficiency value h of the target monitoring sub-area according to the monitoring data packet; h=[ η 2i *v i ]; Among them, θ3 is the number of efficiency evaluation indicators; η 2i is the influencing factor of the i-th efficiency evaluation index; VI is the reference value of the i-th efficiency evaluation index in the target monitoring sub-area; Generate the operating efficiency value of each monitoring sub-area in turn; Establish the operating efficiency value series H of the current evaluation time node, H=(h1, h2…h i …h n ), where h i is the operating efficiency value of the ith monitoring sub-area at the current evaluation time node; n is the number of monitoring sub-areas.

[0051] Specifically, the efficiency evaluation indicators include but are not limited to the average outbound time of containers in the monitoring sub-area, the frequency of repeated dispatch, the average stacking time of containers and other parameters. The larger the operating efficiency value, the higher the space utilization efficiency in the target monitoring sub-area, the faster the overall circulation efficiency of its containers, and the higher the operation delay rate.

[0052] Specifically, when determining whether to generate an update instruction for an agent, it includes: Generate the updated evaluation value s at the current evaluation time node based on the operation efficiency value sequence H; s = e6 * Q6 * Y2(i) * (h i - h') 2 + e7 * Q7 * (h i - Δh) 2 ) where, e6 is the preset sixth weight coefficient; e7 is the preset seventh weight coefficient; Q6 is the preset sixth fixed coefficient; Q7 is the preset seventh fixed coefficient; Y 2（ i) is the selection coefficient; if (h i - h') > 0, Y2 Y2(i) = 1; if (h i - h') < 0, Y2(i) = 0; h' is the preset operation efficiency value threshold; Δh is the average value of all data in the operation efficiency value sequence H; Judge whether to generate an update instruction according to the updated evaluation value s.

[0053] Specifically, the larger the updated evaluation value, the lower the overall scheduling efficiency of the current intelligent agent, the greater the degree of space usage in the yard to be optimized, and the higher the frequency of job delays. It is necessary to optimize and train the current intelligent agent in a timely manner to improve its ability to locate and schedule containers.

[0054] Specifically, normalize all parameters in the model through the preset sixth fixed coefficient and seventh fixed coefficient, so that each parameter in the model is within the same value range.

[0055] Specifically, when judging whether to generate an update instruction according to the updated evaluation value s, it includes: Preset the first updated evaluation value threshold S1 and the second updated evaluation value threshold S2, and S1 < S2; If s < S1, no update instruction is generated at the current evaluation time node; If S1 ≤ s < S2, a first-level update instruction is generated at the current evaluation time node; If s ≥ S2, a second-level update instruction is generated at the current evaluation time node.

[0056] Specifically, the first updated evaluation value threshold and the second updated evaluation value threshold can be set according to historical parameters.

[0057] Specifically, the first-level update instruction means analyzing each monitoring sub-region with low operation efficiency, screening out the optimization data packets of each location search sub-strategy, and updating the intelligent agent through all the optimization data packets.

[0058] Specifically, the secondary update instruction refers to re-dividing the box categories within the intelligent body and reconstructing the corresponding location-finding sub-strategy to complete the update of the intelligent body.

[0059] According to the first concept of the present application, the AI ​​intelligent agent processes the characteristic parameters of the containers to be stacked, selects the corresponding positioning sub-strategy, and quickly selects the appropriate stacking point through the positioning sub-strategy and the electronic monitoring map. It also intelligently allocates the stacking positions through the overall planning of all the containers to be stacked, thereby reducing the moving distance within the yard and avoiding repeated scheduling.

[0060] According to the second concept of the present application, multiple initial sub-strategies are generated by reasonably allocating the number series of the first-level stacking points of each container to be stacked, and each initial sub-strategy is simulated and analyzed according to the simulation sub-model to determine the overall space utilization and scheduling efficiency of the yard within the prediction period, and the corresponding stacking sub-strategy is set according to the optimization result to improve the stacking efficiency of containers at the port terminal and optimize the use of yard space.

[0061] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and substitutions can be made without departing from the technical principles of the present application. These improvements and substitutions should also be regarded as the scope of protection of the present application.

Claims

1. An automatic positioning method for container stacking positions based on AI, characterized in that: Including: Set multiple monitoring sub - regions based on port yard parameters and construct a yard electronic map; Construct an agent according to historical stacking parameters. The agent sets a first - level stacking plan based on the processing results of the parameters of the containers to be stacked; Generate the operation efficiency values of each monitoring sub - region according to the preset evaluation time nodes, and judge whether to generate an update instruction for the agent based on all the operation efficiency values; Among them, when setting multiple monitoring sub - regions, it includes: Establish a monitoring sub-area sequence A, A=(a1, a2…a i …a n ), where a i is the ith monitoring sub-area; n is the number of monitoring sub-areas.

2. The automatic positioning method for container stacking positions based on A1 as claimed in claim 1 is characterized in that: When constructing an agent, it includes: Set multiple characteristic evaluation indicators according to historical stacking parameters; Construct multiple container categories based on all the characteristic evaluation indicators; Establish a box category sequence B, B=(b1,b2…b i …b m1 ), where b i is the i-th box category; m1 is the number of box categories; Set the i - th container category as the target container category in sequence; Generate a training data packet for the target container category according to historical stacking parameters; Generate a location - finding sub - strategy for the target container category according to the training data packet; Generate the location - finding sub - strategies for each container category in sequence; Construct an agent according to all the location - finding sub - strategies.

3. The automatic positioning method for container stacking positions based on A1 as claimed in claim 2 is characterized in that: When the agent sets a first - level stacking plan based on the processing results of the parameters of the containers to be stacked, it includes: Establish a sequence C of containers to be stacked according to the parameters of the containers to be stacked; C=(c1,c2…c i …c m2 ), where c i is the i-th container to be stacked, m2 is the number of containers to be stacked; Set the i - th container to be stacked as the target container in sequence; The agent extracts the characteristic parameters of the target container and generates a first - level location - finding strategy for the target container; Select multiple first - level stacking points for the target container according to the electronic monitoring map and the first - level location - finding strategy; Establish the first-level stacking point sequence P of the target container, P=(p1, p2…p i …p r ), where p i is the i-th primary stacking point of the target container; r is the number of primary stacking points of the target container; Establish a sequence of first - level stacking points for each container to be stacked in sequence; Construct multiple initial sub - strategies according to all the sequences of first - level stacking points; Set a first - level stacking strategy according to all the initial sub - strategies.

4. The automatic positioning method for container stacking positions based on A1 as claimed in claim 3 is characterized in that: When setting a first - level stacking strategy according to all the initial sub - strategies, it includes: Establish the initial sub-strategy sequence W, W = (w1, w2…w i …w n1 ), where w i is the i-th initial sub-strategy; n1 is the number of initial sub-strategies; Set the i - th initial sub - strategy as the target initial sub - strategy in sequence; Generate a scheduling evaluation value f for the target initial sub - strategy; Generate the scheduling evaluation values for each initial sub - strategy in sequence; Establish a scheduling evaluation value series F, F=(f1, f2…f i …f n1 ), where f i is the scheduling evaluation value of the i-th initial sub-strategy; n1 is the number of initial sub-strategies; Set the maximum value f in the scheduling evaluation value sequence F max The corresponding initial sub-strategy is the first-level stacking strategy.

5. The automatic positioning method for container stacking positions based on A1 as claimed in claim 4 is characterized in that: When generating a scheduling evaluation value f for the target initial sub - strategy, it includes: Establish a prediction period and a simulation sub - model; Generate the expected load - bearing parameters of the target initial sub - strategy within the prediction period according to the simulation sub - model; Generate a load - bearing evaluation value U for the target initial sub - strategy according to the expected load - bearing parameters; Generate the stacking evaluation values of each container to be stacked in the target initial sub - strategy; Establish a stacking evaluation value series D, D = (d1, d2…d i …d m2 ), where d i is the stacking evaluation value of the i-th container to be stacked generated based on the target initial sub-strategy; m2 is the number of containers to be stacked; Generate a scheduling evaluation value f; f=e1*Q1* d i ]+e2*Q2* Y1(i)*(d i -d')]+e3*Q3*U; Wherein, e1 is the preset first weight coefficient; e2 is the preset second weight coefficient; e3 is the preset third weight coefficient; Q1 is the preset first fixed coefficient; Q2 is the preset second fixed coefficient; Q3 is the preset third fixed coefficient; Y 1（ i) is the selection coefficient; if (d i -d')>0,Y1(i)=1; if (d i -d')>0,Y1(i)=0; d' is the preset scheduling evaluation value threshold.

6. The automatic positioning method for container stacking positions based on A1 as claimed in claim 5, characterized in that: When generating a load - bearing evaluation value U for the target initial sub - strategy according to the expected load - bearing parameters, it includes: U=e4*Q4*[ β 1i *(j i -j' i) ]+e5*Q5*[ β 2i *t i ]; Wherein, e4 is the preset fourth weight coefficient; e5 is the preset fifth weight coefficient; Q4 is the preset fourth fixed coefficient; Q5 is the preset fifth fixed coefficient; β 1i is the impact factor of the i-th box category; j i The remaining load of the i-th box category is generated based on the expected load parameters; j' i is the load safety threshold of the i-th box category in the prediction period; θ1 is the number of load evaluation indicators; β 2i is the influencing factor of the i-th load evaluation index; t i is the reference value of the i-th load evaluation index generated based on the expected load parameters.

7. The automatic positioning method for container stacking positions based on A1 as claimed in claim 5, characterized in that: When generating the stacking evaluation values of each container to be stacked in the target initial sub - strategy, it includes: According to the sequence C of containers to be stacked, set c i The container to be evaluated; Generate a stacking evaluation value d for the container to be evaluated according to the target initial sub - strategy; d= or 1i *g i; Among them, θ2 is the number of stacking evaluation indicators; η 1i is the influencing factor of the i-th stacking evaluation index; g i is the reference value of the i-th stacking evaluation index of the container to be evaluated in the target initial sub-strategy.

8. The automatic positioning method for container stacking positions based on A1 as claimed in claim 5, characterized in that: When generating the operation efficiency values of each monitoring sub - region, it includes: According to the monitoring sub-area sequence A, set a i Monitor sub-areas for the target; Obtain the monitoring data packet of the target monitoring sub - region at the current evaluation time node; Generate the operation efficiency value h of the target monitoring sub - region according to the monitoring data packet; h=[ or 2i *v i ]; Among them, θ3 is the number of efficiency evaluation indicators; η 2i is the influencing factor of the i-th efficiency evaluation index; VI is the reference value of the i-th efficiency evaluation index in the target monitoring sub-area; Generate the operation efficiency values of each monitoring sub - region in sequence; Establish the operating efficiency value sequence H of the current evaluation time node, H=(h1, h2…h i …h n ), where h i is the operating efficiency value of the ith monitoring sub-area at the current evaluation time node; n is the number of monitoring sub-areas.

9. The automatic positioning method for container stacking positions based on A1 as claimed in claim 8, characterized in that: When judging whether to generate an update instruction for the agent, it includes: Generate an update evaluation value s at the current evaluation time node according to the sequence H of operation efficiency values; s=e6*Q6* Y2(i)*(h i -h') 2 ]+e7*Q7* (h i -Δh) 2 ] Among them, e6 is the preset sixth weight coefficient; e7 is the preset seventh weight coefficient; Q6 is the preset sixth fixed coefficient; Q7 is the preset seventh fixed coefficient; Y 2（ i) is the selection coefficient; if (h i -h')>0, Y2 (i)=1; if (h i -h') < 0, Y2(i) = 0; h' is the preset operating efficiency value threshold; Δh is the average value of all data in the operating efficiency value series H; Judge whether to generate an update instruction according to the update evaluation value s.

10. The automatic positioning method for container stacking positions based on A1 as claimed in claim 9, characterized in that: When judging whether to generate an update instruction according to the update evaluation value s, it includes: Preset a first update evaluation value threshold S1 and a second update evaluation value threshold S2, and S1 < S2; If s < S1, no update instruction is generated at the current evaluation time node; If S1 ≤ s < S2, a first-level update instruction is generated at the current evaluation time node; If s ≥ S2, a second-level update instruction is generated at the current evaluation time node.

Images