CN-121724551-B - Wharf operation full-flow monitoring system based on Internet of things

Abstract

The invention discloses a wharf operation whole-flow monitoring system based on the Internet of things, which relates to the technical field of wharf operation and comprises the steps of constructing a complete technical link consisting of cargo type association, multi-scale tank granularity generation, tank granularity quantification, virtual tank mapping sand table simulation, tank granularity rewards and optimal tank granularity screening. Through analyzing the change of the box occupation and the cargo type characteristics in the historical box turning operation, multiple box division granularities are generated in a self-adaptive mode, the box turning, box reconstruction and access processes are deduced in the virtual sand table, the influence of different granularities on operation continuity and stability is evaluated in combination with reinforcement learning, the optimal box division granularities suitable for the current operation stage are finally screened, dynamic monitoring and optimization of the storage space operability are achieved, and the problems that the storage space operability is difficult to accurately describe in the existing wharf operation monitoring, the box turning complexity evaluation distortion is caused, and the stability of operation strategies is insufficient are solved.

Inventors

• ZHANG SHENGXUAN
• ZHANG LIANG
• YANG KUN

Assignees

• 上海鼎为物联技术有限公司

Dates

Publication Date

20260505

Application Date

20260213

Claims (5)

1. 1. The wharf operation full-flow monitoring system based on the Internet of things is characterized by comprising a cargo type association module, a multi-scale box granularity generation module, a box granularity quantification module, a virtual box mapping sand table simulation module, a box granularity rewarding module and an optimal box granularity screening module; The cargo type association module is used for acquiring the initial storage yard bin division granularity based on the space autocorrelation length of the bin occupation change in the historical bin turning operation time sequence and combining with cargo type labels; The multi-scale bin granularity generation module is used for performing spatial scale expansion on the initial yard bin division granularity to generate a plurality of second yard bin division granularities different in bin coverage density and bin quantity distribution; The bin granularity quantization module is used for quantizing the storage yard bin division granularity into a coverage bin set range updated by bin states and the minimum bin quantity required by bin turning based on a space autocorrelation structure of bin occupation states and bin communication change in a bin turning shielding release process; The virtual box mapping sand table simulation module is used for quantifying and mapping the division granularity of each second storage yard box to the virtual box mapping sand table, and simulating and deducting the reconstruction of the storage yard box, the box turning path and the access flow to obtain a virtual operation result; the bin granularity rewarding module is used for extracting characteristics based on virtual operation results and generating a yard bin division rewarding value corresponding to each second yard bin division granularity through reinforcement learning; the optimal bin granularity screening module is used for screening the optimal bin division granularity of the second storage yard according to the stability distribution of the bin division rewarding value of the storage yard on the granularity axis, and for monitoring and updating the quay operation in a staged mode; Based on the space autocorrelation length of the box position occupation change in the historical box turning operation time sequence, the initial storage yard box position division granularity is obtained by combining the cargo type label, and the method specifically comprises the following steps: acquiring an occupied state sequence of each bin at a continuous operation time in the historical bin turning operation process, and constructing a space distribution matrix sequence of change of the occupied state of the bin of a storage yard along with time; Carrying out space autocorrelation analysis on the space distribution matrix sequence, calculating an autocorrelation function of the bin occupancy change on the space dimension of the storage yard, and extracting a space correlation length of the occupancy change correlation attenuated to be near zero correlation for the first time; grouping historical box turning operation time sequences according to the cargo type labels, and respectively counting the space correlation lengths corresponding to different cargo types to form a space correlation length set associated with the cargo types; Carrying out weighted aggregation on the space correlation lengths associated with the cargo types to obtain a comprehensive space correlation length reflecting the space coupling characteristic of the current storage yard box turning operation; Taking the comprehensive space correlation length as the minimum stable space unit scale of the box bit occupation change, and determining the initial storage yard box bit division granularity; the stability distribution of the reward value on the granularity axis is divided according to the storage yard boxes, the optimal second storage yard boxes are identified and screened based on a stability interval, and the second storage yard boxes are used for staged monitoring and updating of wharf operation, and the method specifically comprises the following steps: Ordering all the second yard bin division granularities according to the size rule of the spatial scale, and constructing a granularity-rewarding value distribution sequence with bin division granularities as a transverse sequence and corresponding yard bin division rewarding values as a longitudinal sequence; Counting the variation amplitude of the corresponding rewarding value of adjacent granularities segment by adopting a sliding granularity window mode along the granularity sequence of the granularity-rewarding value distribution sequence to form a local variation sequence reflecting the stability of the rewarding value along with the granularity variation; In the local change sequence, identifying a granularity continuous section with the prize value change amplitude continuously lower than a preset change threshold value, and determining the granularity continuous section as a candidate stable section; summarizing the granularity corresponding rewards in each candidate stable interval, and screening the stable interval with the highest overall level of rewards in the interval and the largest interval span as a target stable interval; selecting a second yard bin dividing granularity which corresponds to the average value of the interval closest to the rewarding value in the target stable interval, and determining the second yard bin dividing granularity as the optimal second yard bin dividing granularity; And taking the optimal second yard tank dividing granularity as a tank dividing control standard in the current dock operation period, and continuously collecting actual tank turning operation data in the subsequent operation process.
2. 2. The full-process monitoring system for dock operation based on the internet of things according to claim 1, wherein the performing spatial scale expansion on the initial yard tank division granularity to generate a plurality of second yard tank division granularities different in tank coverage density and tank number distribution specifically comprises: Taking the space unit scale corresponding to the initial storage yard bin dividing granularity as a reference, carrying out equal proportion expansion and contraction on bin dividing units along the storage yard space coordinate axis direction, and constructing a plurality of candidate bin dividing units with different space scales; under the scale of each candidate box division unit, the original box grids of the storage yard are regrouped, adjacent boxes are combined or split into new box units, and a corresponding box division scheme is generated; Respectively counting the space coverage of single box units under each box division scheme and the total number of the box units in a storage yard to form a corresponding relation between the box coverage density and the box number distribution; And marking each bin dividing scheme according to the difference of the bin coverage density and the bin number distribution, and generating a plurality of second storage yard bin dividing granularities different in space coverage density and bin number distribution.
3. 3. The system for monitoring the whole process of wharf operation based on the internet of things according to claim 2, wherein the space autocorrelation structure based on the occupied state of the tank and the tank connection change in the process of turning over and shielding and releasing the tank, the method for quantifying the division granularity of the tank in the storage yard into the coverage tank aggregate range updated by the tank state and the minimum number of the tank required by turning over, specifically comprises the following steps: Aiming at each second yard box division granularity, according to a corresponding box division scheme, mapping the box occupation state in the historical box turning operation time sequence to box units under the granularity to form a box occupation state sequence arranged in time sequence; In the box bit occupation state sequence, the space adjacent relation among the box bit units is taken as constraint, the box bit unit combination presenting synchronous occupation change in the time continuous change process is identified, and the box bit occupation state space association structure under the granularity is constructed; Based on the box occupation state space association structure, expanding the box outwards layer by layer along the target box turning process, determining a box unit set with occupation state change due to shielding relation in the box turning process, and generating a box state updating coverage box set range under corresponding granularity; gradually expanding the box turning shielding release process according to the space stacking relation and the operation reachable sequence of the box units in the range of the box state updating covering box set to form a box communicating sequence reflecting the box unlocking sequence relation; the box position communication sequence is traced back outwards from the target box position turning box position, box position units which must participate in the box turning operation under the condition of meeting shielding release are screened, and the minimum box position set which is required to participate in the operation for completing the target box turning is determined; And updating the bin state to cover the range of the bin set and the scale of the minimum bin set, and taking the bin state as a quantization result corresponding to the bin division granularity of the second storage yard.
4. 4. The system for monitoring the whole process of wharf operation based on the internet of things according to claim 3, wherein the steps of quantifying and mapping the division granularity of each second yard tank bit to a virtual tank bit mapping sand table, and performing simulation deduction on the reconstruction of the yard tank bit, the box turning path and the in-out process to obtain a virtual operation result are as follows: Aiming at each second storage yard box division granularity, reading the corresponding box state to update the range of the covering box set and the minimum box set required by box turning, and constructing a box unit topological structure consistent with the actual storage yard space in the virtual box mapping sand table; In the virtual box mapping sand table, the minimum box set is used as a core operation area, and the box units in the range covering the box set are initialized in state to form a virtual storage yard initial layout consistent with the initial state of the historical box turning operation; Based on the topological structure of the box units and the box communication sequence, gradually executing a box turning shielding releasing process in a virtual box mapping sand table, generating a box turning operation sequence consistent with the box unlocking sequence, and synchronously updating the occupancy state of the corresponding box units; in the execution process of the box turning operation sequence, a moving path set of box turning operation is constructed according to the space adjacent relation of box position units in a virtual storage yard, and a box turning path corresponding to the granularity lower box position dividing scheme is formed; after the box turning operation sequence is completed, an entrance-exit task is introduced into the virtual box mapping sand table, and box allocation and path passing expansion are carried out on the storage yard state after box reconstruction, so that a complete entrance-exit operation process sequence is formed; And summarizing the execution results of the box turning operation sequence, the box turning path and the access operation process sequence in the virtual box mapping sand table, and generating a virtual operation result corresponding to the second storage yard box division granularity.
5. 5. The full-process monitoring system for dock operation based on the internet of things according to claim 4, wherein the feature extraction is performed based on the virtual operation result, and the yard bin division rewards corresponding to each second yard bin division granularity is generated through reinforcement learning, specifically: Aiming at each second yard box division granularity, reading the corresponding virtual operation result, and segmenting the virtual operation process according to the box turning operation execution stage, the box turning completion stage and the in-out operation stage; In the execution stage of the box turning operation, counting the number of box bit units actually participating in the operation in a box turning operation sequence, comparing with the minimum box bit set scale under the granularity to form an operation redundancy characteristic reflecting the deviation degree of the box turning operation complexity, and describing the problem that the partial shielding relationship is not sufficiently marked when the box bit classification granularity is too thick; In the execution stage of the box turning operation, recording the continuous change condition of the occupation state of the box bit units along a box turning path, counting the total number of the box bit units triggered to be updated in the box turning process, and taking the total number as a state disturbance characteristic for describing the problem of frequent state change caused by tiny operation when the box bit division granularity is too fine; In the box turning completion stage, counting the number of box bit unit reconstruction times caused by box bit merging or splitting in the virtual storage yard, and taking the number of box bit reconstruction times in a unit box turning task as a space fragmentation characteristic for reflecting the condition of reduced storage yard space stability when granularity is too fine; In the entrance and exit operation stage, counting the number of times of path re-planning in the exit and exit path, and taking the number of times as a path stability characteristic for reflecting the matching degree between the box division granularity and the whole predictability of the storage yard; Combining the operation redundancy feature, the state disturbance feature, the space fragmentation feature and the path stability feature according to a fixed sequence to form a virtual operation feature vector corresponding to the division granularity of the second storage yard bin; Taking the division granularity of each second storage yard bin as an action index, taking a corresponding virtual operation feature vector as a state input, and taking the maintenance condition of the whole operation continuity of the virtual storage yard after the box turning operation is completed as a return signal to construct a reinforcement learning training sample; Iteratively updating the reinforcement learning process based on the training samples to obtain reinforcement learning mapping relations for outputting the grading rewarding values of the second storage yard boxes; And inputting the virtual operation feature vector corresponding to each second storage yard bin division granularity into the reinforcement learning mapping relation, and outputting a corresponding storage yard bin division rewarding value.

Description

Wharf operation full-flow monitoring system based on Internet of things Technical Field The invention relates to the technical field of wharf operation, in particular to a wharf operation whole-flow monitoring system based on the Internet of things. Background With the wide application of the internet of things in ports and terminals, the conventional full-flow monitoring system for the operation of the terminals continuously senses the environments of containers, loading and unloading equipment and storage yards, realizes unified management of the stacking, box turning and scheduling processes, improves the operation transparency and the information communication capability to a certain extent, and becomes an important technical foundation for supporting the operation of the high-strength terminals. Under the high-density storage yard operation condition, the container presents a multi-layer, multi-column and frequently-adjusted space distribution form, the mapping relation between the sensing information of the internet of things and the actual space structure is directly determined by the division mode of storage yard boxes, and the granularity level influences the expression fineness of the occupied state of the boxes and the understandability of the space relation, so that the method is a non-negligible core link in the whole process monitoring. In the prior art, too high or too low storage yard tank bit division granularity can obviously influence the whole process monitoring of wharf operation based on the Internet of things, and the unbalance of granularity easily causes deviation of space description from actual operation conditions, so that a monitoring system is difficult to truly reflect the crowding degree and the operation feasibility of a storage yard, and the supporting capability of the tank turning risk and the path conflict is weakened. When the storage yard tank bit division granularity is low, a plurality of containers are often merged into the same operable area, the internet of things system can only acquire the approximate location information of the tank body, and local spatial relations such as multi-layer shielding and mutual pressing are difficult to embody, so that the actual complexity of the tank turning operation is underestimated, and deviation is generated between monitoring data and real operation risks. When the storage yard tank bit division granularity is higher, the tank bit state frequently changes along with the tiny movement of the container, the storage yard space is highly discrete and unstable in the system, and the state fluctuation amplification and the frequent adjustment of the scheduling decision are easy to be caused, so that the whole process monitoring is difficult to keep stable space cognition under the dynamic tank turning scene. The existing scheme generally lacks a mechanism for reasonably restricting and matching the division granularity of the storage yard tank bit, is difficult to coordinate by combining the stacking height, the tank type diversity and the real-time congestion change, ensures that the sensing result of the Internet of things stays at the static tank bit layer, cannot be effectively converted into the process feasibility judgment under the space restriction, and restricts the application effect of the whole-process monitoring of the wharf operation based on the Internet of things in the complex storage yard scene. The present invention proposes a solution to the above-mentioned problems. Disclosure of Invention In order to overcome the defects of the prior art, the embodiment of the invention provides a wharf operation whole-flow monitoring system based on the Internet of things, which solves the problems that the operability of a storage yard space is difficult to accurately describe, the evaluation distortion of box turning complexity and the stability of an operation strategy are insufficient in the existing wharf operation monitoring through box division granularity self-adaptive modeling and virtual operation simulation evaluation based on the perception data of the Internet of things. In order to achieve the above purpose, the present invention provides the following technical solutions: A wharf operation whole-flow monitoring system based on the Internet of things comprises a cargo type association module, a multi-scale bin granularity generation module, a bin granularity quantization module, a virtual bin mapping sand table simulation module, a bin granularity rewarding module and an optimal bin granularity screening module, wherein the cargo type association module is used for obtaining an initial yard bin division granularity by combining a cargo type tag according to the space autocorrelation length of bin occupation change in a historical bin turnover operation time sequence, the multi-scale bin granularity generation module is used for carrying out space scale expansion on the initial yard bin division granularity to g