CN-121638557-B - Efficiency prediction system for cooperation of adjacent port industry and port logistics

CN121638557BCN 121638557 BCN121638557 BCN 121638557BCN-121638557-B

Abstract

The invention discloses an efficiency prediction system for cooperation of the adjacent port industry and port logistics, which relates to the technical field of port logistics efficiency prediction and comprises the following steps of collecting residence time of cargoes of the adjacent port industry in a port, and collecting turnover time of trucks transporting the cargoes of the adjacent port industry in the port, so as to obtain cargo residence time data and truck turnover time data; the method is used for solving the problems that when the existing port logistics efficiency prediction technology is used for analyzing and predicting the residence time of cargoes in a port in the adjacent port industry, a prediction model cannot be built according to the historical residence time of the cargoes in the port, and the residence time of the cargoes in the port is objectively and stably predicted.

Inventors

FANG YAN
NING TAO
LIU GUOQING

Assignees

交通运输部水运科学研究所

Dates

Publication Date: 20260512
Application Date: 20251202

Claims (8)

1. The efficiency prediction system for the cooperation of the adjacent port industry and the port logistics is characterized by comprising a data collection module, a layering statistics module, an analysis correction module and an efficiency prediction module; The data collection module is used for collecting the residence time of cargoes in ports of the adjacent port industry and collecting the turnover time of trucks transporting cargoes in ports of the adjacent port industry, so as to obtain the cargo residence time data and the truck turnover time data; The layering statistical module comprises a first processing unit and a second processing unit, wherein the first processing unit is used for layering calculation processing of cargo retention time data to obtain average retention data; The analysis and correction module is used for respectively carrying out analysis and correction processing on the basis of the average retention data and the average turnover data to obtain reference retention turnover data; The efficiency prediction module is used for constructing an efficiency prediction model based on the reference retention turnover data and predicting the average retention time of cargoes; the analysis and correction module is configured with an analysis and correction strategy, and the analysis and correction strategy comprises: Taking k4 data from front and back of TL (i) as adjacent data of TL (i), calculating average value and standard deviation of the adjacent data, and respectively recording as GP (i) and GB (i) in sequence, wherein k4 is set number; If the absolute difference between TL (i) and GP (i) is larger than k5 GB (i), marking TL (i) as a suspected abnormal point, otherwise marking TL (i) as a normal data point, wherein k5 is a set proportionality coefficient; If TL (i) is a suspected abnormal point, taking TL (i-k 6) to TL (i) to perform linear fitting to obtain a fitted slope, and marking the fitted slope as E1, and taking TL (i) to TL (i+k6) to perform linear fitting to obtain a fitted slope, and marking the fitted slope as E2; Recording the gradient difference CE corresponding to TL (i) as |E1-E2|, repeatedly calculating gradient differences corresponding to all data in the retention time sequence, taking the gradient differences of the k6 data before TL (i) as a front neighborhood gradient difference set, and taking the gradient differences of the k6 data after TL (i) as a rear neighborhood gradient difference set, wherein k6 is the set number; Calculating the average value and standard deviation of the front neighborhood slope difference set, and sequentially marking the average value and standard deviation as HP1 and HB1, and calculating the average value and standard deviation of the rear neighborhood slope difference set, and sequentially marking the average value and standard deviation as HP2 and HB2; If the absolute difference between CE and HP1 is greater than k7 HB1 and the absolute difference between CE and HP2 is greater than k7 HB2, then the TL (i) is marked as an event outlier, otherwise the TL (i) is marked as a normal data point, and all event outliers in the retention time sequence are repeatedly acquired.
2. The system for predicting efficiency of a cooperation between a near-harbor industry and a harbor logistics according to claim 1, wherein the data collection module is configured with a data collection policy, the data collection policy comprising: The method comprises the steps of marking any port to be predicted as a first port, marking any adjacent port industry corresponding to the first port as a first adjacent port industry, marking goods which need to be transported through the first port by the first adjacent port industry as first goods, marking a container carrying the first goods as a first container, marking a truck for transporting the first container as a first truck, and marking a first load carrying the first container as a first truck; The time period from the arrival of any one first container at the first port to the departure of the first port is recorded as the residence time of the corresponding first container at the port; The period of time that any one first truck takes from entering the first port to leaving the first port is taken as the corresponding first truck's turn-around time at the port.
3. The system for predicting efficiency of a cooperation between a near-harbor industry and a harbor logistics according to claim 2, wherein the data collection strategy further comprises: Setting the collection time period as T1, and recording the collection time period as a first time period for any one of the collection time periods; in a first time period, collecting the turnover time of each first truck at a port, and recording the turnover time information of the trucks in the first time period; and periodically and repeatedly acquiring cargo retention time information and truck turnover time information of each acquisition time period, and respectively recording the cargo retention time information and the truck turnover time information as cargo retention time data and truck turnover time data according to time sequence.
4. The system for predicting efficiency of a cooperation between a near-harbor industry and a harbor logistics according to claim 3, wherein the first processing unit is configured with a first processing policy, the first processing policy comprising: Arranging cargo retention time information in a first time period according to a sequence from small to large, marking the information as a first retention sequence, calculating an average value and a standard deviation of the first retention sequence, marking the information as an AP and an AB according to sequential distribution, removing retention time which is not positioned in [ AP-k1 x AB, AP+k1 x AB ] in the first retention sequence, and obtaining a second retention sequence after finishing, wherein k1 is a set proportionality coefficient; Setting interval duration as t2, setting a plurality of continuous duration intervals according to t2, and sequentially recording as duration interval 1-duration intervals n1, n1 as the total number of the set duration intervals; according to the second retention sequence, counting the number of retention time contained in each of the time length sections 1-2, and respectively recording the time length section with the largest number of the contained retention time and the time length section with the largest number of the contained retention time as a first section and a second section.
5. The system for predicting efficiency in cooperation with port logistics in a clinical industry of claim 4, wherein the first processing strategy further comprises: The method comprises the steps of expanding a first interval to two sides by one CB to obtain a first dense interval, expanding a second interval to two sides by one CB to obtain a second dense interval, taking a union of the first dense interval and the second dense interval as an effective dense interval, marking the retention time of the second retention sequence in the effective dense interval as normal retention time, counting the total number of the normal retention time, and marking the total number as AF; Uniformly dividing the effective dense interval into n2 subintervals, sequentially recording the n2 subintervals as dense subintervals 1-dense subintervals n2, and counting the number of normal residence time contained in each dense subinterval, wherein n2 is the set number; The method comprises the steps of recording any one normal retention time as a first retention time LT, recording the number of normal retention times contained in a dense subinterval where the first retention time is positioned as BF, recording BF/AF as the weight of the first retention time, and recording (BF/AF) LT as the weighted retention time of the first retention time; Calculating QT/AQ, which is recorded as the representative retention time of the first time period, repeatedly calculating the representative retention time of each acquisition time period, arranging the representative retention time in time sequence, recording the representative retention time sequence as the average retention data.
6. The system for predicting efficiency in cooperation with port logistics in a clinical port of claim 5, wherein the second processing unit is configured with a second processing strategy, the second processing strategy comprising: Arranging the turnover time information of the trucks in the first time period according to the sequence from small to large, recording the turnover time information as a first turnover sequence, calculating the average value and standard deviation of the first turnover sequence, recording the average value and standard deviation as DP and DB according to the sequence distribution, and eliminating the residence time which is not positioned in [ DP-k2 DB, DP+k2 DB ] in the first turnover sequence, so as to obtain a second turnover sequence after finishing, wherein k2 is a set proportionality coefficient; A nuclear density curve is made for the second turnover sequence, the lowest point between two peaks on the nuclear density curve is obtained and is marked as a valley point, the second turnover sequence is divided into two parts by taking the valley point as a threshold value and is respectively marked as a first turnover rotor sequence and a second turnover rotor sequence; Calculating a median and discrete coefficients of the first peripheral rotor sequence, respectively marking the median and discrete coefficients as CE and CV in sequence, and calculating a corresponding correction representative value of the first peripheral rotor sequence as XT1, wherein XT1=CE (1-k 3: CV), and k3 is a set proportionality coefficient; The method comprises the steps of respectively obtaining the number of turnover time contained in a first turnover rotor sequence and a second turnover rotor sequence, respectively marking the turnover time as WF1 and WF2 in sequence, calculating [ WF 1/(WF 1+WF2) ]x1+ [ WF 2/(WF 1+WF2) ]xXT2, marking the turnover time as a representative turnover time of a first time period, repeatedly calculating the representative turnover time of each acquisition time period, marking the turnover time as a turnover time sequence in time sequence, and marking the turnover time sequence as average turnover data.
7. The system for predicting efficiency of a cooperation between a near-harbor industry and a harbor logistics of claim 6, wherein the analyzing and correcting strategy further comprises: If TL (i) is an event abnormal point, arranging adjacent data of TL (i) according to the sequence of the sizes, acquiring a median, recording the median as neighborhood median data, and replacing TL (i) by using the neighborhood median data; Repeatedly acquiring a correction sequence corresponding to the turnover time sequence, recording the correction sequence as a reference turnover sequence, recording the reference retention sequence and the reference turnover sequence as reference retention turnover data of a first adjacent port industry, and repeatedly acquiring the reference retention turnover data of a plurality of adjacent port industries.
8. The system for efficiency prediction for cooperation with port logistics in a temporary port industry of claim 7, wherein the efficiency prediction module is configured with an efficiency prediction strategy, the efficiency prediction strategy comprising: Constructing an initial prediction model based on BiLSTM models, wherein the initial prediction model comprises an input layer, a bidirectional LSTM layer, a Dropout layer, a full connection layer and an output layer, setting model input as a reference retention sequence and a corresponding reference turnover sequence, and outputting the model output as representative retention time of each acquisition time period in the future; And carrying out model training on the initial prediction model by using the reference retention turnover data, obtaining an efficiency prediction model after completion, and predicting the representative retention time of each acquisition time period in the future of the first harbor industry by using the reference retention turnover data of the first harbor industry to obtain an efficiency prediction result corresponding to the cooperation of the first harbor industry and the harbor logistics.

Description

Efficiency prediction system for cooperation of adjacent port industry and port logistics Technical Field The invention relates to the technical field of port logistics efficiency prediction, in particular to an efficiency prediction system for cooperation of the adjacent port industry and port logistics. Background The port logistics efficiency prediction technology is a comprehensive technical system for providing data support for port scheduling optimization, port industry cooperation and logistics resource allocation by pre-judging the port logistics efficiency core index change trend in a specific period in advance through data preprocessing, model construction and training based on the multi-source operation data of the port logistics whole flow. The importance of the method is that the method can accurately predict the cooperative efficiency of the adjacent port industry and the port logistics, the importance of the method runs through a full chain of port operation optimization, the cost reduction and synergy of the adjacent port industry and regional economic competitiveness are improved, the change of key indexes corresponding to the accurate prediction cooperative efficiency can be adjusted pertinently for the port, the operation smoothness of the logistics is maximized, the reliable cooperative efficiency prediction can guide the industry production plan of the adjacent port industry to avoid the interruption risk of a supply chain, the accurate cooperative efficiency prediction can provide basis for macroscopic decisions such as port space layout, multi-intermodal system perfection and the like, the traditional port logistics efficiency prediction technology can often use data envelope analysis, namely DEA (data analysis) to analyze and predict the retention time of cargoes in the port of the adjacent port industry, DEA (data analysis and interference analysis) is highly sensitive to the data quality, the stability of a prediction result is poor, and the DEA is often required to collect various data, the input and output indexes are selected to have subjectivity, and influence the objectivity of the result, and the traditional port logistics efficiency prediction technology can provide basis for the objective prediction of the retention time of cargoes in the port industry and the retention time of the cargo is stably predicted at the port time according to the retention time of the cargo in the port prediction model. Disclosure of Invention The method aims at solving at least one of the technical problems in the prior art to a certain extent, and solves the problems that when the retention time of cargoes in ports in the adjacent port industry is analyzed and predicted by the existing port logistics efficiency prediction technology, a prediction model cannot be built according to the historical retention time of the cargoes in the ports and the retention time of the cargoes in the ports is objectively and stably predicted by collecting the retention time of the cargoes in the ports of the adjacent port industry and collecting the turnover time of trucks transporting the cargoes in the ports of the adjacent port industry, obtaining average retention data and average turnover data by carrying out layered calculation processing respectively, analyzing and correcting the average retention data respectively, obtaining reference retention turnover data, and constructing an efficiency prediction model based on the reference retention turnover data. In order to achieve the above purpose, the application provides an efficiency prediction system for cooperation of adjacent port industry and port logistics, which comprises a data collection module, a layering statistics module, an analysis correction module and an efficiency prediction module; The data collection module is used for collecting the residence time of cargoes in ports of the adjacent port industry and collecting the turnover time of trucks transporting cargoes in ports of the adjacent port industry, so as to obtain the cargo residence time data and the truck turnover time data; The layering statistical module comprises a first processing unit and a second processing unit, wherein the first processing unit is used for layering calculation processing of cargo retention time data to obtain average retention data; The analysis and correction module is used for respectively carrying out analysis and correction processing on the basis of the average retention data and the average turnover data to obtain reference retention turnover data; The efficiency prediction module comprises a construction unit and a prediction unit, wherein the construction unit is used for constructing an efficiency prediction model based on reference retention turnover data, and the prediction unit is used for predicting the average retention time of cargoes. Further, the data collection module is configured with a data collection policy comprising: The method comprises th