Search

CN-122022626-A - Building material logistics intelligent management method and system based on multi-source data feedback

CN122022626ACN 122022626 ACN122022626 ACN 122022626ACN-122022626-A

Abstract

The invention discloses a building material logistics intelligent management method and system based on multi-source data feedback, which relate to the technical field of building material logistics management and currently propose the following scheme, wherein the scheme comprises the steps of obtaining construction data of a building material receiving place, wherein the construction data comprises BIM model version data, inspection data and inspection acceptance data; the method comprises the steps of obtaining logistics data of transportation building materials, comprising vehicle real-time data and a transfer warehouse thermodynamic diagram, wherein the vehicle real-time data comprise vehicle position information, load information and oil consumption information, obtaining real-time road data, wherein the road data comprise height and weight information and congestion indexes, analyzing BIM model version data, and calculating progress deviation rate of construction nodes. According to the scheme, BIM model data are analyzed in real time, the progress deviation rate is calculated by combining the inspection result, and the floating interval of the distribution time window is intelligently adjusted so as to adapt to the progress fluctuation of a construction site, and the situation that building materials are accumulated or delayed to be delivered to influence construction due to early delivery is avoided.

Inventors

  • WANG YONGWU
  • LI ZONGPING
  • SUN YATING

Assignees

  • 山东山盟环保工程有限公司

Dates

Publication Date
20260512
Application Date
20260122

Claims (10)

  1. 1. The building material logistics intelligent management method based on multi-source data feedback is characterized by comprising the following steps of: Acquiring construction data of a building material receiving place, wherein the construction data comprises BIM model version data and supervision acceptance data; Obtaining logistics data of transport building materials, wherein the logistics data comprises vehicle real-time data and a transfer warehouse thermodynamic diagram, and the vehicle real-time data comprises vehicle position information, load information and oil consumption information; Acquiring real-time road data, wherein the road data comprises height and weight limiting information and congestion indexes; Analyzing BIM model version data, calculating progress deviation rate of construction nodes, and generating building material distribution time windows with floating intervals; Respectively training a scheduling model based on construction data and logistics data, aggregating model weights through a federal learning server, and outputting time window priority, an optimal path and a vehicle matching scheme; Matching the empty return vehicle with a building rubbish recycling order, and completing the Dutch auction based on price, route and load compatibility three-dimensional grading; and dynamically correcting the weight distribution proportion of the floating interval and the federal model of the time window according to the actual distribution deviation and the construction progress change.
  2. 2. The method of claim 1, wherein the analyzing the BIM model version data, calculating a progress deviation rate of the construction node, and generating a building material distribution time window with a floating section, specifically comprises: analyzing BIM model version data, and extracting actual completion quantity, original planning quantity and Gantt chart time axis of construction nodes; Based on a Gantt chart time axis, acquiring a distribution time window corresponding to the building material; taking the ratio of the actual completion amount to the original planning amount as a progress deviation rate; dynamically adjusting the floating range of the time window according to a preset threshold interval, wherein when the progress deviation rate is greater than the upper limit of the threshold interval, the floating interval of the time window is +/-1 hour, and when the progress deviation rate is within the threshold interval, the floating interval is pressed An hour expansion, when the progress deviation rate is smaller than the lower limit of the threshold interval, the floating interval is set to be +/-0.5 hour, the Is the progress deviation rate.
  3. 3. The method according to claim 2, wherein the training the scheduling model based on the construction data and the logistics data respectively, aggregating model weights by the federal learning server, outputting the time window priority, the optimal path and the vehicle matching scheme, specifically comprises: based on an LSTM network of time sequence prediction, constructing a scheduling model of a constructor, and inputting the progress deviation rate building material stock water level into the scheduling model to obtain the emergency score and the time window priority of each building material class; Constructing a logistics square scheduling model based on a path optimization model of the graph neural network, and inputting logistics data into the scheduling model to obtain an optimal path and vehicle matching scheme; The logistics party transports building materials to a building material receiving place according to the time window priority, the optimal path and the vehicle matching scheme, and the construction party records and updates the supervision acceptance data; Setting a transportation period, extracting prediction acceptance data in BIM model version data, comparing the prediction acceptance data with supervision acceptance data, and if the deviation value of the prediction acceptance data and the supervision acceptance data is greater than 5%, inputting an LSTM layer weight list and a graph neural network side weight list into a federal learning server, outputting global weights, and replacing corresponding weights in a scheduling model; extracting LSTM layer weight list from constructor scheduling model Extracting the edge weight list of the graph neural network from the logistic side scheduling model ; The weight aggregation model of the federal learning server is as follows: ; In the formula, As a result of the global weight being given, And Respectively a construction side weight and a logistics side weight.
  4. 4. The method of claim 3, wherein the path optimization model based on the graph neural network constructs a scheduling model of the logistic party, and inputs logistic data into the scheduling model to obtain an optimal path and vehicle matching scheme, and the method specifically comprises the following steps: Constructing a dynamic directed graph, wherein static nodes of the dynamic directed graph are building material receiving ground coordinates and a transfer warehouse thermodynamic diagram, the dynamic nodes are vehicle position information, and the edges are real-time road data; adopting a space-time diagram convolution network, combining a spatial attention mechanism and time convolution, and outputting path cost prediction and vehicle cargo matching scoring; The log value and the category data are respectively standardized and embedded with codes, and the graph structure is updated every 15 minutes to reflect real-time road condition changes; combining the global weight output by the federal learning server through a multi-objective loss function to jointly train a scheduling model; And searching and selecting an optimal path based on the mixed A-algorithm and the Monte Carlo tree, and solving global optimal matching of the vehicle based on the Hungary algorithm.
  5. 5. The method according to claim 1, wherein matching the construction waste recovery order for the empty return vehicle completes the netherlands auction based on price, route, load compatibility three-dimensional scoring, in particular comprising: Taking vehicle position information, load information and driver preset parameters as idle vehicle information, wherein the driver preset parameters comprise a maximum detour distance and a maximum waiting time; building a construction waste order pool, wherein the construction waste order pool comprises loading point coordinates, loading quantity, loading time and unit quotation; Screening orders according to three pre-screening conditions of space compatibility, load matching and time matching, and grading the orders by using a grading model; Matching empty vehicle information for orders in the building rubbish order pool by adopting a Holland type auction process, wherein the Holland type auction process comprises initial quotation, price reduction bidding and bargaining conditions; The scoring model is: ; In the formula, The order is scored and the order is scored, To offer a price for the current unit, / For the highest/lowest price of the same class of orders in the day, For the path distance of the vehicle location information from the coordinates of the loading point, In order to be able to load the goods, In order to be a value of the payload information, 、 、 Is a weight coefficient.
  6. 6. The method according to claim 1, wherein dynamically correcting the ratio of the floating interval of the time window to the federal model weight distribution according to the actual distribution deviation and the construction progress change comprises: Setting acquisition periods, and acquiring actual arrival time deviation and construction progress variation of a vehicle in each period and carrying out standardized treatment on the actual arrival time deviation and construction progress variation; From the formula Analyzing a construction progress influence coefficient, wherein, In order for the construction progress to influence the coefficient, In order to change the amount of construction progress, Is the original planning quantity; judging the emergency degree according to the construction progress influence coefficient, and correspondingly adjusting the floating interval; The weight of the model construction party and the logistics Fang Quan weight are adjusted according to the actual arrival time deviation of the vehicle, and the latest three deviation data are calculated according to And (5) weighting and accumulating.
  7. 7. The building material logistics intelligent management system based on multi-source data feedback is characterized by being used for realizing the building material logistics intelligent management method based on multi-source data feedback as claimed in any one of claims 1-6, and comprising the following steps: the construction data analysis module is used for analyzing BIM model version data, calculating the progress deviation rate of construction nodes and generating a building material distribution time window with a floating section; The federal scheduling decision module is used for respectively training scheduling models based on construction data and logistics data, aggregating model weights through a federal learning server and outputting time window priority, an optimal path and a vehicle matching scheme; The return auction matching module is used for matching the empty return vehicle with the building rubbish recycling order and completing the Dutch auction based on price, route and load compatibility three-dimensional grading; And the dynamic correction module is used for dynamically correcting the weight distribution proportion of the floating interval and the federal model of the time window according to the actual distribution deviation and the construction progress change.
  8. 8. The intelligent building material logistics management system based on multi-source data feedback of claim 7, wherein the federal scheduling decision module specifically comprises: the LSTM emergency degree prediction unit is used for generating building material emergency degree scores through an LSTM network based on the progress deviation rate and building material stock water levels; The GNN path optimizing unit is used for constructing a dynamic directed graph and outputting an optimal path by combining a space-time graph convolution network; And the federal weight aggregation unit is used for calculating global weights by using a weight aggregation model of the federal learning server and updating weight parameters of the scheduling model.
  9. 9. The intelligent building material logistics management system based on multi-source data feedback of claim 7, wherein the return auction matching module specifically comprises: The order pre-screening unit is used for screening orders according to three pre-screening conditions of space compatibility, load matching and time matching; The three-dimensional grading unit is used for grading the orders by using a grading model; The netherlands auction engine is used for matching empty vehicle information for orders in the building rubbish order pool by adopting the netherlands auction process, wherein the netherlands auction process comprises initial quotation, price reduction bidding and success conditions.
  10. 10. The intelligent building material logistics management system based on multi-source data feedback of claim 7, wherein the dynamic correction module specifically comprises: the deviation analysis unit is used for setting acquisition periods, acquiring actual arrival time deviation and construction progress variation of the vehicle in each period, and carrying out standardized processing on the actual arrival time deviation and construction progress variation; the floating interval adjusting unit is used for judging the emergency degree according to the construction progress influence coefficient and correspondingly adjusting the floating interval; Weight reassigning unit for adjusting model construction side weight and logistics Fang Quan weight according to actual arrival time deviation of vehicle, and for the latest three times of deviation data according to And (5) weighting and accumulating.

Description

Building material logistics intelligent management method and system based on multi-source data feedback Technical Field The invention relates to the technical field of building material logistics management, in particular to an intelligent building material logistics management method and system based on multi-source data feedback. Background In the existing building material logistics management technology, a construction party and a logistics party usually adopt independent data systems and scheduling rules, construction progress management mainly depends on BIM models and manual inspection, logistics scheduling is based on fixed scheduling and static path planning, and part of systems introduce GPS tracking and electronic fence technologies, but construction and logistics data are still in a splitting state and lack of a real-time cooperative mechanism. In addition, the return vehicle scheduling relies on manual matching or fixed partners, and dynamic optimization is not achieved. The traditional building material logistics management has systematic cooperative defects that firstly, building material distribution and on-site actual demand are disjointed due to incapability of timely responding to site progress change caused by dependence on fixed scheduling, secondly, a construction party and logistics party data are mutually independent, a cooperative modeling mechanism is lacked, so that path planning and vehicle scheduling can only be based on local information, the overall cooperative efficiency is low, and finally, a return empty-load vehicle cannot be intelligently matched with building waste recycling requirements, so that the utilization rate of transportation resources is insufficient, and therefore, the intelligent building material logistics management method and system based on multi-source data feedback are provided to solve the problem. Disclosure of Invention In order to solve the technical problems, the technical scheme provides an intelligent building material logistics management method and system based on multi-source data feedback, which solve the problems that the traditional building material logistics management provided in the background technology has systematic cooperative defects, firstly, construction material distribution and on-site actual demand are disjointed due to incapability of timely responding to construction site progress change caused by dependence on fixed scheduling, secondly, constructors and logistics data are mutually independent, a cooperative modeling mechanism is lacking, so that path planning and vehicle scheduling can only be based on local information, the overall cooperative efficiency is low, and finally, a return empty-load vehicle cannot be intelligently matched with building garbage recycling requirements, and the utilization rate of transportation resources is insufficient. In order to achieve the purpose, the technical scheme adopted by the invention is that the invention provides an intelligent building material logistics management method based on multi-source data feedback, which comprises the following steps: Acquiring construction data of a building material receiving place, wherein the construction data comprises BIM model version data and supervision acceptance data; Obtaining logistics data of transport building materials, wherein the logistics data comprises vehicle real-time data and a transfer warehouse thermodynamic diagram, and the vehicle real-time data comprises vehicle position information, load information and oil consumption information; Acquiring real-time road data, wherein the road data comprises height and weight limiting information and congestion indexes; Analyzing BIM model version data, calculating progress deviation rate of construction nodes, and generating building material distribution time windows with floating intervals; Respectively training a scheduling model based on construction data and logistics data, aggregating model weights through a federal learning server, and outputting time window priority, an optimal path and a vehicle matching scheme; Matching the empty return vehicle with a building rubbish recycling order, and completing the Dutch auction based on price, route and load compatibility three-dimensional grading; and dynamically correcting the weight distribution proportion of the floating interval and the federal model of the time window according to the actual distribution deviation and the construction progress change. Preferably, analyzing BIM model version data, calculating progress deviation rate of construction nodes, and generating building material distribution time window with floating section, specifically including: analyzing BIM model version data, and extracting actual completion quantity, original planning quantity and Gantt chart time axis of construction nodes; Based on a Gantt chart time axis, acquiring a distribution time window corresponding to the building material; taking the rati