Research On Stochastic Optimization Method Of Truck Dispatching In Open-pit Coal Mine Based On Intelligent Travelling Time Forecast

With the development of artificial intelligence and unmanned mining technology,the open-pit mine automatic truck dispatching system will usher in new changes.The application of positioning systems and integrated sensors will bring the latest monitoring tools to the truck dispatching system;the full use of deep learning methods,GPS/BDS positioning and navigation technology,Io T cloud and edge computing will change the communication between practitioners and vehicles.On the basis of big data such as GPS/BDS positioning,speed change and attitude of vehicles,theories and methods of data analysis and intelligent processing such as deep learning and statistics can be used to further optimize open-pit mine truck dispatching strategies,reduce transportation costs,and promote intelligence Chemical open-pit mine construction.On the basis of drawing on and improving the predecessor’s open-pit mine dispatching system modeling scheme,this paper constructed an open-pit mine vehicle state data transmission system under the Io T cloud framework.Based on the big data of vehicle travel,we predicted the travel time of vehicle dispatch by forming a method according to a deep neural network.Based on probability statistics and deep learning theory,we modeled and analyzed the data of truck dispatching of short-term plans in an open-pit coal mine,optimized the truck-shovel pre-allocation plan for the next stage,and established an open-air stochastic optimization model of a coal mine truck dispatching system and its effectiveness evaluation method.The main research results obtained in this paper are as follows:(1)Since mine transportation has special round-trip features,deep learning methods can be used to predict the travel time of open-pit mine trucks.For this reason,the transportation path of an open-pit mine obtained from the mining area planning map was used to analyze the truck trip data.The results showed that the probability distribution of the time-consuming activities of different activities such as mining,loading,transportation,and unloading may have two different distributions: normal and lognormal.The travel time is greatly affected by the truck drivers and the loading time.Combining the position and vehicle status data recorded by the differential positioning of the GPS/BDS module,an intelligent edge device that collects mining vehicle status data was developed,which realized the data collection of mining trucks behavior and collected an enhanced data set of truck driving behavior.(2)In order to solve the problem of mining vehicle travel time prediction,the collected vehicle idling speed and other vehicle state data are classified by an improved1D-CNN(one-dimension convolutional neural network)to distinguish truck waiting and driving behaviors,slope driving and horizontal driving behavior.Through extracting the behavior characteristics of truck driving on different road sections,we solved the problem of identifying the features of mining trucks.Two different deep neural network predicting models,based on LSTM(long short-term memory)and CNN,were established.Two-dimensional input vectors were constructed according to the data features of the driver and different road sections.The 2D deep network was used to learn the travel time data and predicted the mining truck travelling time.The results show that the 1D-CNN network with inception structure can achieve a truck status recognition rate of more than 98%;in terms of truck travel time prediction results,if the learning effect of the designed CNN network was in the error tolerance range,the accuracy rate is about 90%.The two models have weaker generalization ability to different transportation paths with large differences.(3)In order to solve the problem of truck dispatch optimization and feasibility assessment in open-pit mines,a stochastic optimization method for truck dispatch in the short-term plan of open-pit mines was established.Since traditional truck dispatching seldom considered the randomness in mine transportation,this article converted the optimization goal based on the minimum cost,and simplified and converted the driving path and other variables in the target into time variables.Comparing to the traditional optimization model,several stochastic variables were introduced to represent randomness such as travel time,and a multi-chromosome genetic algorithm was proposed to solve the stochastic optimization problem of truck shovel scheduling.The results show that the optimization method reduces the idle time of mining equipment,can increase the utilization rate of mining vehicles by about 3%,reduces transportation costs,and the number of heavy trucks uphill by about 7%.In summary,the stochastic optimization model for truck dispatching in open-pit mines designed in this paper can predict and optimize truck dispatching in the next short-term planning stage based on truck travelling time predictions,and verify the feasibility of the predicted results to improve mine transportion performance.The dissertation contains 46 figures,18 tables,and 152 references.