Railway engineering construction material demand fluctuation prediction method based on arrival time limit

Abstract

The invention provides a railway engineering construction material demand fluctuation prediction method based on arrival time limit. The method comprises the following steps: predicting cargo arrival time probability distribution in different paths according to a vehicle driving track, material source distribution, a transportation path and weather and road conditions; a material supply demand prediction model is constructed and solved by considering the transportation arrival time limit, the transportation cost, the guarantee supply requirement, the vehicle guarantee and the storage capacity, and the material supply demand quantity in different time periods is obtained; and deducing the stock of materials based on a Monte Carlo simulation method in combination with the arrival time limit and the material supply demand prediction, and outputting an engineering material ordering strategy. And integrating a material demand volatility prediction method and developing a demand prediction simulation system. According to the method, a material supply demand prediction technology which firstly predicts the cargo transportation time probability distribution and then predicts the construction material supply demand in different time periods is provided, accurate prediction of the engineering material demand is realized, and data support is provided for scientific formulation of a purchase plan and reasonable adjustment of a construction organization.

Description

Technical Field

[0001] The present invention relates to the technical field of material demand forecasting, and in particular to a method for forecasting the volatility of material demand for railway engineering construction based on delivery time limit. Background Art

[0002] Super-large railway construction projects in complex and dangerous areas are often located in complex mountainous areas. Due to the shortage of construction materials along the line, uneven resource distribution, fragile transportation facilities along the line, imperfect road network, frequent geological disasters along the line and serious traffic congestion, the supply of materials in most sections is difficult, and even the normal construction may be affected by the interruption of material supply. The on-site material supply plan is mainly based on the experience of the staff of the section material department, and the weather, road conditions and demand changes are not considered enough. The material supply plan management is relatively extensive, and there is a lack of scientific and accurate supply plan prediction methods. Therefore, it is of great significance to study the spatiotemporal evolution of multiple transportation demands, continuous fluctuations in supply, and emergency reserve guarantees under the influence of multiple factors, break through the volatility prediction technology of material supply demand for super-large railway construction in complex and dangerous areas, and develop a material supply demand prediction simulation system based on this, which is of great significance to accurately grasp the supply and demand of construction materials and scientifically ensure supply.

[0003] Existing related technologies include:

[0004] 1. Probability distribution prediction of delivery time limit

[0005] The existing delivery time prediction methods are mainly based on statistical models or machine learning algorithms, which predict the arrival time through historical data. These methods can often only give a deterministic single-point prediction result, that is, predict a specific arrival time point. They cannot fully consider the impact of factors such as traffic conditions, road conditions, weather effects, and individual differences of drivers on transportation time. They cannot provide the interval distribution of delivery time, cannot effectively describe the uncertainty of the prediction results, and lack comprehensive consideration of the uncertainty of the prediction results.

[0006] 2. Supply and demand forecast

[0007] In terms of material demand forecasting methods, foreign countries started research on material demand forecasting earlier, and generally speaking, they are divided into classical forecasting methods and modern forecasting methods. Classical forecasting methods mainly include time series forecasting methods, regression analysis methods, etc. Modern forecasting methods mainly include neural network forecasting methods, gray system theory, fuzzy forecasting theory, etc. At the same time, domestic scholars mainly focus on research on logistics demand forecasting methods and systems, railway planning and transportation demand plans, etc. The existing results have been applied to urban logistics demand forecasting and emergency material demand forecasting under special circumstances. In terms of engineering practice, in order to ensure the timely and accurate supply of materials, my country's telecommunications industry and petroleum industry have carried out research on the forecasting and management of material demand plans based on big data technology. However, in the railway industry, there is a lack of research on the refinement, accuracy and timeliness of the preparation of material demand plans for engineering construction. Since most of my country's super-large construction projects are located in the eastern and central regions with better social and economic foundations, good processing and manufacturing capabilities, and relatively complete material storage and transportation supporting conditions, most materials can be supplied nearby and in a timely manner. Most of the existing material demand plans are prepared on an annual or monthly basis. For example, the existing Wuhan-Guangzhou High-Speed ​​Railway, Beijing-Shanghai High-Speed ​​Railway and other construction projects all prepare annual and monthly material demand plans based on construction organization plans.

[0008] 3. Implementation of Demand Forecasting Simulation System

[0009] The realization of demand forecasting simulation system can provide a basis for generating plans, determining business policies, and making management decisions. However, traditional forecasting systems are usually developed based on static forecasting models, which are less able to reflect the real-time changes in material demand and difficult to accurately predict supply and demand. At present, the solution to this problem mainly adopts two methods: improving model accuracy and integrating multiple technologies.

[0010] At present, the disadvantages of the existing railway construction material demand volatility prediction methods include:

[0011] (1) Insufficient refinement in material demand forecasting: Existing material demand plans are mostly compiled on an annual or monthly basis. There is no precedent for accurately forecasting railway project material demand plans and conducting theoretical research and practical application of precise project logistics distribution. However, super-large railway construction projects in complex and dangerous areas have the characteristics of long construction cycles, large material supply and demand, shortage of construction materials along the line, and uneven resource distribution. The traffic conditions they face are extremely fragile. Existing construction material demand forecasting methods are difficult to adapt to the complexity of the changing laws of super-large construction materials.

[0012] (2) Material demand forecasts lack consideration for supply security: Since most of my country's super-large construction projects are located in the eastern and central regions with relatively good social and economic foundations, good processing and manufacturing capabilities, and relatively complete material storage and transportation supporting conditions, most materials can be supplied nearby and in a timely manner. On-site material supply plans are mainly issued based on manual experience and lack consideration of factors such as transportation conditions and demand changes. Summary of the invention

[0013] The embodiment of the present invention provides a method for predicting the volatility of railway engineering construction material demand based on the delivery time limit, so as to effectively predict the supply demand of railway engineering construction materials.

[0014] In order to achieve the above object, the present invention adopts the following technical scheme.

[0015] A method for predicting the volatility of railway construction material demand based on delivery time limit, comprising:

[0016] According to the vehicle's driving trajectory, cargo source distribution, transportation route, weather and road conditions, the Gaussian process regression-based prediction method is used to predict the probability distribution of cargo arrival time on different routes.

[0017] Combined with the study of the probability distribution of cargo delivery time, a material supply demand forecasting model for railway engineering construction is constructed by considering factors such as delivery time limit, transportation cost, supply guarantee requirements, vehicle ownership and storage capacity, and the material supply demand in different time periods is predicted;

[0018] Based on the forecast of material delivery time and material supply demand, combined with the actual project progress and material consumption, the Monte Carlo simulation algorithm is used to deduce the time series changes of material inventory, and the material ordering strategy, that is, the material ordering quantity and the material plan release time, is obtained;

[0019] A demand forecasting simulation system is developed by integrating the probability distribution prediction method of cargo arrival time, the material supply demand prediction method and the material ordering strategy prediction method with geographic information data processing. The demand forecasting simulation system uses a railway engineering construction material demand volatility prediction method based on arrival time limit to predict the supply demand of various materials in different sections and work sites and the time to issue the supply plan.

[0020] Preferably, the prediction method based on Gaussian process regression according to the vehicle driving trajectory, cargo source distribution, transportation route and weather and road conditions predicts the probability distribution of cargo arrival time on different routes, including:

[0021] Comprehensively consider road conditions, transportation environment, transportation vehicles and date characteristics to obtain historical cargo transportation trajectory data and historical weather data, and use Gaussian process regression algorithm based on the historical cargo transportation trajectory data and historical weather data to build a cargo transportation travel time prediction model by mining vehicle transportation trajectories;

[0022] Before a cargo transportation starts, the route, transportation start date and transportation vehicle information of the cargo transportation process are obtained according to the cargo transportation plan. The route, transportation start date and transportation vehicle information are processed and input into the cargo transportation travel time prediction model. The cargo transportation travel time prediction model outputs the predicted value of the arrival time of the cargo transportation process.

[0023] Preferably, the processing of the route, transportation start date and transportation vehicle information and inputting them into the cargo transportation travel time prediction model, the cargo transportation travel time prediction model outputting the predicted value of the arrival time during the cargo transportation process, includes:

[0024] The route, transportation start date and transportation vehicle raw data are processed to obtain the characteristic values ​​related to road conditions, transportation environment, vehicle type and date. The processing process of the raw data is as follows:

[0025] Transportation environment: If you want to obtain weather information, the required information includes date and location. The time is determined by the data returned by the vehicle track query interface. For the location, the longitude and latitude returned by the vehicle track query interface are matched to the specific location through the Baidu map coordinate picking system to obtain the weather information for the specific time and location;

[0026] Date feature and transport vehicle: Date feature refers to whether it is a holiday. The time in the trajectory data can be used to determine whether the transport process is on a holiday. The type of transport vehicle can be obtained by reading the trajectory data.

[0027] Road conditions: Obtain the road conditions corresponding to the trajectory by matching the read trajectory data to the specific road section;

[0028] The cargo transportation travel time prediction model finds a potential function to express the relationship between the feature data and the target vector by learning the data in the training set, which will affect the feature data x of the cargo transportation travel time. i Defined as:

[0029] x i =(d i ,t i ,q i ,c i )

[0030] Where, d i =(d 1i ,d 2i ,…,d 6i ) is the road condition feature vector; t i is the date feature vector; q iis the meteorological information feature vector, c i is the vehicle type feature vector;

[0031] The travel time of the cargo on each road is defined as y i , the cargo transportation travel time prediction model is expressed by the following formula:

[0032] f(x)~GP(m(x),k(x,x′))

[0033] Where m(x) is the mean function, assumed to be a constant, and k(x,x′) is the covariance function (or kernel function) used to measure the similarity between the input data x and x′;

[0034] The training process of the freight transport travel time prediction model is defined by the following formula:

[0035] f=[f(x 1 ),f(x 2 ),……,f(x n )] T

[0036] Where f is the predicted value vector of all training data points, n is the number of training data, and assuming that the observed value is y = [y 1 ,y 2 ,……,y n ] T ;

[0037] The prediction process of the freight transportation travel time prediction model is expressed as follows:

[0038] p(y * ∣x * ,X,y)=N(μ * ,σ *2 )

[0039] where y * is the output value of the new data point to be predicted, x * The features of the new data point to be predicted, X is the features of the training data, y is the output of the training data, μ * is the mean of the predicted values, σ *2 is the variance of the predicted value, calculated as follows:

[0040]

[0041] where k * is the covariance vector of the predicted data point and all training data points, and K is the covariance matrix between the training data points.

[0042] Preferably, the material supply demand forecasting model for railway engineering construction is constructed by combining the research on the probability distribution of cargo delivery time, taking into account delivery time limit, transportation cost, supply guarantee requirements, vehicle ownership and storage capacity factors, and forecasting the material supply demand in different time periods, including:

[0043] Based on the above-mentioned study on the probability distribution of cargo arrival time, taking into account the arrival time limit, transportation costs, supply requirements, vehicle ownership and storage capacity factors, and taking cost minimization as the goal, a material supply demand forecasting model for railway engineering construction is constructed under the framework of storage theory. The constraints of the material supply demand forecasting model are set based on the time series balance of material inventory, material supply, storage capacity, vehicle ownership and actual material consumption factors.

[0044] The actual on-site construction progress of the railway project and the actual material consumption, delivery time limit, transportation costs, supply guarantee requirements, vehicle ownership and storage capacity data are processed and input into the material supply demand forecasting model. The heuristic algorithm is used to solve the material supply demand forecasting model, and the optimal material arrival quantity for each day in the forecast period is calculated to obtain the construction material supply demand in different time periods.

[0045] Preferably, the material supply demand forecasting model is based on the actual construction organization plan and actual material consumption at the project site, with inventory capacity, supply guarantee and transportation capacity as constraints, and the minimum average cost as the objective function. The meanings of relevant symbols in the material demand forecasting model are shown in Table 1;

[0046] Table 1 Symbolic definitions of the material supply demand forecasting model

[0047]

[0048]

[0049] The objective function of the material supply demand forecasting model is composed of the following formula:

[0050] Taking the minimum average ordering cost, storage cost and freight cost in one cycle as the goal, the objective function is constructed as follows:

[0051]

[0052] The constraints of the material demand forecasting model are composed of the following formula:

[0053] Material inventory constraint: indicates that the material inventory on day t is equal to the material inventory on day t-1 plus the material arrival on day t minus the material consumption on day t;

[0054] S(t)=S(t-1)+A(t)-C(t)

[0055] Inventory capacity constraint: indicates that the inventory of materials on day t should meet the material demand for the next Δt days and be less than the inventory capacity;

[0056] D(t+Δt)≤S(t)≤N

[0057] Material arrival time constraint: means that the material arrival quantity on day t is equal to the order quantity before the delivery time σ days;

[0058] A(t)=O(t-σ)

[0059] Material arrival quantity constraint: The material arrival quantity on day t is equal to the number of vehicles that arrive on day t multiplied by the transport capacity of a single vehicle;

[0060] A(t)=M*E(t)

[0061] Material delivery time probability distribution constraint: indicates that the material delivery time σ follows the probability distribution P(σ);

[0062] σ~P(σ)

[0063] Material consumption constraint: indicates that the material consumption on day t is determined by the actual project progress and the actual quota consumption;

[0064] C(t)=f(AP,AC)

[0065] Material demand constraint: indicates that the material demand in the next Δt days is determined by the planned project progress and the estimated quota consumption;

[0066] D(t+Δt)=g(PP,PC)

[0067] The solution process of the heuristic algorithm includes:

[0068] Step 1: Initialize the population

[0069] Coding rules: Assume that the number of vehicles that arrive on day t is A(t), and the number of vehicles that arrive every day in the forecast period T is composed of an array A=[A(1), A(2), …, A(T)], and convert the decimal array into a binary array, A′=[A′(1), A′(2), …, A′(T)];

[0070] Set the evolution generation counter n = 0, set the maximum evolution generation N, and randomly generate M individuals as the initial population P(0);

[0071] Step 2: Individual Assessment

[0072] Calculate the fitness F[P(m)] of each individual in the population P(m), substitute the number of vehicles that deliver goods represented by different individuals into the objective function, calculate the average cost, and based on this, derive the fitness of the individual:

[0073] F[P(m)]=1 / K[P(m)]

[0074] Step 3: Select the operation

[0075] Apply the selection operator to the population, and directly pass the optimized individuals to the next generation or generate new individuals through pairing and crossover and then pass them to the next generation. The selection operation is based on the fitness evaluation of individuals in the population. Based on the calculation results of the fitness of each individual in the population, each individual is given a probability of being selected, and the next generation of basic population is formed through multiple sampling.

[0076] Step 4: Crossover

[0077] Apply the crossover operator to the population. The so-called crossover refers to the operation of replacing and recombining part of the structure of two parent individuals to generate new individuals. With a certain probability p, some parent individuals are selected from the next generation basic population for crossover, and individuals are reconstructed by cutting off from a certain position.

[0078] Step 5: Mutation operation

[0079] Applying mutation operators to the population means changing the gene values ​​at certain loci of the individual strings in the population. Using binary numbers to construct individuals, the values ​​at a certain locus in the individual are transformed from 0 to 1 with a certain probability p. After the population P(n) undergoes selection, crossover, and mutation operations, the next generation population P(n+1) is obtained.

[0080] Step 6: Termination condition judgment

[0081] If n=N, the individual with the maximum fitness obtained in the evolution process is output as the optimal solution, the calculation is terminated, the daily arrival volume of materials within the prediction period is output, and the material supply demand in different time periods is obtained.

[0082] Preferably, the material ordering strategy, i.e., the material ordering quantity and the material plan release time, is obtained by using a Monte Carlo simulation algorithm to deduce the time series change of material inventory based on the material delivery time and material supply demand forecast, combined with the actual project progress and material consumption, including:

[0083] The material inventory simulation process based on Monte Carlo simulation is as follows:

[0084] Input: number of simulations M; simulation duration T; initial material inventory D(0); material supply days requirement Δt; material delivery time σ probability distribution P(σ); material storage capacity N; actual project progress AP; actual quota consumption AC; planned project progress PP, estimated quota consumption PC;

[0085] process:

[0086] Step 1: Start the simulation of material demand volatility forecasting for the i-th time;

[0087] Step 2: Deducing the material inventory S(t) on day t:

[0088] Step 2.1: Calculate the material consumption C(t) on day t based on the project progress AP and the actual quota consumption AC;

[0089] Step 2.2: Obtain the material delivery volume A(t) on the tth day based on the output of the material supply demand forecasting model;

[0090] Step 2.3: Calculate the material inventory S(t) on day t based on the material inventory S(t-1) on day t-1, the material consumption C(t) on day t, and the material delivery volume A(t) on day t;

[0091] Step 3: Calculate the material demand D(t+Δt) within the next Δt days based on the planned project progress PP and the estimated quota consumption PC;

[0092] Step 4: Sample from the delivery time probability distribution P(σ) to determine the delivery time σ of the materials when scheduled on the tth day t ;

[0093] Step 5: Determine the order time t-σ for materials arriving on day t t ;

[0094] Step 6: If t=T, end the i-th simulation; otherwise, t=t+1, return to step 2;

[0095] Step 7: If i=M, end the simulation; otherwise, i=i+1, return to step 1;

[0096] Output: Material ordering strategy, which includes the material ordering quantity in different time periods and the planned release time of materials.

[0097] Preferably, the integrated cargo delivery time probability distribution prediction method, material supply demand prediction method and material ordering strategy prediction method are combined with geographic information data processing to develop a demand prediction simulation system, which uses a railway engineering construction material demand volatility prediction method based on delivery time limit to predict the supply demand of various materials at different bidding sections and work sites and the time of issuing supply plans, including:

[0098] The prediction process of the probability distribution of the cargo arrival time, the prediction process of the construction material supply demand in different time periods and the material inventory deduction process based on Monte Carlo simulation are integrated, and a demand prediction simulation system is developed in combination with geographic information data processing. The demand prediction simulation system uses a prediction method for the volatility of railway engineering construction material demand based on the arrival time limit, taking into account multiple factors such as the actual demand for engineering materials, the material transportation network and route selection, the probability distribution of the arrival time limit of different types of materials and the material storage capacity, to achieve the prediction of the supply demand of various materials in different sections and work sites and the time of issuing the supply plan;

[0099] The calculation process of the demand forecasting simulation system includes:

[0100] Step 1: Basic parameter settings, including selecting the forecast initial time, forecast days, supply guarantee days, bidding section, warehouse and supply source, and obtaining information on environmental parameters, warehouse storage capacity, storage cost and actual material consumption;

[0101] Step 2: Calculate the probability distribution of delivery time;

[0102] Step 2.1: Obtain weather conditions, road conditions, transportation environment and vehicle type parameters for each road section;

[0103] Step 2.2: Load the delivery time probability distribution prediction model, pass the route section information, including the section mileage, road grade and road width average speed, into the time probability distribution prediction model, and calculate the volatility delivery time probability distribution prediction result;

[0104] Step 3: Calculation of material supply demand forecast;

[0105] Step 3.1: Combined with the calculation results of the delivery time probability distribution, the input parameters include the forecast period, actual consumption, delivery time, number of vehicles, single vehicle transportation, capacity freight rate, storage cost, ordering cost, initial inventory, inventory capacity, transportation distance and supply guarantee days, as well as basic parameters including population size and mutation probability;

[0106] Step 3.2: Load the material supply demand forecasting model, input the forecasting period, actual consumption, delivery time, number of vehicles, single vehicle transportation, capacity freight rate, storage cost, ordering cost, initial inventory, inventory capacity, transportation distance and supply guarantee days into the material demand forecasting model, and calculate the material supply demand forecasting results for different time periods;

[0107] Step 4: Calculation of material ordering strategy;

[0108] Step 4.1: Combined with the probability distribution of delivery time limit and the calculation results of material supply demand forecast, the input parameters include simulation time, initial material inventory, material supply guarantee days, material delivery time, material storage capacity, actual project progress and actual material consumption;

[0109] Step 4.2: Load the material inventory simulation algorithm based on Monte Carlo simulation and output the corresponding material ordering strategy;

[0110] Step 5: Display the demand forecast results and output the material supply demand and material ordering strategy, that is, output the supply demand of various materials in different sections and work sites and the time when the supply plan is issued.

[0111] It can be seen from the technical solutions provided by the above-mentioned embodiments of the present invention that the method of the present invention proposes a material supply and demand forecasting technology that first predicts the probability distribution of cargo arrival time along different routes and then predicts the supply and demand of construction materials in different time periods. A material supply and demand forecasting model for super-large railway engineering construction is constructed under the framework of storage theory to achieve accurate forecasting of engineering material demand, thereby providing data support for scientifically formulating procurement plans and reasonably adjusting construction organizations.

[0112] Additional aspects and advantages of the present invention will be given in part in the following description, which will become obvious from the following description, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0113] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0114] Figure 1 A schematic diagram of a method for predicting the volatility of railway construction material demand based on delivery time provided in an embodiment of the present invention;

[0115] Figure 2 A schematic diagram of a method for predicting freight transport travel time provided by an embodiment of the present invention;

[0116] Figure 3 A schematic diagram of a data processing portion of a cargo travel time prediction model provided by an embodiment of the present invention;

[0117] Figure 4 A schematic diagram of a construction material supply demand forecasting method by time period provided in an embodiment of the present invention;

[0118] Figure 5A schematic diagram of an implementation scheme of a demand forecasting simulation system integrating a geographic information system and a demand forecasting model provided by an embodiment of the present invention;

[0119] Figure 6 A schematic diagram of an implementation scheme for constructing a material demand database for a demand forecasting simulation system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0120] The embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0121] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or coupling. The term "and / or" used herein includes any unit and all combinations of one or more associated listed items.

[0122] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with the meanings in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless defined as herein.

[0123] To facilitate understanding of the embodiments of the present invention, several specific embodiments will be further explained below with reference to the accompanying drawings, and each embodiment does not constitute a limitation on the embodiments of the present invention.

[0124] The embodiment of the present invention combines the characteristics of on-site material supply and proposes a method for predicting the volatility of railway construction material demand based on the delivery time limit, including the probability distribution prediction of material delivery time limit, material supply demand prediction, material inventory simulation and demand prediction simulation system development. This method combines the study of the probability distribution of freight transportation travel time, constructs a material supply demand prediction method based on Monte Carlo simulation, and realizes the refined prediction of material demand.

[0125] The embodiment of the present invention proposes a method for predicting the supply and demand of railway engineering materials. It combines the prediction results of the probability distribution of the arrival time of goods on different routes, considers factors such as the arrival time limit, transportation costs, supply requirements, vehicle ownership, and storage capacity, builds a corresponding material demand prediction model and solves it, and performs material inventory deduction based on Monte Carlo simulation, thereby realizing an accurate prediction of the demand for engineering material supply. By refining the road network, accurately matching truck trajectories, and extracting features such as the arrival time limit, transportation costs, supply requirements, and storage capacity, and then combining the storage theory framework to build a material supply and demand prediction model, this technical solution can comprehensively consider various influencing factors, accurately predict the demand for various categories of materials, and realize the scientific supply of materials on the railway engineering site.

[0126] Through field research, the present invention found that the main contradiction in the demand for materials on the construction site of super-large railway projects is supply guarantee, and the current on-site bidding section lacks an accurate material supply calculation method. The present invention is based on solving practical problems in the construction process of super-large railway projects in complex and dangerous areas, with the goal of finely predicting the supply demand of construction materials, and carries out research according to the technical route of "freight transportation travel time prediction - construction material supply demand prediction by time period - material inventory simulation deduction - system platform development".

[0127] The implementation principle diagram of a method for predicting the volatility of railway construction material demand based on the delivery time limit proposed in an embodiment of the present invention is as follows: Figure 1 As shown, the processing steps include the following:

[0128] Step S10, predicting the probability distribution of cargo delivery time on different routes. Inputting vehicle driving trajectory, cargo source distribution, transportation route, weather and road conditions, and using the prediction method based on Gaussian process regression, predicting the probability distribution of cargo delivery time on different routes.

[0129] Step S20, predict the supply demand of construction materials in different time periods. Combined with the study of the probability distribution of cargo delivery time, based on the construction progress of the project site and the actual material consumption and demand for construction materials, taking into account factors such as delivery time limit, transportation cost, supply guarantee requirements, vehicle ownership and storage capacity, a corresponding material demand prediction model is constructed, and a heuristic algorithm is designed to solve it and obtain the material supply demand.

[0130] Step S30, deduction of material inventory time series changes. Combining the results of the cargo delivery time and material supply demand research, combined with the actual project progress and material consumption, the Monte Carlo simulation method is used to deduce the material inventory, and the corresponding engineering material ordering strategy is obtained, that is, the order quantity of various materials in different sections and work sites within the forecast period and the planned release time of materials.

[0131] Step S40, material demand forecasting simulation system development. The demand forecasting simulation system is developed by integrating the cargo arrival time probability distribution forecasting method, material supply demand forecasting method and material ordering strategy forecasting method with geographic information data processing. The demand forecasting simulation system uses a map as a carrier to display the system business data content in a visual way. Based on the railway engineering construction material demand volatility forecasting method, the demand for various materials in different sections and work sites and the time of issuing supply plans are predicted.

[0132] Specifically, the above step S10 includes: a method for predicting the travel time of freight transportation based on Gaussian process regression: by mining vehicle transportation trajectories and weather data, a freight transportation travel time prediction model is constructed based on the Gaussian process regression algorithm. The model comprehensively considers road conditions, transportation environment, transportation vehicles and date characteristics to improve the prediction performance of the model. The implementation principle of the method for predicting the travel time of freight transportation provided by the embodiment of the present invention is as follows: Figure 2 As shown, from Figure 2 It can be seen that there are mainly the following steps to build a travel time prediction model:

[0133] 1: Original data acquisition: The model comprehensively considers road conditions, transportation environment, transportation vehicles and date characteristics, so the original data required include historical cargo transportation trajectory data and historical weather data.

[0134] 2: Data processing: This part mainly uses data mining technology to obtain feature values ​​through a series of transformations on the original data, which are used as feature space data for training model parameters.

[0135] 3: Model construction: The present invention constructs a cargo transportation travel time prediction model based on the Gaussian process regression algorithm. The basic idea of ​​the Gaussian process regression prediction method is: by assuming the prior probability that the learning sample obeys the Gaussian process, and then combining the Bayesian theory to obtain the corresponding posterior probability, the maximum likelihood method is used to obtain the corresponding optimal hyperparameters.

[0136] 4: Model application: Apply the constructed model to the actual cargo transportation process. Before a cargo transportation starts, the information about the transportation route, transportation start date and transportation vehicle can be obtained according to the cargo transportation plan. The known information can be processed and input into the travel time prediction model to obtain the predicted value of the transportation time for this transportation process.

[0137] The specific steps are:

[0138] (1) Data processing

[0139] Figure 3 This is a schematic diagram of the data processing part of a freight travel time prediction model provided by an embodiment of the present invention. In order to obtain the training set data for training model parameters, the original data needs to be processed to obtain the characteristic values ​​related to road conditions, transportation environment, vehicle type and date. The processing process of the original data is as follows:

[0140] ① Transportation environment: If you want to obtain weather information, the required information includes date and location. The time can be determined from the data returned by the vehicle track query interface. For the location, first match the specific location through the Baidu map coordinate picking system according to the longitude and latitude returned by the vehicle track query interface. Combine the information of the two, first obtain the website of the China Meteorological Administration, and then parse the webpage to obtain the weather information of the specific time and location, including whether there is rain or snow and visibility.

[0141] ② Date features and transport vehicles: Date features refer to whether it is a holiday. The time in the trajectory data can be used to determine whether the transportation process is on a holiday. The type of transport vehicle is also obtained by reading the trajectory data.

[0142] ③Road conditions: Road conditions are mainly obtained by matching the read trajectory data to a specific road section to obtain the road conditions corresponding to the trajectory. Taking a certain road as an example, the specific process of matching the trajectory data to the road is to first determine the longitude and latitude coordinates of the starting point of the road: starting point

[0143] [97.0967068,30.6611156], end point [96.5987455,30.87088213], where the horizontal coordinate is longitude (lon) and the vertical coordinate is latitude (lat); secondly, the distances between different trajectory points and the starting point in a single transportation process are calculated and the smallest distance is retained (as shown in the table below, the closest distances between the trajectory point in the first transportation trajectory and the starting point of this road are 587982 and 622889 respectively, and the third is 21.035134 and 55.022815 respectively), and the trajectory points A and B in a single transportation process within 200m from the starting point of the road are retained (the third transportation trajectory is retained). According to the time recorded at points A and B, the transportation time of this transportation trajectory on this road can be obtained, and the road condition data including road width, number of lanes, design speed of the road, and line grade can be obtained from the road name. After each trajectory matching operation is completed, if there is a transportation trajectory on the matched road, the average speed of the transportation process (speed = distance / time) will be recorded at a specific location in the database. After all trajectory matching operations are completed, the arithmetic average method is used to calculate the historical average speed of all vehicles passing through each road.

[0144] Table 2 Trajectory data matching road process table

[0145]

[0146]

[0147] (2) Model construction

[0148] Gaussian Process Regression (GP) models the Gaussian distribution of a function and uses known data points to estimate the parameters of the distribution, thereby achieving prediction of new data points and providing uncertainty estimates for the prediction results.

[0149] This model finds a potential function that can express the relationship between feature data and target vector by learning the data in the training set, and the feature data x that affects the travel time of cargo transportation is i Defined as:

[0150] x i =(d i ,t i ,q i ,c i )

[0151] Where, d i =(d 1i ,d 2i ,…,d6i ) is the road condition feature vector; t i is the date feature vector; q i is the meteorological information feature vector, c i is the vehicle type feature vector.

[0152] The target vector, i.e. the travel time of the cargo on each road segment, is defined as y i .

[0153] Therefore, the model of the present invention can be expressed by the following formula:

[0154] f(x)~GP(m(x),k(x,x′))

[0155] Where m(x) is the mean function, assumed to be a constant, and k(x,x′) is the covariance function (or kernel function) used to measure the similarity between the input data x and x′. The training process of the model can be defined as follows:

[0156] f=[f(x 1 ),f(x 2 ),……,f(x n )] T

[0157] Where f is the predicted value vector of all training data points, n is the number of training data, and assuming that the observed value is y = [y 1 ,y 2 ,……,y n ] T The prediction process of the model can be expressed as follows:

[0158] p(y * ∣x * ,X,y)=N(μ * ,σ *2 )

[0159] where y * is the output value of the new data point to be predicted, x * The features of the new data point to be predicted, X is the features of the training data, y is the output of the training data, μ * is the mean of the predicted values, σ *2 is the variance of the predicted value, calculated as follows:

[0160]

[0161] where k * is the covariance vector of the predicted data point and all training data points, and K is the covariance matrix between the training data points.

[0162] Specifically, the above step S20 includes: Time-divided construction material supply and demand forecasting technology: constructing a railway engineering construction material supply and demand forecasting model under the framework of storage theory, forecasting the engineering material demand based on the actual construction organization plan and the actual quota consumption of materials at the engineering site, and taking inventory capacity, supply guarantee, transportation capacity, etc. as constraints, and taking the minimum average cost as the objective function. A solution algorithm for the engineering construction material supply and demand forecasting model based on a heuristic algorithm is designed to obtain the material supply demand in different time periods. The implementation principle diagram of a time-divided construction material supply and demand forecasting method provided by an embodiment of the present invention is shown in FIG. Figure 4 As shown. Figure 4 It can be seen that the construction of a time-divided material supply demand forecasting method mainly consists of the following steps:

[0163] 1: Model construction: A railway engineering construction material supply demand forecasting model is constructed under the framework of storage theory, based on the actual construction organization plan and actual material demand at the project site, with inventory capacity, supply guarantee, transportation capacity, etc. as constraints, and the minimum average cost as the objective function.

[0164] 2: Algorithm design: According to the material supply demand forecasting model, an algorithm for solving the engineering construction material supply demand forecasting model based on a heuristic algorithm is designed. The material supply demand in different time periods is obtained through the algorithm solution.

[0165] The specific steps are:

[0166] (1) Construction of construction material supply demand forecasting model by time period

[0167] The meanings of relevant symbols in the definition model are shown in Table 3.

[0168] Table 3 Symbol definitions of construction material supply demand forecasting model

[0169]

[0170] Taking the minimum average ordering cost, storage cost, and freight cost within a period as the goal, the objective function is constructed as follows:

[0171]

[0172] The model constraints consist of the following:

[0173] Material inventory constraint. It means that the material inventory on day t is equal to the material inventory on day t-1 plus the material arrival on day t minus the material consumption on day t.

[0174] S(t)=S(t-1)+t(t)-C(t)

[0175] Inventory capacity constraint. It means that the inventory of materials on day t should meet the material demand for the next Δt days and be less than the inventory capacity.

[0176] D(t+Δt)≤S(t)≤N

[0177] Material arrival time constraint: It means that the material arrival quantity on day t is equal to the order quantity before the delivery time σ days.

[0178] A(t)=O(t-σ)

[0179] Material arrival quantity constraint. It means that the material arrival quantity on day t is equal to the number of vehicles that arrive on day t multiplied by the transport capacity of a single vehicle.

[0180] A(t)=M*E(t)

[0181] The material delivery time probability distribution constraint. It means that the material delivery time σ follows the probability distribution P(σ).

[0182] σ~P(σ)

[0183] Material consumption constraint. Indicates that the material consumption on day t is determined by the actual project progress and the actual quota consumption.

[0184] C(t)=f(AP,AC)

[0185] Material demand constraint. It means that the material demand in the next Δt days is determined by the planned project progress and the estimated quota consumption.

[0186] D(t+Δt)=g(PP,PC)

[0187] (2) Heuristic algorithm design

[0188] The model is solved using a heuristic algorithm. The algorithm flow is as follows:

[0189] Step 1: Initialize the population

[0190] Coding rules: Assume that the number of vehicles that arrive on day t is A(t), and the number of vehicles that arrive every day in the forecast period T is composed of an array A = [A(1), A(2), …, A(T)]. To improve the calculation efficiency of the algorithm, the decimal array is converted into a binary array, A′ = [A′(1), A′(2), …, A′(T)].

[0191] Set the evolution generation counter n=0, set the maximum evolution generation N, and randomly generate M individuals as the initial population P(0).

[0192] Step 2: Individual Assessment

[0193] Calculate the fitness F[P(m)] of each individual in the population P(m). Substitute the number of vehicles that have arrived with materials represented by different individuals into the objective function, calculate the average cost, and based on this, derive the fitness of the individual:

[0194] F[P(m)]=1 / K[P(m)]

[0195] Step 3: Select the operation

[0196] The selection operator is applied to the population. The purpose of selection is to directly pass on the optimized individuals to the next generation or to generate new individuals through pairing and crossover and then pass on to the next generation. The selection operation is based on the fitness evaluation of individuals in the population. Based on the calculation results of the fitness of each individual in the population, each individual is given a probability of being selected, and the next generation of basic population is formed through multiple sampling.

[0197] Step 4: Crossover

[0198] Apply the crossover operator to the population. Crossover refers to the operation of replacing and recombining parts of the structures of two parent individuals to generate new individuals. Through crossover, the search ability of the genetic algorithm is greatly improved. With a certain probability p, some parent individuals are selected from the next generation basic population for crossover, and individuals are reconstructed by cutting them off at a certain position.

[0199] Step 5: Mutation operation

[0200] Apply the mutation operator to the population. That is, change the gene values ​​at certain loci of the individual strings in the population. Since this algorithm uses binary numbers to construct individuals, the value at a certain locus in the individual is transformed from 0 to 1 with a certain probability p. After the population P(n) is selected, crossed, and mutated, the next generation population P(n+1) is obtained.

[0201] Step 6: Termination condition judgment

[0202] If n=N, the individual with the maximum fitness obtained in the evolution process is taken as the optimal solution output, the calculation is terminated, the optimal material arrival quantity for each day in the prediction period is output, and the material supply demand in different time periods is obtained.

[0203] Specifically, the above step S30 includes: simulation and deduction of material inventory time series changes: based on the material transportation travel time and material supply demand forecast research, combined with the actual project progress and material consumption, the Monte Carlo simulation algorithm is used to deduce the time series changes of material inventory, and the corresponding engineering material ordering strategy is obtained, that is, the material ordering quantity and the material plan release time;

[0204] The material inventory deduction process based on Monte Carlo simulation mainly includes the following steps:

[0205] 1: Simulation: Combined with the cargo arrival time and material supply demand forecast, the material inventory is simulated based on the Monte Carlo simulation method to obtain the corresponding engineering material ordering strategy.

[0206] 2: Technical application: Apply the constructed material supply and demand forecasting method to the actual material ordering, and form the corresponding material supply and demand OD pairs based on the actual material demand points and material supply point distribution on site. Combined with the study on the probability distribution of the arrival time of the goods, the arrival time forecast value of each supply and demand OD pair can be obtained. Input the initial date, forecast time, supply guarantee days, initial inventory and other parameters as well as the probability forecast result of the arrival time limit into the material supply and demand forecasting model to obtain the material supply demand in different time periods. Further, through Monte Carlo simulation deduction, the material inventory, arrival and ordering situation within the deduction period can be obtained.

[0207] The specific steps are as follows:

[0208] Input: number of simulations M; simulation duration T; initial material inventory D(0); material supply days requirement Δt; material arrival time σ probability distribution P(σ); material storage capacity N; actual project progress AP; actual quota consumption AC; planned project progress PP, estimated quota consumption PC.

[0209] process:

[0210] Step 1: Start the simulation of material demand volatility forecasting for the i-th time;

[0211] Step 2: Deducing the material inventory S(t) on day t:

[0212] Step 2.1: Calculate the material consumption C(t) on day t based on the project progress AP and the actual quota consumption AC;

[0213] Step 2.2: Obtain the material delivery volume A(t) on the tth day based on the output of the material supply demand forecasting model;

[0214] Step 2.3: Calculate the material inventory S(t) on day t based on the material inventory S(t-1) on day t-1, the material consumption C(t) on day t, and the material delivery volume A(t) on day t;

[0215] Step 3: Calculate the material demand D(t+Δt) within the next Δt days based on the planned project progress PP and the estimated quota consumption PC.

[0216] Step 4: Sample from the delivery time probability distribution P(σ) to determine the delivery time σ of the materials when scheduled on the tth day t ;

[0217] Step 5: Determine the order time t-σ for materials arriving on day t t ;

[0218] Step 6: If t=T, end the i-th simulation; otherwise, t=t+1, return to step 2;

[0219] Step 7: If i=M, end the simulation; otherwise, i=i+1, return to step 1.

[0220] Output: Material sequence order quantity range.

[0221] Specifically, the above step S40 includes: an implementation scheme of a demand forecasting simulation system integrating geographic information data and the method for predicting the volatility of demand for railway engineering construction materials based on the delivery time limit provided by the embodiment of the present invention is as follows: Figure 5 shown.

[0222] The system is based on a method for predicting the volatility of railway construction material demand based on the delivery time limit. It considers factors such as the actual demand for construction materials, material transportation network and route selection, the probability distribution of delivery time limits for different types of materials, and material storage capacity, and predicts the supply demand for various materials at different bidding sections and work sites, as well as the time when supply plans will be issued. The system reconstructs material demand data in a digital way, builds a multi-dimensional database of material demand, and designs functional modules such as map display, material demand forecasting, and material supply early warning, to provide auxiliary decision-making support for on-site understanding of the material supply situation of each bidding section and the issuance of supply plans by each construction unit. The specific implementation steps are:

[0223] (1) Data processing

[0224] The material demand data is reconstructed digitally to realize the standardization, systematization, visualization and display of material demand data. The data contained in the material demand database is mainly divided into multiple dimensions such as geographic data, business data, and calculation data. Figure 6 As shown. Among them, geographic data includes basic geographic data such as provinces, cities, counties, and water systems, and professional geographic data such as railway networks and highway network construction roads. It is a display carrier for business and computing data, and is also the entrance for system spatial query and user interaction; business data includes entity data such as bidding sections, work sites, work surfaces, warehouses, contracts, and sources of goods. It is the physical carrier for realizing functions such as demand forecasting and material supply guarantee; computing data includes vehicle trajectories, weather conditions, road information, material reporting requirements, forecasting parameters, etc. It is the parameter data required for model calculation, and the calculation results can be visualized. The data processing process includes:

[0225] Step 1: Establish a geographic information grid, divide the space into multiple grids by city, obtain real-time weather conditions in the grid, and establish the topological structure of the public network and railway network in the study area to calculate the optimal path between the source of goods and the warehouse;

[0226] Step 2: Digitally reconstruct the material demand data to achieve standardized, systematized, and visualized operation and display of the material demand data, which is used to calculate the probability distribution of delivery time and material supply demand.

[0227] (2) Prediction method integration

[0228] The method for predicting the probability distribution of cargo arrival time, the method for predicting the supply and demand of construction materials in different time periods, and the method for deducing the material inventory based on Monte Carlo simulation are integrated, and a demand prediction simulation system is developed in combination with geographic information data processing. The demand prediction simulation system uses a method for predicting the volatility of demand for railway construction materials based on arrival time limits, taking into account factors such as the actual demand for construction materials, material transportation network and route selection, probability distribution of arrival time limits for different types of materials, and material storage capacity, to achieve predictions of the supply demand for various materials at different bidding sections and work sites and the time when supply plans are issued. The system calculation process includes:

[0229] Step 1: Basic parameter settings, including selecting the initial forecast time, forecast days, supply guarantee days, bidding section, warehouse, source of goods, etc., and obtaining information such as environmental parameters, warehouse storage capacity, storage cost, actual material consumption, etc.

[0230] Step 2: Calculate the probability distribution of delivery time;

[0231] Step 2.1: Obtain parameters such as weather conditions, road conditions, transportation environment, and vehicle type for each road section.

[0232] Step 2.2: Load the transport time probability distribution prediction model, pass the route section information, including section mileage, road grade, road width and average speed, into the model, and calculate the volatility time probability distribution prediction result.

[0233] Step 3: Calculation of material supply demand forecast;

[0234] Step 3.1: Combined with the calculation results of the delivery time probability distribution, the input parameters include the forecast period, actual consumption, delivery time, number of vehicles, single vehicle transportation, capacity freight rate, storage cost, ordering cost, initial inventory, inventory capacity, transportation distance, supply guarantee days, and basic parameters including population size, mutation probability, etc.

[0235] Step 3.2: Load the material supply demand forecasting model, pass the forecasting period, actual consumption, delivery time, number of vehicles, single vehicle transportation, capacity freight rate, storage cost, ordering cost, initial inventory, inventory capacity, transportation distance, supply guarantee days, etc. into the model, and calculate the material supply demand forecasting results for different time periods.

[0236] Step 4: Calculation of material ordering strategy;

[0237] Step 4.1: Combined with the probability distribution of delivery time limit and the calculation results of material supply demand forecast, the input parameters include simulation time, initial material inventory, material supply guarantee days, material delivery time, material storage capacity, actual project progress, and actual material consumption;

[0238] Step 4.2: Load the material inventory simulation algorithm based on Monte Carlo simulation and output the corresponding engineering material ordering strategy;

[0239] Step 5: Display the demand forecast results and output the engineering material supply demand and ordering strategy, that is, the supply demand of various materials in different sections and work sites and the time when the supply plan is issued.

[0240] (3) Functional design

[0241] The main functional modules of the material demand forecasting simulation system include map display, demand forecasting, supply forecasting, and results export, as shown in Table 4.

[0242] Table 4 System function overview

[0243]

[0244]

[0245] In summary, in terms of delivery time prediction, the present invention makes full use of trajectory data and historical weather data, combines the road condition characteristics obtained by path matching mining, and adopts Gaussian process regression to output the predicted distribution and evaluate the distribution. This method effectively solves the problems of insufficient consideration of influencing factors and lack of uncertainty quantification in the prior art, and provides a more accurate, comprehensive and reliable solution for cargo transportation time prediction.

[0246] In terms of material supply demand forecasting, existing railway construction projects usually prepare annual and monthly material demand plans based on construction organization plans under good material transportation conditions. However, the material transportation conditions of super-large railway engineering construction projects in complex and dangerous areas are complex, and the key point of material management lies in material supply guarantee. In order to ensure sufficient material supply during railway construction, this technology aims to accurately predict the supply demand of construction materials, and combines the study of the probability distribution of delivery time to propose a time-divided construction material supply demand forecasting method, so as to accurately grasp the on-site material demand and supply status, and ensure that materials will not be in short supply, nor will the supply of materials exceed the storage capacity of the warehouse.

[0247] In terms of the implementation of the demand forecasting simulation system, geographic information technology is introduced to provide accurate geographic grid information, and information such as road division and real-time weather conditions in each area is implemented in the model calculation. Comprehensive consideration of various supply and demand factors is added to refine the forecast of construction material supply demand. The combination of time forecasting and demand forecasting models is constructed to improve the accuracy and reliability of model calculations, and the forecasting simulation results are intuitively displayed through geographic information visualization.

[0248] Those skilled in the art can understand that the accompanying drawings are only schematic diagrams of an embodiment, and the modules or processes in the accompanying drawings are not necessarily required to implement the present invention.

[0249] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention or certain parts of the embodiments.

[0250] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0251] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for predicting the volatility of railway construction material demand based on delivery time limit, characterized in that: include: According to the vehicle's driving trajectory, cargo source distribution, transportation route, weather and road conditions, the Gaussian process regression-based prediction method is used to predict the probability distribution of cargo arrival time on different routes. Combined with the study of the probability distribution of cargo delivery time, a material supply demand forecasting model for railway engineering construction is constructed by considering factors such as delivery time limit, transportation cost, supply guarantee requirements, vehicle ownership and storage capacity, and the material supply demand in different time periods is predicted; Based on the forecast of material delivery time and material supply demand, combined with the actual project progress and material consumption, the Monte Carlo simulation algorithm is used to deduce the time series changes of material inventory, and the material ordering strategy, that is, the material ordering quantity and the material plan release time, is obtained; A demand forecasting simulation system is developed by integrating the probability distribution prediction method of cargo arrival time, the material supply demand prediction method and the material ordering strategy prediction method with geographic information data processing. The demand forecasting simulation system uses a railway engineering construction material demand volatility prediction method based on arrival time limit to predict the supply demand of various materials in different sections and work sites and the time to issue the supply plan.

2. The method according to claim 1, characterized in that The prediction method based on Gaussian process regression, which predicts the probability distribution of cargo arrival time along different routes according to vehicle driving trajectory, cargo source distribution, transportation route, weather and road conditions, includes: Comprehensively consider road conditions, transportation environment, transportation vehicles and date characteristics to obtain historical cargo transportation trajectory data and historical weather data, and use Gaussian process regression algorithm based on the historical cargo transportation trajectory data and historical weather data to build a cargo transportation travel time prediction model by mining vehicle transportation trajectories; Before a cargo transportation starts, the route, transportation start date and transportation vehicle information of the cargo transportation process are obtained according to the cargo transportation plan. The route, transportation start date and transportation vehicle information are processed and input into the cargo transportation travel time prediction model. The cargo transportation travel time prediction model outputs the predicted value of the arrival time of the cargo transportation process.

3. The method according to claim 2, characterized in that The processing of the route, transportation start date and transportation vehicle information is input into the cargo transportation travel time prediction model, and the cargo transportation travel time prediction model outputs the predicted value of the arrival time during the cargo transportation process, including: The route, transportation start date and transportation vehicle raw data are processed to obtain the characteristic values ​​related to road conditions, transportation environment, vehicle type and date. The processing process of the raw data is as follows: Transportation environment: If you want to obtain weather information, the required information includes date and location. The time is determined by the data returned by the vehicle track query interface. For the location, the longitude and latitude returned by the vehicle track query interface are matched to the specific location through the Baidu map coordinate picking system to obtain the weather information for the specific time and location; Date feature and transport vehicle: Date feature refers to whether it is a holiday. The time in the trajectory data can be used to determine whether the transport process is on a holiday. The type of transport vehicle can be obtained by reading the trajectory data. Road conditions: Obtain the road conditions corresponding to the trajectory by matching the read trajectory data to the specific road section; The cargo transportation travel time prediction model finds a potential function to express the relationship between the feature data and the target vector by learning the data in the training set, which will affect the feature data x of the cargo transportation travel time. i Defined as: x i =(d i ,t i ,q i ,c i ) Where, d i =(d 1i ,d 2i ,…,d 6i ) is the road condition feature vector; t i is the date feature vector; q i is the meteorological information feature vector, c i is the vehicle type feature vector; The travel time of the cargo on each road is defined as y i , the cargo transportation travel time prediction model is expressed by the following formula: f(x)~GP(m(x),k(x,x′)) Where m(x) is the mean function, assumed to be a constant, and k(x,x′) is the covariance function (or kernel function) used to measure the similarity between the input data x and x′; The training process of the freight transport travel time prediction model is defined by the following formula: f=[f(x1),f(x2),……,f(x n )] T Where f is the predicted value vector of all training data points, n is the number of training data, and the observed value is y = [y1, y2, ..., y n ] T ; The prediction process of the freight transportation travel time prediction model is expressed as follows: p(y * ∣x * ,X,y)=N(μ * ,s *2 ) where y * is the output value of the new data point to be predicted, x * The features of the new data point to be predicted, X is the features of the training data, y is the output of the training data, μ * is the mean of the predicted values, σ *2 is the variance of the predicted value, calculated as follows: where k * is the covariance vector of the predicted data point and all training data points, and K is the covariance matrix between the training data points.

4. The method according to claim 3, characterized in that The above-mentioned study on the probability distribution of cargo delivery time takes into account the delivery time limit, transportation cost, supply guarantee requirements, vehicle ownership and storage capacity factors to build a material supply demand forecasting model for railway engineering construction, and predicts the material supply demand in different time periods, including: Based on the above-mentioned study on the probability distribution of cargo arrival time, taking into account factors such as arrival time limit, transportation cost, supply guarantee requirements, vehicle ownership and storage capacity, and taking cost minimization as the goal, a material supply demand forecasting model for railway engineering construction is constructed under the framework of storage theory, and the constraints of the material supply demand forecasting model are set based on the time series balance of material inventory, material supply guarantee, storage capacity, vehicle ownership and actual material consumption factors; The actual on-site construction progress of the railway project and the actual material consumption, delivery time limit, transportation costs, supply guarantee requirements, vehicle ownership and storage capacity data are processed and input into the material supply demand forecasting model. The heuristic algorithm is used to solve the material supply demand forecasting model, and the optimal material arrival quantity for each day in the forecast period is calculated to obtain the construction material supply demand in different time periods.

5. The method according to claim 4, characterized in that The material supply demand forecasting model is based on the actual construction organization plan and actual material consumption of the project site, with inventory capacity, supply guarantee and transportation capacity as constraints, and the minimum average cost as the objective function. The meanings of the relevant symbols in the material demand forecasting model are shown in Table 1; Table 1 Symbolic definitions of the material supply demand forecasting model The objective function of the material supply demand forecasting model is composed of the following formula: Taking the minimum average ordering cost, storage cost and freight cost in one cycle as the goal, the objective function is constructed as follows: The constraints of the material demand forecasting model are composed of the following formula: Material inventory constraint: indicates that the material inventory on day t is equal to the material inventory on day t-1 plus the material arrival on day t minus the material consumption on day t; S(t)=S(t-1)+A(t)-C(t) Inventory capacity constraint: indicates that the inventory of materials on day t should meet the material demand for the next Δt days and be less than the inventory capacity; D(t+Δt)≤S(t)≤N Material arrival time constraint: means that the material arrival quantity on day t is equal to the order quantity before the delivery time σ days; A(t)=O(t-σ) Material arrival quantity constraint: The material arrival quantity on day t is equal to the number of vehicles that arrive on day t multiplied by the transport capacity of a single vehicle; A(t)=M*E(t) Material delivery time probability distribution constraint: indicates that the material delivery time σ follows the probability distribution P(σ); σ~P(σ) Material consumption constraint: indicates that the material consumption on day t is determined by the actual project progress and the actual quota consumption; C(t)=f(AP,AC) Material demand constraint: indicates that the material demand in the next Δt days is determined by the planned project progress and the estimated quota consumption; D(t+Δt)=g(PP,PC) The solution process of the heuristic algorithm includes: Step 1: Initialize the population Coding rules: Assume that the number of vehicles that arrive on day t is A(t), and the number of vehicles that arrive every day in the forecast period T is composed of an array A=[A(1), A(2), …, A(T)], and convert the decimal array into a binary array, A′=[A′(1), A′(2), …, A′(T)]; Set the evolution generation counter n = 0, set the maximum evolution generation N, and randomly generate M individuals as the initial population P(0); Step 2: Individual Assessment Calculate the fitness F[P(m)] of each individual in the population P(m), substitute the number of vehicles that deliver goods represented by different individuals into the objective function, calculate the average cost, and based on this, derive the fitness of the individual: F[P(m)]=1 / K[P(m)] Step 3: Select the operation Apply the selection operator to the population, and directly pass the optimized individuals to the next generation or generate new individuals through pairing and crossover and then pass them to the next generation. The selection operation is based on the fitness evaluation of individuals in the population. Based on the calculation results of the fitness of each individual in the population, each individual is given a probability of being selected, and the next generation of basic population is formed through multiple sampling. Step 4: Crossover Apply the crossover operator to the population. The so-called crossover refers to the operation of replacing and recombining part of the structure of two parent individuals to generate new individuals. With a certain probability p, some parent individuals are selected from the next generation basic population for crossover, and individuals are reconstructed by cutting off from a certain position. Step 5: Mutation operation Applying mutation operators to the population means changing the gene values ​​at certain loci of the individual strings in the population. Using binary numbers to construct individuals, the values ​​at a certain locus in the individual are transformed from 0 to 1 with a certain probability p. After the population P(n) undergoes selection, crossover, and mutation operations, the next generation population P(n+1) is obtained. Step 6: Termination condition judgment If n=N, the individual with the maximum fitness obtained in the evolution process is output as the optimal solution, the calculation is terminated, the daily arrival volume of materials within the prediction period is output, and the material supply demand in different time periods is obtained.

6. The method according to claim 5, characterized in that The material delivery time and material supply demand forecast is combined with the actual project progress and material consumption, and the Monte Carlo simulation algorithm is used to deduce the time series change of material inventory to obtain the material ordering strategy, that is, the material ordering quantity and the material plan release time, including: The material inventory simulation process based on Monte Carlo simulation is as follows: Input: number of simulations M; simulation duration T; initial material inventory D(0); material supply days requirement Δt; material delivery time σ probability distribution P(σ); material storage capacity N; actual project progress AP; actual quota consumption AC; planned project progress PP, estimated quota consumption PC; process: Step 1: Start the simulation of material demand volatility forecasting for the i-th time; Step 2: Deducing the material inventory S(t) on day t: Step 2.1: Calculate the material consumption c(t) on day t based on the project progress AP and the actual quota consumption AC; Step 2.2: Obtain the material delivery volume A(t) on the tth day based on the output of the material supply demand forecasting model; Step 2.3: Calculate the material inventory S(t) on day t based on the material inventory S(t-1) on day t-1, the material consumption C(t) on day t, and the material delivery volume A(t) on day t; Step 3: Calculate the material demand D(t+Δt) within the next Δt days based on the planned project progress PP and the estimated quota consumption PC; Step 4: Sample from the delivery time probability distribution P(σ) to determine the delivery time σ of the materials when scheduled on the tth day t ; Step 5: Determine the order time t-σ for materials arriving on day t t ; Step 6: If t=T, end the i-th simulation; otherwise, t=t+1, return to step 2; Step 7: If i=M, end the simulation; otherwise, i=i+1, return to step 1; Output: Material ordering strategy, which includes the material ordering quantity in different time periods and the planned release time of materials.

7. The method according to claim 6, characterized in that The integrated cargo delivery time probability distribution prediction method, material supply demand prediction method and material ordering strategy prediction method are combined with geographic information data processing to develop a demand prediction simulation system. The demand prediction simulation system uses a railway engineering construction material demand volatility prediction method based on delivery time limit to predict the supply demand of various materials at different bidding sections and work sites and the time of issuing supply plans, including: The prediction process of the probability distribution of the cargo arrival time, the prediction process of the construction material supply demand in different time periods and the material inventory deduction process based on Monte Carlo simulation are integrated, and a demand prediction simulation system is developed in combination with geographic information data processing. The demand prediction simulation system uses a prediction method for the volatility of railway engineering construction material demand based on the arrival time limit, taking into account multiple factors such as the actual demand for engineering materials, the material transportation network and route selection, the probability distribution of the arrival time limit of different types of materials and the material storage capacity, to achieve the prediction of the supply demand of various materials in different sections and work sites and the time of issuing the supply plan; The calculation process of the demand forecasting simulation system includes: Step 1: Basic parameter settings, including selecting the forecast initial time, forecast days, supply guarantee days, bidding section, warehouse and supply source, and obtaining information on environmental parameters, warehouse storage capacity, storage cost and actual material consumption; Step 2: Calculate the probability distribution of delivery time; Step 2.1: Obtain weather conditions, road conditions, transportation environment and vehicle type parameters for each road section; Step 2.2: Load the delivery time probability distribution prediction model, pass the route section information, including the section mileage, road grade and road width average speed, into the time probability distribution prediction model, and calculate the volatility delivery time probability distribution prediction result; Step 3: Calculation of material supply demand forecast; Step 3.1: Combined with the calculation results of the delivery time probability distribution, the input parameters include the forecast period, actual consumption, delivery time, number of vehicles, single vehicle transportation, capacity freight rate, storage cost, ordering cost, initial inventory, inventory capacity, transportation distance and supply guarantee days, as well as basic parameters including population size and mutation probability; Step 3.2: Load the material supply demand forecasting model, input the forecasting period, actual consumption, delivery time, number of vehicles, single vehicle transportation, capacity freight rate, storage cost, ordering cost, initial inventory, inventory capacity, transportation distance and supply guarantee days into the material demand forecasting model, and calculate the material supply demand forecasting results for different time periods; Step 4: Calculation of material ordering strategy; Step 4.1: Combined with the probability distribution of delivery time limit and the calculation results of material supply demand forecast, the input parameters include simulation time, initial material inventory, material supply guarantee days, material delivery time, material storage capacity, actual project progress and actual material consumption; Step 4.2: Load the material inventory simulation algorithm based on Monte Carlo simulation and output the corresponding material ordering strategy; Step 5: Display the demand forecast results and output the material supply demand and material ordering strategy, that is, output the supply demand of various materials in different sections and work sites and the time when the supply plan is issued.

Images