A kind of rail traffic dispatching method based on big data and Internet of Things
Technical field
Patent of the present invention belongs to rail traffic dispatching algorithm, specifically, being related to a kind of based on big data and Internet of Things
Rail traffic dispatching algorithm.
Background technology
Existing rail traffic dispatching algorithm is all using prediction algorithm, such as genetic algorithm, population by mathematical modeling
Algorithm etc. carries out passenger flow estimation, is then dispatched a car according to prediction, scheduling of stopping;And in the environment for currently thering is big data to support
Under, it can be based on history big data completely, carry out more simple and effective passenger flow estimation, while in the present of Internet of Things high speed development
My god, each website and train passenger flow data can be obtained in real time, and the realization for dynamic adjustment in real time provides possibility.
Invention content
The purpose of the present invention is establish a kind of based on big data and Internet of Things, open, the track friendship that can continue to optimize
Logical dispatching method.
Present invention requirement first has the following conditions:1st, the data warehouse of different moments train passenger number has been established;2、
The device of monitoring train current passenger number in real time has been established, and passes through Internet of Things and synchronizes and be sent to background server and each train,
Backstage scheduling control information being capable of each train of delivered in real time simultaneously, it is ensured that Real-Time Scheduling can be realized.
Technical scheme of the present invention includes two parts:1st, using the volume of the flow of passengers of big data prediction daily different moments, and with
This is according to the basic data for determining daily different moments headway;2nd, between the data monitored according to Internet of Things are to dispatching a car
It is finely adjusted every the time.
Specific embodiment
Above-mentioned part 1 predicts the volume of the flow of passengers of daily different moments using big data, and determines daily on this basis not
The basic thought of the basic data of headway is in the same time:Any time volume of the flow of passengers depends on the moment corresponding event
Feature, when event includes month belonging to the moment, week, date, hour, festivals or holidays, the new term begins time, school of school have a holiday or vacation
Between, various shocking flashes and other dependent events, if obtaining disturbance degree of each event to the volume of the flow of passengers by big data, just
More accurately the volume of the flow of passengers of any time can be predicted, specific technical solution includes:
(1) track circuit ridership P is defined sometime to put the sum of all train passenger numbers in current orbit circuit, it will be over the years
This track circuit ridership is sampled at regular intervals, is stored in data warehouse;
(2) average value of this track circuit ridership over the years is obtained in the method for arithmetic average, is denoted as respectively
The then history average of this track circuit ridershipSimultaneously by the average value of this track circuit ridership over the yearsBy linear regression method, the average value of this track circuit next year ridership is obtained, withIt represents,
At this timeNot real data, but to the prediction data of next year average passenger number;
(3) disturbance degree of the different event to the volume of the flow of passengers is obtained, is denoted as the weights of different event, method is:By with different things
Part E is index, retrieves all P for meeting event E conditions from data warehouse, the condition for meeting event E equipped with k, the k
It is denoted as P respectively1、P2……Pk, then meet the mean value of the track circuit ridership of the eventThen event E is to passenger flow
Measure the weights of disturbance degree
(4) according to each event weights, the passenger flow forecast amount at following each moment is obtained, method is:When analysis is following each successively
Corresponding event is carved, the weights of its corresponding all event are subjected to arithmetic average, obtain the weights λ at the moment, method is:
If a certain moment corresponds to m event, this m event is λ respectively to the weights of volume of the flow of passengers disturbance degree1、λ2……λm, then moment
WeightsThe passenger at the track circuit moment predicts that number isWhereinTo be predicted multiplying for year
Objective number mean value;
(5) headway at the moment is determined according to the plan carrying number of any moment passenger flow forecast amount, each train, side
Method is:If the time of the complete a circuit of train operation, each Train operation plan carrying number was P for TP, then headway t=
T·PP/PF, this headway is basic data, in actual moving process, further according to the data of part 2 Internet of Things monitoring
Headway is finely adjusted.
Above-mentioned part 2 includes according to the technical solution that the data that Internet of Things monitors are finely adjusted headway:
(1) this track circuit current passenger number P is collected;
(2) by currently practical ridership P and current time prediction ridership PFRatio, current time bias ratio μ=P/ is obtained
PF, then the prediction ridership of subsequent time be modified to initial value and be multiplied by bias ratio, i.e. PF=μ PF(equal sign is assignment herein), will repair
P after justFValue substitutes into formula t=TPP/PFIn, acquire the headway t after adjustment.
Embodiment
Assuming that the passenger data of known a certain track circuit the first three years,Then
History average passenger numberNext year prediction average passenger number is calculated for first three annual data with linear regression
It obtains,
Assuming that count in history all January average passenger numbers be 910 people, several 920 people of Monday average passenger, the morning
8:10 ridership, 1080 people, can obtain January, Monday, 8:The disturbance degree of 10 these three events is with λ1、λ2、λ3It represents, respectively:At this time if a certain moment of prediction just only
Three above event is triggered, then the ridership that the moment is expected
Assuming that the time of complete one time of this circuit train operation is 100 minutes, i.e. T=100, every time Train operation plan carrying number is
100 people, i.e. PP=100, then headway t=TPP/PF=100100/989.4=10.1 minutes;
Assuming that sampling interval duration is 10 minutes, and all mornings 8 in history:20 average passenger number is 1077 people, and
Above-mentioned 8:10 actual passenger numbers are 942 people, then current time bias ratio μ=P/PF=942/989.4=0.952, subsequent time
Predicted passenger number isAccording to deviation
Rate is adjusted PF=μ PF=0.952988.38=940.94, at this time headway be adjusted to t=TPP/PF=
100100/940.94=10.63, i.e. headway were adjusted to 10.63 minutes from 10.1 minutes.
Compared with prior art, this method has preferable open, accuracy and adaptivity, can be with research
Deeply, new event is continuously increased, event is more perfect, and accuracy is higher, while can also be according to the continuous dynamic of actual conditions
Amendment headway cause train operation more efficiently.