CN113742879A - Method and device for designing and optimizing scheduling model based on passenger flow simulation - Google Patents

Method and device for designing and optimizing scheduling model based on passenger flow simulation Download PDF

Info

Publication number
CN113742879A
CN113742879A CN202010407472.4A CN202010407472A CN113742879A CN 113742879 A CN113742879 A CN 113742879A CN 202010407472 A CN202010407472 A CN 202010407472A CN 113742879 A CN113742879 A CN 113742879A
Authority
CN
China
Prior art keywords
passenger flow
driver
passengers
departure
time
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010407472.4A
Other languages
Chinese (zh)
Other versions
CN113742879B (en
Inventor
熊林海
周金明
赵丽
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Xingzheyi Intelligent Transportation Technology Co ltd
Original Assignee
Nanjing Xingzheyi Intelligent Transportation Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing Xingzheyi Intelligent Transportation Technology Co ltd filed Critical Nanjing Xingzheyi Intelligent Transportation Technology Co ltd
Priority to CN202010407472.4A priority Critical patent/CN113742879B/en
Publication of CN113742879A publication Critical patent/CN113742879A/en
Application granted granted Critical
Publication of CN113742879B publication Critical patent/CN113742879B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/10—Office automation; Time management
    • G06Q10/109—Time management, e.g. calendars, reminders, meetings or time accounting
    • G06Q10/1093—Calendar-based scheduling for persons or groups
    • G06Q10/1097—Task assignment
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • Tourism & Hospitality (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Evolutionary Computation (AREA)
  • Computer Hardware Design (AREA)
  • General Engineering & Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Geometry (AREA)
  • Data Mining & Analysis (AREA)
  • Educational Administration (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Train Traffic Observation, Control, And Security (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a design method, an optimization method and a device of a scheduling model based on passenger flow simulation, wherein the design method mainly comprises the following steps: setting scheduling control parameters; predicting the number of passengers in each station in every x minutes according to the passenger flow of the stations corresponding to the same type of dates in the history, and predicting the number of passengers in the vehicle after departure of each station according to the number of passengers getting on or off the station corresponding to the number of the passengers in each shift in the history in the days; determining departure timetables and driver allocation rules according to the limiting conditions; and (3) fully utilizing historical passenger flow rules, distributing vehicles and drivers to carry out actual simulation operation according to the optimal real load rate, and finally obtaining an actual available driving plan.

Description

Method and device for designing and optimizing scheduling model based on passenger flow simulation
Technical Field
The invention relates to the field of big data and intelligent traffic research, in particular to a design method, an optimization method and a device of a scheduling model based on passenger flow simulation.
Background
Along with the rapid development of the society of China, the living standard of people is increasingly improved, the scale of urban areas is also continuously enlarged, public transport is also greatly developed, urban transport workers have various forms, wherein public transport is the most common transport means of public transport systems, and in the process of realizing the invention, the inventor finds that at least the following problems exist in the prior art: at present, most of urban bus departure schedules are generated according to manual experience, and the departure schedules are manually formed, so that public travel waiting time is long, the phenomena of 'train crossing' and 'large interval' are obvious, the bus congestion is serious in peak time, and the no-load phenomenon of buses in peak time generally exists; therefore, generating a departure schedule according with the passenger flow rule is particularly important; meanwhile, the bus company artificially makes a person and vehicle allocation plan for the driving schedule, so that part of drivers often work in an overload mode, part of drivers are relatively idle, and the eating and rest time of the drivers cannot be considered, so that the generation of the scientific and reasonable driving plan is more important.
Disclosure of Invention
In order to overcome the defects of the prior art, the embodiment of the disclosure provides a method, a method and a device for designing and optimizing a scheduling model based on passenger flow simulation, which make full use of historical passenger flow rules and distribute vehicles and drivers to perform actual simulation operation according to the optimal real-load rate, so that the rationality of a driving plan is improved, and the effects of cost reduction and efficiency improvement are achieved. The technical scheme is as follows:
in a first aspect, an embodiment of the present disclosure provides a method for designing a shift scheduling model based on passenger flow simulation, where the method includes the following steps:
step 1, setting scheduling control parameters: determining the earliest departure time of the first shift of the line, taking the earliest departure time as the first shift for departure, and recording the earliest departure time as a1(ii) a Minimum value r of rest time of driverminAnd maximum value rmaxThe maximum number of people N in the platformmax(ii) a Minimum value of departure interval dminAnd maximum value dmaxThe optimum load factor is hop。
Step 2, predicting the number of people in each station every x minutes according to the passenger flow of the stations corresponding to the same type of dates in the history; and predicting the number of passengers in the vehicle after departure at each stop according to the number of passengers getting on or off at the corresponding class on the historical multiple days, wherein the actual load rate is the number of passengers in the vehicle/the rated nuclear load of the bus.
Step 3, determining departure timetable and driver allocation rule according to limiting conditions
Note a1The next equidirectional shift of (A)3In the same way, a2The next equidirectional shift of (A)4Then go upward for shift a3The dispatching rule of the driver is that if the driver d1Execution of a1Post-execution of a2Is less than a minimum value rminThe driver is rescheduled, otherwise the driver d1Continue to execute a2(ii) a Execution of a2Is recorded as d by the driver2。
If a1The real load rate reaches the optimal real load rate hopWhen the departure interval is greater than or equal to the minimum value dminWhen the next shift is sent, the next shift a3If the driver d2Execution of a2Then, a is continuously executed3Is less than a minimum value rminThe driver is rescheduled, otherwise it is performed by the driver arriving first in the opposite direction, and if not, a new driver is rescheduled.
If a1The real load rate of the power grid does not reach the optimal real load rate hopWhen the number of people in the station is more than or equal to the maximum value N, the next vehicle a is sent3If the driver d2Execution of a2Then, a is continuously executed3Is less than a minimum value rminThe driver is rescheduled, otherwise it is performed by the driver arriving first in the opposite direction, and if not, a new driver is rescheduled.
If a1The real load rate of the power grid does not reach the optimal real load rate hopAnd when the number of people in the station does not reach the maximum value N, the departure interval is more than or equal to the maximum value dmaxIf so, the next vehicle a is sent at the moment3If the driver d2Execution of a2Then, a is continuously executed3Is less than a minimum value rminThe driver is rescheduled, otherwise it is performed by the driver arriving first in the opposite direction, and if not, a new driver is rescheduled.
Determining the downlink shift a in the same manner4The departure driver rule of (1).
And sequentially determining departure and driver allocation rules of all the shifts according to the method.
In a second aspect, an embodiment of the present disclosure provides a method for optimizing a shift scheduling model based on passenger flow simulation, where the method includes the following steps:
the method comprises the following steps of adopting Markov decision to carry out reinforcement learning on a driving plan obtained by a design method of a scheduling model based on passenger flow simulation, and intelligently generating an optimal driving plan, wherein the specific method comprises the following steps:
markov decision is taken as (S, A, T, r, S)0) Wherein, in the step (A),
s refers to the driving plan (state space),
a refers to adjusting (increasing, decreasing or not changing) the optimum real load rate hopAnd executing a method (action space) for designing a scheduling model based on passenger flow simulation;
t is the transition probability space
r is a function of the reward, r is,
Figure BDA0002491861950000021
Figure BDA0002491861950000022
alpha is weight value, K is number of segments, q is passenger unit time and other vehicle cost, HiIs the number of passengers in the i period, Δ tiIs the departure interval in the ith time period, p is the cost per kilometer of the bus, L is the one-way route mileage, TiIs the time span of the ith time period, C is the passenger fare, J is the number of drivers, tjThe corresponding working hours of the driver j, and the unit working hour wage of the driver D.
The reinforcement learning process is as follows:
(1)s0is an initial state, specifically: setting the value range of the real load rate as [ hmin,hmax]In [ h ]min,hmax]Internal random generation of optimal real load rate hopBased on hopCalculating a reward function Q according to the driving plan obtained by the method for designing the scheduling model based on the passenger flow simulation0。
(2) In [ h ]min,hmax]Increasing or decreasing the optimum loading rate hopObtaining a new state s according to the method for designing the scheduling model based on the passenger flow simulation1Calculating a reward function r1＝Q1-Q0If r is1Increasing, then giving the action a0Greater probability distribution pi (a)0|s0) Otherwise, giving a smaller probability;
recording the motion track as(s)0,a0,r1,s1,a1,r2…), noting the cumulative prize for that track as R ═ R
Figure BDA0002491861950000031
Note its probability distribution as tauπ＝∏p(st+1|st,at)π(at|st) Wherein p(s)t+1|st,at) For transition probabilities, at which time it is desirable to maximize the reward, i.e.
Figure BDA0002491861950000032
Wherein the discount coefficient gamma belongs to [0,1) ]tTo the power of t of γ, decreases with increasing t.
(3) Setting the maximum iteration times, repeating the steps (1) to (2), recording reward values corresponding to all action tracks, recording the maximum reward as V, and recording the corresponding strategy as the optimal strategy as tau*The corresponding optimum state is s*I.e. an optimal driving plan.
In a third aspect, the disclosed embodiment provides a device of a scheduling model based on passenger flow simulation, the device includes a parameter setting unit, a passenger flow prediction unit, and a scheduling unit, the units are electrically connected in sequence;
the parameter setting unit is used for executing the step 1 of the design method of the scheduling model based on passenger flow simulation in any one of all possible implementation methods;
the passenger flow prediction unit is used for executing the step 2 of the design method of the scheduling model based on passenger flow simulation in any one of all possible implementation methods;
the scheduling unit is used for executing the step 3 of the design method of the scheduling model based on passenger flow simulation in any one of all possible implementation methods.
Compared with the prior art, one of the technical schemes has the following beneficial effects: the method has the advantages that through setting constraint conditions such as normal departure intervals, real load rates, driver rest time length and reasonable driver working hours, historical passenger flow rules are fully utilized, vehicles and drivers are distributed according to the optimal real load rates to carry out actual simulation operation, finally, the practical available driving plans are obtained, the reasonable constraint conditions can be set according to optimal parameters in the actual operation, the practical available driving plans can be automatically generated without any complex algorithm, generation of timetables and the driving plans can be carried out simultaneously, the practical passenger flow rules are completely met, and resource allocation optimization is achieved. And generating an optimization model of the driving plan through Markov decision, taking a resource optimization configuration target as a learning target, and continuously training and learning through a machine to obtain an optimal scheduling strategy and an optimal driving plan.
Detailed Description
In order to clarify the technical solution and the working principle of the present invention, the embodiments of the present disclosure will be described in further detail below. All the above optional technical solutions may be combined arbitrarily to form the optional embodiments of the present disclosure, and are not described herein again.
The terms "step 1," "step 2," "step 3," and the like in the description and claims of this application are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It should be understood that the data so used may be interchanged under appropriate circumstances such that the embodiments of the application described herein may, for example, be implemented in an order other than those described herein.
The following adopts a passenger flow simulation scenario in the bus field as an exemplary application scenario in the embodiment of the present application, but the technical scheme of the present application is not limited to the bus field, and can also be used in the fields of enterprise buses, subways, and the like, which adopt similar operation modes as buses.
In a first aspect, an embodiment of the present disclosure provides a method for designing a shift scheduling model based on passenger flow simulation, where the method includes the following steps:
step 1, setting scheduling control parameters: determining the earliest departure time of the first shift of the line, taking the earliest departure time as the first shift for departure, and recording the earliest departure time as a1(ii) a Minimum value r of rest time of driverminAnd maximum value rmaxThe maximum number of people N in the platformmax(ii) a Minimum value of departure interval dminAnd maximum value dmaxThe optimum load factor is hop。
Preferably, the earliest departure time of the first shift of the route is determined and is taken as the departure time of the first shift, and the earliest departure time is marked as a1The method specifically comprises the following steps: acquiring the time of the first and last uplink shifts of a certain line shift as t1,t2]The first and last down shift time is [ t ]3,t4]Finding out the earliest time min (t) of the first shift of the line1,t3) Departure as the first shift, marked as a1And arranging the first driver as d1(ii) a In the bus shift schedule, assume a1For ascending, the ascending shift is a1,a3,a5…, the downward shift is a2,a4,a6,…
And 2, predicting the number of people in each station every x minutes according to the passenger flow of the stations corresponding to the same type of dates (festivals and holidays, weather, chills and hots are consistent) of the historical days.
Preferably, the number of people in each station per x minutes is predicted according to the passenger flow of the station corresponding to the same type of dates in a plurality of historical days, and the method specifically comprises the following steps: obtaining a high-flat peak period according to a Fisher optimal segmentation method: according to the corresponding passenger flow data (distinguishing working days and double holidays) of all the historical sites, segmenting the high and flat peak of the passenger flow, and designing a high and flat peak period generation model: note all every T of site1Time particle size (preferably, T)115 minutes) is { X } (i.e., one-way full time)1,X2,…,XNIn Fisher's optimal segmentation method, the total variance is
Figure BDA0002491861950000051
Supposing that the passenger flow data in the all-day station is divided into K groups according to time intervals, namely the classification number is K, then
Figure BDA0002491861950000052
Wherein
Figure BDA0002491861950000053
Is the mean value of the intrastation passenger of the jth time period,
Figure BDA00024918619500000510
is the mean of the passenger flow in the whole day station, wherein
Figure BDA0002491861950000054
Is the sum of squared deviations in the group,
Figure BDA0002491861950000055
the square sum of the differences between groups.
Definition of
Figure BDA0002491861950000056
DKThe classification number is K, namely when the deviation square sum in the group is minimum, the corresponding segmentation method is the optimal segmentation method; by minimizing the sum of squared deviations D within the groupKAnd obtaining an optimal separation method, so that the difference among the samples of the same type is minimum, and the difference among the samples of all types is maximum.
Traversing the classification number K from 1 to n (i.e., from 1 to 9), n ∈ 9, 15]For each value of K, the corresponding D is calculatedKDecay rate based on minimizing sum of squared deviations within a group
Figure BDA0002491861950000057
Take alphaKAnd the maximum corresponding classification number K is the optimal classification number, and the time interval optimal division method is obtained according to the optimal classification number.
Because the travel rules of working days and non-working days are greatly different, working days and double holidays are divided; because the passenger flow rules of different peak periods in the same day are different, when the peak periods are divided, the historical passenger flow rules of different periods are obtained according to the optimal time division method, the actual passenger flow quantity in each period is obtained, and the passenger flow quantity in the h-th period is recorded as fh
Predicting the number of passengers in each station per x minutes according to the Poisson distribution, namely the probability function corresponding to the number k of passengers arriving at the station in x minutes (x is 1-30min) is as follows:
Figure BDA0002491861950000058
the number of passengers in the station at x minutes in the time period is obtained as follows: argmax P (X ═ k).
Where lambda is the average incidence of passenger arrivals per unit of time,
Figure BDA0002491861950000059
i.e. λ over period h equals the average number of passengers over historical h period, where fhIs the number of passengers arriving at the station in time period h, thIs the time span (in minutes) of period h.
And predicting the number of passengers in the vehicle after departure of each station according to the number of passengers on or off a bus corresponding to the same type of date (holidays, weather, cold and summer holidays) on multiple historical days, wherein the actual load rate is the number of passengers in the vehicle/the rated bus load.
Preferably, the number of passengers in the vehicle after departure of each station is predicted according to the number of passengers getting on or off the vehicle in the past corresponding to the same type of dates in multiple days, and the method specifically comprises the following steps: predicting all every T intervals according to the number of passengers on and off each site in history2Time particle size (preferably, T)21-30 minutes), i.e., the predicted traffic at the ith stop is
Figure BDA0002491861950000061
Figure BDA0002491861950000062
Figure BDA0002491861950000063
Representing the time granularity corresponding to the historical value of the passenger flow; the same method can predict the passenger flow, and the number of passengers in the vehicle after departure at the ith station can be predicted according to the predicted number of passengers on the vehicle
Figure BDA0002491861950000064
The predicted passenger flow is:
Figure BDA0002491861950000065
Figure BDA0002491861950000066
the presentation time granularity corresponds to the passenger flow history value.
Further, in order to maximize the benefit of passengers, the predicted passenger flow at the ith station is
Figure BDA0002491861950000067
The calculation formula of (2) is replaced by:
Figure BDA0002491861950000068
wherein the content of the first and second substances,
Figure BDA0002491861950000069
the presentation time granularity corresponds to the historical value of the passenger flow,
Figure BDA00024918619500000610
represents the median of the plurality of history values, and n represents the number of history values equal to or greater than the median.
Step 3, determining departure timetable and driver allocation rule according to limiting conditions
Note a1The next equidirectional shift of (A)3In the same way, a2The next equidirectional shift of (A)4Then go upward for shift a3The dispatching rule of the driver is that if the driver d1Execution of a1Post-execution of a2Is less than a minimum value rminThe driver is rescheduled, otherwise the driver d1Continue to execute a2(ii) a Execution of a2Is recorded as d by the driver2(d1And d2Possibly the same driver or different drivers).
If a1The real load rate reaches the optimal real load rate hopWhen the departure interval is greater than or equal to the minimum value dminWhen the next shift is sent, the next shift a3If the driver d2Execution of a2Then, a is continuously executed3Is less than a minimum value rminThe driver is rescheduled, otherwise it is performed by the driver arriving first in the opposite direction, and if not, a new driver is rescheduled.
If a1The real load rate of the power grid does not reach the optimal real load rate hopWhen the number of people in the station is more than or equal to the maximum value N, the next vehicle a is sent3If the driver d2Execution of a2Then, a is continuously executed3Is less than a minimum value rminThe driver is rescheduled, otherwise it is performed by the driver arriving first in the opposite direction, and if not, a new driver is rescheduled.
If a1The real load rate of the power grid does not reach the optimal real load rate hopAnd when the number of people in the station does not reach the maximum value N, the departure interval is more than or equal to the maximum value dmaxIf so, the next vehicle a is sent at the moment3If the driver d2Execution of a2Then, a is continuously executed3Is less than a minimum value rminThe driver is rescheduled, otherwise it is performed by the driver arriving first in the opposite direction, and if not, a new driver is rescheduled.
Determining the downlink shift a in the same manner4The departure driver rule of (1).
And sequentially determining departure and driver allocation rules of all the shifts according to the method.
Preferably, the limiting conditions for determining the departure time table and the driver allocation rule further include that the total times of the driver working in one day are less than a predetermined value, and a new driver is arranged if the times of the driver working in one day are greater than or equal to the predetermined value.
Preferably, if the rest period occurs during a meal time period and the driver does not eat, the minimum value of the rest period is dmin+ e, maximum value dmax+ e, e is the length of the meal.
In a second aspect, an embodiment of the present disclosure provides a method for optimizing a shift scheduling model based on passenger flow simulation, where the method includes the following steps:
the method comprises the following steps of adopting Markov decision to carry out reinforcement learning on a driving plan obtained by a design method of a scheduling model based on passenger flow simulation, and intelligently generating an optimal driving plan, wherein the specific method comprises the following steps:
markov decision is taken as (S, A, T, r, S)0) Wherein, in the step (A),
s refers to the driving plan (state space),
a refers to adjusting (increasing, decreasing or not changing) the optimum real load rate hopAnd executing a method (action space) for designing a scheduling model based on passenger flow simulation;
t is the transition probability space
r is a function of the reward, r is,
Figure BDA0002491861950000071
Figure BDA0002491861950000072
alpha is weight value, K is number of segments, q is passenger unit time and other vehicle cost, HiIs the number of passengers in the i period, Δ tiIs the departure interval in the ith time period, p is the cost per kilometer of the bus, L is the one-way route mileage, TiIs the time span of the ith time period, C is the passenger fare, J is the number of drivers, tjThe corresponding working hours of the driver j, and the unit working hour wage of the driver D.
The reinforcement learning process is as follows:
(1)s0is an initial state, specifically: setting the value range of the real load rate as [ hmin,hmax]In [ h ]min,hmax]Internal random generation of optimal real load rate hopBased on hopCalculating a reward function Q according to the driving plan obtained by the method for designing the scheduling model based on the passenger flow simulation0。
(2) In [ h ]min,hmax]Internal random augmentationOr reducing the optimum loading rate hopObtaining a new state s according to the method for designing the scheduling model based on the passenger flow simulation1Calculating a reward function r1＝Q1-Q0If r is1Increasing, then giving the action a0Greater probability distribution pi (a)0|s0) Otherwise, giving a smaller probability;
recording the motion track as(s)0,a0,r1,s1,a1,r2…), noting the accumulated reward for the track as
Figure BDA0002491861950000081
Figure BDA0002491861950000082
Noting that its probability distribution is
Figure BDA0002491861950000083
Wherein p(s)t+1|st,at) For transition probabilities, at which time it is desirable to maximize the reward, i.e.
Figure BDA0002491861950000084
Wherein the discount coefficient gamma belongs to [0,1) ]tTo the power of t of γ, decreases with increasing t.
(3) Setting the maximum iteration times, repeating the steps (1) to (2), recording reward values corresponding to all action tracks, recording the maximum reward as V, and recording the corresponding strategy as the optimal strategy as tau*The corresponding optimum state is s*I.e. an optimal driving plan.
Preferably, for other new lines, if the difference or fluctuation between the passenger flow law of the new line and the passenger flow law of the optimized line is within a given threshold value range, the new line can directly apply the strategy tau*And obtaining the optimal driving plan.
In a third aspect, the disclosed embodiment provides a device of a scheduling model based on passenger flow simulation, the device includes a parameter setting unit, a passenger flow prediction unit, and a scheduling unit, the units are electrically connected in sequence;
the parameter setting unit is used for executing the step 1 of the design method of the scheduling model based on passenger flow simulation in any one of all possible implementation methods;
the passenger flow prediction unit is used for executing the step 2 of the design method of the scheduling model based on passenger flow simulation in any one of all possible implementation methods;
the scheduling unit is used for executing the step 3 of the design method of the scheduling model based on passenger flow simulation in any one of all possible implementation methods.
Preferably, the device further comprises an optimization unit, wherein the optimization unit is configured to execute the step of the optimization method based on the passenger flow simulation scheduling model in any one of all possible implementation methods.
It should be noted that, when the device for a scheduling model based on passenger flow simulation provided in the foregoing embodiment executes a method for designing a scheduling model based on passenger flow simulation or a method for optimizing a scheduling model based on passenger flow simulation, only the division of the functional modules is illustrated, and in practical applications, the function distribution may be completed by different functional modules as needed, that is, the internal structure of the device may be divided into different functional modules to complete all or part of the functions described above. In addition, the passenger flow simulation-based scheduling model device provided by the embodiment and the passenger flow simulation-based scheduling model design and optimization method embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment and is not described herein again.
The invention has been described above by way of example, it is obvious that the specific implementation of the invention is not limited by the above-described manner, and that various insubstantial modifications are possible using the method concepts and technical solutions of the invention; or directly apply the conception and the technical scheme of the invention to other occasions without improvement and equivalent replacement, and the invention is within the protection scope of the invention.

Claims (10)

1. A design method of a scheduling model based on passenger flow simulation is characterized by comprising the following steps:
step 1, setting scheduling control parameters: determining the earliest departure time of the first shift of the line, taking the earliest departure time as the first shift for departure, and recording the earliest departure time as a1(ii) a Minimum value r of rest time of driverminAnd maximum value rmaxThe maximum number of people N in the platformmax(ii) a Minimum value of departure interval dminAnd maximum value dmaxThe optimum load factor is hop；
Step 2, predicting the number of people in each station every x minutes according to the passenger flow of the stations corresponding to the same type of dates in the history; predicting the number of passengers in the vehicle after departure at each stop according to the number of passengers getting on or off at each shift corresponding to the same type of dates in a plurality of historical days, wherein the actual load rate is the number of passengers in the vehicle/the rated nuclear load of the bus;
step 3, determining departure timetable and driver allocation rule according to limiting conditions
Note a1The next equidirectional shift of (A)3，a2The next equidirectional shift of (A)4Then go upward for shift a3The dispatching rule of the driver is that if the driver d1Execution of a1Post-execution of a2Is less than a minimum value rminThe driver is rescheduled, otherwise the driver d1Continue to execute a2(ii) a Execution of a2Is recorded as d by the driver2；
If a1The real load rate reaches the optimal real load rate hopWhen the departure interval is greater than or equal to the minimum value dminWhen the next shift is sent, the next shift a3If the driver d2Execution of a2Then, a is continuously executed3Is less than a minimum value rminIf the driver is not the driver who arrives first in the opposite direction, a new driver is rescheduled;
if a1The real load rate of the power grid does not reach the optimal real load rate hopWhen the number of people in the station is more than or equal to the maximum value N, the next vehicle a is sent3If the driver d2Execution of a2Then, a is continuously executed3Is less than a minimum value rminIf the driver is not the driver who arrives first in the opposite direction, a new driver is rescheduled;
if a1The real load rate of the power grid does not reach the optimal real load rate hopAnd when the number of people in the station does not reach the maximum value N, the departure interval is more than or equal to the maximum value dmaxIf so, the next vehicle a is sent at the moment3If the driver d2Execution of a2Then, a is continuously executed3Is less than a minimum value rminIf the driver is not the driver who arrives first in the opposite direction, a new driver is rescheduled;
determining the downlink shift a in the same manner4The departure driver allocation rule of (1);
and sequentially determining departure and driver allocation rules of all the shifts according to the method.
2. The design method of the scheduling model based on the passenger flow simulation as claimed in claim 1, wherein the step 2 predicts the number of people in each station per x minutes according to the passenger flow of the stations corresponding to the same type of dates in the historical multiple days as follows: obtaining a high-flat peak period according to a Fisher optimal segmentation method: segmenting passenger flow peak height and peak height according to the passenger flow data corresponding to all historical stations, and designing a peak height time period generation model: note all every T of site1Time-granular in-station traffic of { X }1,X2,…,XNIn Fisher's optimal segmentation method, the total variance is
Figure FDA0002491861940000021
Supposing that the passenger flow data in the all-day station is divided into K groups according to time intervals, namely the classification number is K, then
Figure FDA0002491861940000022
Wherein
Figure FDA0002491861940000023
Is the number j of the passenger flow in the stationThe value of the one or more of the one,
Figure FDA0002491861940000024
is the mean of the passenger flow in the whole day station, wherein
Figure FDA0002491861940000025
Is the sum of squared deviations in the group,
Figure FDA0002491861940000026
is the sum of squared deviations between the groups,
definition of
Figure FDA0002491861940000027
DKThe classification number is K, namely when the deviation square sum in the group is minimum, the corresponding segmentation method is the optimal segmentation method;
traversing the classification number K from 1 to n, n ∈ [9, 15 ]]For each value of K, the corresponding D is calculatedKDecay rate based on minimizing sum of squared deviations within a group
Figure FDA0002491861940000028
Take alphaKThe maximum corresponding classification number K is the optimal classification number, and a time interval optimal division method is obtained according to the optimal classification number;
obtaining historical passenger flow rules in different time periods according to the time period optimal division method to obtain the actual passenger flow quantity in each time period, and recording the passenger flow quantity in the h time period as fh；
Predicting the number of passengers in each station per x minutes according to the Poisson distribution, namely the probability function corresponding to the number k of passengers arriving at the station in x minutes (x is 1-30min) is as follows:
Figure FDA0002491861940000029
the number of passengers in the station at x minutes in the time period is obtained as follows: argmax P (X ═ k);
where lambda is the average incidence of passenger arrivals per unit of time,
Figure FDA00024918619400000210
wherein f ishIs the number of passengers arriving at the station in time period h, thIs the time span of period h.
3. The method for designing the scheduling model based on the passenger flow simulation as claimed in any one of claims 1 or 2, wherein the step 2 predicts the number of passengers in the vehicle after departure of each station according to the number of passengers getting on or off the corresponding shift on the same type of dates in a plurality of historical days, and specifically comprises the following steps: predicting all every T intervals according to the number of passengers on and off each site in history2Time particle size (preferably, T)21-30 minutes), i.e., the predicted traffic at the ith stop is
Figure FDA00024918619400000211
Figure FDA00024918619400000212
Figure FDA00024918619400000213
Representing the time granularity corresponding to the historical value of the passenger flow; the same method can predict the passenger flow, and the number of passengers in the vehicle after departure at the ith station can be predicted according to the predicted number of passengers on the vehicle
Figure FDA00024918619400000214
4. The method as claimed in claim 3, wherein the predicted passenger flow at the ith station is
Figure FDA0002491861940000031
The calculation formula of (2) is replaced by:
Figure FDA0002491861940000032
wherein the content of the first and second substances,
Figure FDA0002491861940000033
the presentation time granularity corresponds to the historical value of the passenger flow,
Figure FDA0002491861940000034
represents the median of the plurality of history values, and n represents the number of history values equal to or greater than the median.
5. The method as claimed in claim 3, wherein the step 3 of determining the departure schedule and the constraint conditions of the driver configuration rules further comprises that the total number of the driver's work trips per day is less than a predetermined value, and a new driver is arranged if the number of the driver's work trips is greater than or equal to the predetermined value.
6. The method for designing a shift scheduling model based on passenger flow simulation according to any one of claims 4 or 5, wherein if the rest time occurs in the meal time period and the driver does not eat the meal in step 3, the minimum value of the rest time is dmin+ e, maximum value dmax+ e, e is the length of the meal.
7. A method for optimizing a scheduling model based on passenger flow simulation is characterized by comprising the following steps:
the method comprises the following steps of adopting Markov decision to carry out reinforcement learning on a driving plan obtained by a design method of a scheduling model based on passenger flow simulation, and intelligently generating an optimal driving plan, wherein the specific method comprises the following steps:
markov decision is taken as (S, A, T, r, S)0) Wherein, in the step (A),
s refers to the driving plan (state space),
a refers to adjusting (increasing, decreasing or not changing) the optimum real load rate hopAnd executing a baseA method (action space) for designing a scheduling model in passenger flow simulation;
t is the transition probability space
r is a function of the reward, r is,
Figure FDA0002491861940000035
Figure FDA0002491861940000036
alpha is weight value, K is number of segments, q is passenger unit time and other vehicle cost, HiIs the number of passengers in the i period, Δ tiIs the departure interval in the ith time period, p is the cost per kilometer of the bus, L is the one-way route mileage, TiIs the time span of the ith time period, C is the passenger fare, J is the number of drivers, tjThe corresponding working hours of the driver j, and D is the unit working hour wage of the driver;
the reinforcement learning process is as follows:
(1)s0is an initial state, specifically: setting the value range of the real load rate as [ hmin,hmax]In [ h ]min,hmax]Internal random generation of optimal real load rate hopBased on hopCalculating a reward function Q according to the driving plan obtained by the method for designing the scheduling model based on the passenger flow simulation0，
(2) In [ h ]min,hmax]Increasing or decreasing the optimum loading rate hopObtaining a new state s according to the method for designing the scheduling model based on the passenger flow simulation1Calculating a reward function r1＝Q1-Q0If r is1Increasing, then giving the action a0Greater probability distribution pi (a)0|s0) Otherwise, giving a smaller probability;
recording the motion track as(s)0,a0,r1,s1,a1,r2…), noting the accumulated reward for the track as
Figure FDA0002491861940000041
Figure FDA0002491861940000042
Note its probability distribution as tauπ＝∏p(st+1|st,at)π(at|st) Wherein p(s)t+1|st,at) For transition probabilities, at which time it is desirable to maximize the reward, i.e.
Figure FDA0002491861940000043
Wherein the discount coefficient gamma belongs to [0,1) ]tIs the t power of gamma and decreases with the increase of t;
(3) setting the maximum iteration times, repeating the steps (1) to (2), recording reward values corresponding to all action tracks, recording the maximum reward as V, and recording the corresponding strategy as the optimal strategy as tau*The corresponding optimum state is s*I.e. an optimal driving plan.
8. The method as claimed in claim 7, wherein for other new lines, if the difference or fluctuation between the traffic law of the new line and the traffic law of the optimized line is within a given threshold, the new line can directly apply the policy τ*And obtaining the optimal driving plan.
9. A device of a scheduling model based on passenger flow simulation is characterized by comprising a parameter setting unit, a passenger flow prediction unit and a scheduling unit, wherein the units are electrically connected in sequence;
the parameter setting unit is used for the step 1 of the passenger flow simulation-based scheduling model design method of any one of claims 1 to 6;
the passenger flow prediction unit is used for executing the step 2 of the passenger flow simulation-based shift scheduling model design method of any one of claims 1 to 6;
the scheduling unit is used for executing the step of step 3 of the method for designing the scheduling model based on the passenger flow simulation as claimed in any one of claims 1 to 6.
10. The passenger flow simulation-based scheduling model device according to claim 9, further comprising an optimization unit for performing the steps of the passenger flow simulation-based scheduling model optimization method according to any one of claims 7-8.
CN202010407472.4A 2020-05-14 2020-05-14 Design method, optimization method and device of scheduling model based on passenger flow simulation Active CN113742879B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010407472.4A CN113742879B (en) 2020-05-14 2020-05-14 Design method, optimization method and device of scheduling model based on passenger flow simulation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010407472.4A CN113742879B (en) 2020-05-14 2020-05-14 Design method, optimization method and device of scheduling model based on passenger flow simulation

Publications (2)

Publication Number Publication Date
CN113742879A true CN113742879A (en) 2021-12-03
CN113742879B CN113742879B (en) 2024-05-03

Family

ID=78723415

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010407472.4A Active CN113742879B (en) 2020-05-14 2020-05-14 Design method, optimization method and device of scheduling model based on passenger flow simulation

Country Status (1)

Country Link
CN (1) CN113742879B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005212641A (en) * 2004-01-30 2005-08-11 Mitsubishi Electric Corp Station congestion degree estimation system
CN109034566A (en) * 2018-07-11 2018-12-18 南京行者易智能交通科技有限公司 A kind of intelligent dispatching method and device based on passenger flow above and below bus station
CN109753694A (en) * 2018-12-13 2019-05-14 东南大学 Small and medium-sized cities Transit Network Design method based on overall process trip detecting period

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005212641A (en) * 2004-01-30 2005-08-11 Mitsubishi Electric Corp Station congestion degree estimation system
CN109034566A (en) * 2018-07-11 2018-12-18 南京行者易智能交通科技有限公司 A kind of intelligent dispatching method and device based on passenger flow above and below bus station
CN109753694A (en) * 2018-12-13 2019-05-14 东南大学 Small and medium-sized cities Transit Network Design method based on overall process trip detecting period

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
刘继国;: "基于遗传算法的公交排班系统研究", 控制与信息技术, no. 06, 5 December 2019 (2019-12-05) *

Also Published As

Publication number Publication date
CN113742879B (en) 2024-05-03

Similar Documents

Publication Publication Date Title
Anderson Subways, strikes, and slowdowns: The impacts of public transit on traffic congestion
CN109034566B (en) A kind of intelligent dispatching method and device based on passenger flow above and below bus station
Shang et al. Bus timetabling considering passenger satisfaction: An empirical study in Beijing
CN101572011B (en) System and method for intelligently dispatching and managing urban public transports
CN111160686B (en) Design method and device of driving plan generation model
CN107564270A (en) A kind of intelligent public transportation dispatching method for running
CN111325483B (en) Electric bus scheduling method based on battery capacity prediction
CN107704950A (en) A kind of city rail train figure optimization method based on trip requirements and energy saving of system
CN113222387A (en) Multi-objective scheduling and collaborative optimization method for hydrogen fuel vehicle
CN107274665A (en) Bus transport capacity resource method and system for planning
CN107679654A (en) A kind of scale of parking Prediction Control system and implementation method
Pencheva et al. Evaluation of passenger waiting time in public transport by using the Monte Carlo method
CN110033143A (en) A kind of intelligent multichannel line programming dispatching and management method and system
Bie et al. Bus scheduling of overlapping routes with multi-vehicle types based on passenger OD data
CN111882097A (en) Vehicle maintenance personnel scheduling method and system
CN110211379A (en) A kind of public transport method for optimizing scheduling based on machine learning
CN110598971B (en) Responsive bus service planning method based on ant colony algorithm
CN111445091B (en) Intelligent decision-making method for taxi driver in passenger transport center
CN113742879A (en) Method and device for designing and optimizing scheduling model based on passenger flow simulation
CN114331231B (en) Method and system for scheduling public transport
CN111785015A (en) Public transport real-time regulation and control information system and scheduling method based on genetic algorithm
CN106682759A (en) Battery supply system for electric taxi, and network optimization method
CN114492951A (en) Resource allocation method, device and equipment based on passenger flow prediction
CN107578120A (en) It is a kind of that utility car share-car dispatching method is done based on electronic map
CN106251025A (en) A kind of Power Material distribution route commending system and method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant