CN115017720A - Travel activity chain generation method based on nested dynamic discrete selection - Google Patents
Travel activity chain generation method based on nested dynamic discrete selection Download PDFInfo
- Publication number
- CN115017720A CN115017720A CN202210711903.5A CN202210711903A CN115017720A CN 115017720 A CN115017720 A CN 115017720A CN 202210711903 A CN202210711903 A CN 202210711903A CN 115017720 A CN115017720 A CN 115017720A
- Authority
- CN
- China
- Prior art keywords
- selection
- travel
- traveler
- utility function
- stage
- 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.)
- Pending
Links
- 230000000694 effects Effects 0.000 title claims abstract description 101
- 238000000034 method Methods 0.000 title claims abstract description 23
- 101100100125 Mus musculus Traip gene Proteins 0.000 claims abstract description 39
- 239000011159 matrix material Substances 0.000 claims abstract description 6
- 230000007704 transition Effects 0.000 claims abstract description 6
- 239000000126 substance Substances 0.000 claims description 4
- 230000009467 reduction Effects 0.000 claims description 3
- 230000001186 cumulative effect Effects 0.000 claims description 2
- 238000004088 simulation Methods 0.000 description 15
- 238000004364 calculation method Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 230000000875 corresponding effect Effects 0.000 description 5
- 230000000052 comparative effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000006399 behavior Effects 0.000 description 3
- 238000010923 batch production Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000004931 aggregating effect Effects 0.000 description 1
- 230000032823 cell division Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
Images
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Databases & Information Systems (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Computer Hardware Design (AREA)
- Entrepreneurship & Innovation (AREA)
- Evolutionary Computation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Data Mining & Analysis (AREA)
- Geometry (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a travel activity chain generation method based on nested dynamic discrete selection, which comprises the following steps: s1, defining parameters of the dynamic discrete selection model, and constructing an all-day trip activity chain model of a traveler; s2, defining a state transition matrix, and establishing a relation between the state-selection pairs of the continuous stages; s3, on the premise of rational traveler and utility maximization decision criteria, deducing an expected utility function of the traveler in each stage; s4, defining the distribution of the disturbance items to obtain a closed-form nested dynamic discrete selection model; s5, defining an instant utility function according to the selection characteristics, and calculating an option specific utility function; (6) and (5) estimating parameters of various instant utility functions in the step (5), and generating a travel activity chain. According to the invention, disturbance item distribution based on a nested selection model is introduced, selection correlation of travel modes is carved, and a complete travel activity chain of a traveler can be predicted better.
Description
Technical Field
The invention relates to a travel activity chain generation method, in particular to a travel activity chain generation method based on nested dynamic discrete selection.
Background
At present, the social economy and urbanization of China are continuously and rapidly developed, the urban scale is continuously enlarged, the population is rapidly gathered, the automobile holding capacity is rapidly increased, the third industry mainly for entertainment and service accounts for nearly 77 percent, and the social and economic activities of people are more diversified and richer than the activities mainly for commuting. The travel is derived demand generated by activities, the activity demand of residents is reasonably and accurately predicted in limited urban space and main activity time, the urban traffic congestion problem is solved, accurate traffic management means are adopted, the travel activities of the residents are effectively induced, and the necessary premise of optimizing the urban traffic supply and demand structure is optimized. All of the above provide a serious challenge for urban traffic demand prediction.
The traditional traffic demand model mainly adopts a four-stage method, the OD thinking mode has insufficient response capability to urban space construction environment, space relation, space activity, space network and space policy, and only the aggregate utility generated by a traveler can be considered, so that the traffic demand model has the problems of single analysis angle and poor accuracy. The traffic demand model based on activities takes a travel activity chain as an analysis unit, takes each traveler as an analysis object, and belongs to a non-centralized travel demand prediction model. The model considers that the traveler arranges activities according to self intention and connects the activities with the traveler to form a chain, thereby making up the defects of the traditional model in the modeling principle.
The existing traffic demand prediction models based on activities are mainly divided into models based on constraint conditions, models based on discrete selection, models based on computational processes and models based on intelligent agents according to modeling methods. The general problem of the current research is that the time, the place, the purpose and the travel mode of a single trip are taken as research objects, and the constraint and influence relationship between different trips is ignored, for example, if the place, the travel mode and even the existence of activities of the next activity are influenced by the long duration of the previous activity; the possible activity going combinations in the future will influence the activity decision (activity duration, etc.) of the current stage traveler. In addition, the decision related to the travel activities belongs to a multi-dimensional selection problem, and often has certain correlation, so that the selection behaviors of travelers can be reflected wrongly by neglecting the correlation of the travel activities in time and space, and further lower demand prediction precision is caused.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to provide a travel activity chain generation method based on nested dynamic discrete selection, which can predict the complete travel activity chain of a traveler.
The technical scheme is as follows: the method for generating the travel activity chain comprises the following steps:
s1, defining parameters of the dynamic discrete selection model according to the properties of the Markov chain, and constructing an all-day trip activity chain model of the traveler;
s2, defining a state transition matrix, and establishing a relation between the state-selection pairs of the continuous stages;
s3, on the premise of rational traveler and utility maximization decision criteria, deducing an expected utility function of the traveler in each stage;
s4, defining the distribution of the disturbance items to obtain a closed-form nested dynamic discrete selection model;
s5, defining an instant utility function according to the selection characteristics of the trip purpose, the trip location and the trip mode, and calculating an option specific utility function;
and S6, estimating parameters of various instant utility functions in the step S5 by using known data, generating a trip activity chain of the traveler according to the obtained expected utility function, and verifying the aggregate result of activity type time arrangement, trip mode selection and place selection by using the observation data.
Further, in step S1, the parameters of the dynamic discrete selection model include a time interval t, a current location l, a current activity type r, a historical activity record h, a historical travel mode record c, and a time period τ during which current activities are performed cumulatively; for the state variable x of the traveler at each stage k k Then x k =(t k ,l k ,r k ,h k ,c k ,τ k );
The traveler needs to make some travel activity choices at each stage k to form a whole-day travel activity chain: t 'represents two choices of stay or trip, r' represents trip purpose choice, l 'represents trip place choice, and m' represents trip mode choice;
r 'if traveler chooses to stay' k =r k ,l′ k =l k ,m′ k =m stay ;
If the traveler selects to go, r' k Is the next activity to be performed,/' k Is the location of the next campaign, m' k Representing a trip mode;
if with a k Indicating a selection made at stage k, then a k =(t′ k ,r′ k ,l′ k ,m′ k ) (ii) a Let the selection set be influenced by state variables and be denoted C kj (x k ) Then a is k ∈C kj (x k );
The socioeconomic attribute and the travel attribute set of the traveler are represented by y, including a family attribute H, a personal attribute I, a travel distance TD, a travel time TT, and a travel cost TC.
Further, in step S2, the position and activity type of the k +1 th stage correspond to the selection of the k-th stage, and the link between the state-selection pairs of the successive stages is represented as: f (l) k+1 |x k ,a k )=l′ k And f (r) k+1 |x k ,a k )=r′ k ;
the two time variables t and tau need to be updated respectively according to the stay or trip selection, when stay, t and tau both go with the time, then f (tau) k+1 |x k ,a k )=τ k +1 and f (t) k+1 |x k ,a k )=t k +1;
When the trip is selected, the accumulated time period variable is zeroed, f (τ) k+1 |x k ,a k ) When 0, the time period variable needs to be added with the travel time.
Further, in the step S3, the value of ε k To represent a perturbation term, then ε k =(ε k (a k1 ),ε k (a k2 ),…,ε k (a kS ) In a batch process), wherein,when indicated at stage k, corresponds to option a kj S represents the number of options at this stage;
setting up a stage utility function V containing disturbance term when a traveler selects a certain option at each stage k (x k ,y,ε k ) The expression of (a) is: traveler's current immediate utility function u k (x k ,y,a k ) The sum of the reduced future utility function and the perturbation term;
then a utility function V 'is expected' k (x k Y) is:
wherein, V k Representing a stage utility function V k (x k ,y,ε k ),A desired function representing a perturbation term.
Further, in step S4, four nests are defined to cover all the selections, denoted by j;
setting to classify the belonged nests according to the travel modes in the options, including staying nest j stay Car nest j auto Public transport nest j transit And a non-motorized nest j non-motorized Where the set of four nests is denoted by Ω, then j ∈ Ω, and the selection set in nest j is C kj (x k ),a k ∈C kj (x k );
Option a for at nest j k In other words, the perturbation term is split according to the following formula:
ε k (a k )=ε kj +ε kj (a k )
wherein epsilon kj Option a for all in-nest j k Are said to be equal, epsilon kj (a k ) Is an option-specific perturbation term and is specific to all a k ∈C kj (x k ) To say,. epsilon kj (a k ) Independent from each other and obey the same Gumbel distribution;
for expected utility function V' k (x k Y), the maximum term in nest j is first selected and then compared between different nests, then:
according to Gumbel distribution properties, we obtain:
wherein the content of the first and second substances,
v k (a k ) The utility function which represents the option specificity is obtained by adding the instant utility function and the reduced future utility function; θ ∈ (0, 1) represents a reduction factor for future utility.
Further, in the step S5, the immediate utility function u of the stage k k (x k ,y,a k ) Is composed of four functions including stay selection utilityPurpose selection utilityLocation selection utilityAnd travel mode selection utility
finally, the option-specific utility function is calculated starting from the last phase K, for all a K ,v K (a K )=u K (a K ) And calculating an expected utility function V' k (x k Y); for stage K ∈ { K-1, K-2, …, 2, 1}, u is calculated k (a k ) And matching to possible state variables of the next stage according to the state transition matrix, thereby selecting a specific utility functionUntil the first stage.
Compared with the prior art, the invention has the following remarkable effects:
1. a dynamic discrete selection model based on a Markov process is introduced to consider prospective behaviors of travelers in travel activity related selection, decisions such as travel modes, travel purposes, travel time and the like at different time intervals in one day are associated through a complete model framework, and the defects of redundancy, disorder and difficulty in tandem of an existing travel activity chain generation model are overcome.
2. Disturbance item distribution based on a nested selection model is introduced, and selection correlation of a travel mode can be described, so that a complete travel activity chain of a traveler can be well predicted, and the requirement of a relevant traffic department for refined traffic demand management is met.
Drawings
FIG. 1 is a general flow chart of the present invention;
FIG. 2 is a diagram of analyzing zone traffic cell divisions;
FIG. 3(a) a graph of comparative observation and simulation plots of departure times of activities on the shift,
FIG. 3(b) is a graph showing the comparison of observation and simulation of the departure time of the school trip activity,
FIG. 3(c) is a graph of comparative observation and simulation plots of departure times of pickup activities,
FIG. 3(d) a graph of the comparison of observation and simulation of the departure time of a shopping trip activity,
FIG. 3(e) a graph of comparative observation and simulation of the departure time of the outgoing party travel activity,
FIG. 3(f) a graph of the comparison of observation and simulation at the departure time of a social travel activity,
FIG. 3(g) is a diagram of a comparative observation and simulation distribution of the departure time of the home-trip activity,
FIG. 3(h) a graph of the comparison of observation and simulation for the departure times of other entertainment travel activities;
FIG. 4 is a graph showing a comparison result of the allocation rate of the trip mode;
figure 5(a) observations selected for various traffic cell locations,
figure 5(b) is a simulation result for individual traffic cell site selection,
fig. 5(c) shows the absolute values of the simulation results and observation results selected for each traffic cell point.
Detailed Description
The invention is described in further detail below with reference to the drawings and the detailed description.
As shown in fig. 1, the method for generating a travel activity chain of the present invention includes the following steps:
and S1, abstracting the whole day trip activity chain of the traveler into a mathematical modeling problem, and defining the related concepts of the following dynamic discrete selection model by using the properties of the Markov chain.
A day is discretized into specific time periods, with 15 minutes as time intervals, denoted by t, and then divided into 96 analysis units. The current position is denoted by l. The current activity type is denoted by r and includes work, school, pick-up, shopping, dining, social, home-returning and other entertainment activities. The historical activity record is represented by h, and the historical travel pattern record is represented by c. The period during which the current activity is accumulated is denoted by τ. All the above properties are state variables of the traveler at each stage k, with x k Is shown, then x k =(t k ,l k ,r k ,h k ,c k ,τ k )。
The traveler needs to make some travel activity selections at each stage k to form a whole-day travel activity chain. First the traveler needs to decide whether to continue with the current activity or to end the current activity and perform the next activity by going, with t' representing the two choices of stay or go. Further, r ' represents travel purpose selection, l ' represents travel place selection, and m ' represents travel mode selection. R 'if traveler chooses to stay' k =r k ,l′ k =l k ,m′ k =m stay R 'if the traveler selects to go' k Is the next activity to be performed,/' k Is the location of the next campaign, m' k Representing travel modes including walking, riding, car riding, bus riding and subway. If with a k Indicating a selection made at stage k, then a k =(t′ k ,r′ k ,l′ k ,m′ k ). Assume that the selection set is influenced by a state variable and is denoted C kj (x k ) I.e. a k ∈C kj (x k )。
And y represents a socioeconomic attribute and a travel attribute set of the traveler, including a family attribute H, a personal attribute I, a travel distance TD, a travel time TT and a travel cost TC.
S2, defining a state transition matrix, linking the state-selection pairs of the successive stages by f (x) k+1 |x k ,a k ) And (4) showing. The position and activity type of the (k + 1) th stage corresponds to the selection of the previous stage, i.e. f (l) k+1 |x k ,a k )=l′ k And f (r) k+1 |x k ,a k )=r′ k . Travel activity and travel pattern history require the storage of non-redundant information, so both are updated only when the traveler chooses to travel, i.e. when t' k In order to go out of the journey,and wherein when y does not belong to x,equal to x ≧ y; if not, then the mobile terminal can be switched to the normal mode,equal to x. The two time variables t and τ need to be updated separately according to the stay or trip options, when stay, both t and τ are over time, i.e., f (τ) k+1 |x k ,a k )=τ k +1 and f (t) k+1 |x k ,a k )=t k + 1; when the trip is selected, the accumulated time period variable is zeroed, f (τ) k+1 |x k ,a k ) At 0, the session variable needs to be added with the travel time, which is obtained by the hadamard open platform API and needs to be converted to the number of intervals at 15 minute intervals.
And S3, assuming that the traveler is rational, and all the selection processes adopt a utility maximization decision criterion to deduce an expected utility function of the traveler in each stage.
Because the state variable x and the attribute variable y cannot reflect all influence factors such as weather, mood, trip preference and the like faced by a traveler in decision making to a certain extent, the use of epsilon k To represent the above-mentioned non-observable influencing factor, i.e. the perturbation term. Since the non-observable factors are related to different options, ε k =(ε k (a k1 ),ε k (a k2 ),…,ε k (a kS ) In a batch process), wherein,when indicated at stage k, corresponds to option a kj S represents the number of options at this stage. The dynamic discrete selection model assumes that a traveler selects an option at each stage to maximize the current immediate utility function u of the traveler k (x k ,y,a k ) The sum of the reduced future utility function and the perturbation term, the result is called the stage utility function containing the perturbation term, V k (x k ,y,ε k ) And (4) showing. From the above definitions, the following equation (in order to make the symbolic representation clearer, u is used) k (a k ) Represents u k (x k ,y,a k ) And V k Represents V k (x k ,y,ε k ),ε k+1 =(ε k+1,1 ,…ε k+1,S )):
Where θ ∈ (0, 1) denotes the reduction factor for future utility, and E denotes the expectation function. Due to the existence of the random disturbance term, the calculation of the stage utility function is relatively complex, a numerical simulation method is needed, and a large amount of time is consumed. To solve the problem, the disturbance term needs to be integrated to obtain the expected utility function V' k (x k Y) and is represented by V 'in the formula' k Is shown, i.e.
And S4, defining the distribution of the disturbance terms, obtaining a nested dynamic discrete selection model with a closed form, simplifying the calculation process, and further deducing a calculation formula of an expected utility function and a conditional selection probability.
First four nests are defined to cover all the choices, denoted by j. Suppose that the home cell is classified according to the travel mode in the option, including the stay cell j stay Nest j of small car auto Public transport nest j transit And a non-motorized nest j non-motorized . The set of four nests is denoted by Ω, i.e., j ∈ Ω, and the selection set in nest j is C kj (x k ) I.e. a k ∈C kj (x k )。
Options for at nest ja k In other words, the perturbation term can be split according to the following formula:
ε k (a k )=ε kj +ε kj (a k ) (3)
wherein epsilon kj Option a for all in-nest j k Are said to be equal, epsilon kj (a k ) Is an option-specific perturbation term and is specific to all a k ∈C kj (x k ) To say, epsilon kj (a k ) Independent of each other and obey the same Gumbel distribution, and the distribution scale parameter is rho j And the position parameter is 0. For the maximized function in the expected utility function formula, it is split into two parts, first selecting the largest term in nest j, and then comparing between different nests. Can be written as:
because for all a k ∈C kj (x k ),ε kj (a k ) Independent of each other and subject to the same Gumbel distribution, so that, according to the nature of this distribution, it is possible to obtain:
wherein the content of the first and second substances,
v k (a k ) Is in the complete form v k (x k ,y,a k ) The utility function which indicates the option is specific is obtained by adding the instant utility function and the reduced future utility function; e denotes the natural constantNumber, rho j Representing the Gumbel distribution scale parameter specific to nest j. According to the linear transfer property of the Gumbel distribution (i.e. if the random variable A-Gumbel (η, μ), then A + b-Gumbel (η + b, μ), b represents any real number), one can define:
wherein the content of the first and second substances,
ε′ kj ~Gumbel(0,ρ j )
thus, the desired utility function can be reduced to
Suppose the perturbation term ε of all nests j kj Independent, aggregate disturbance termsIndependent of each other and obey the same Gumbel distribution, with a scale parameter of 1 and a position parameter of 0. Further derived from the nature of the distribution:
where γ represents the euler constant. According to the splitting form of the maximization function, a conditional selection probability formula, namely the probability P (j | Ω, x) of selecting the nest j in all the nests Ω can be vividly obtained k Y), multiplied by the choice of option a in nest j k Probability P (a) of k |j,x k Y). Combining two independent identically distributed perturbation terms (i.e.And ε kj (a k ) ) the resulting closed logit calculation form, the conditional choice probability formula is:
j 'and a' k And represents j and a k The meanings are the same (for example, buses, subways and automobiles exist at present, and the probability of selecting the automobiles is calculated at present, a k Denotes an automobile a' k Representing buses, subways, and automobiles). The conditional selection probability can be directly passed throughAnd v k (a k ) And (4) calculating. The following needs to define an immediate utility function u of the option-specific utility functions k (x k ,y,a k ) Thus obtaining a complete calculation framework.
S5, defining an instant utility function according to the selection characteristics of the trip purpose, the trip location and the trip mode, and providing a process for calculating the option specific utility function through a reverse derivation process.
The immediate utility function of stage k is composed of four functions includingAndand selecting utility functions corresponding to the stay selection utility function, the target selection utility function, the place selection utility function and the travel mode respectively. When the traveler chooses to stop the traveler,otherwise
The dwell selection utility is influenced by the type of activity r currently being performed, the time variable t, and the cumulative period variable τ, i.e.
β end And beta duration A utility parameter representing an activity end time and an activity duration. Where r, τ, t represent dummy variables of corresponding meaning. The utility of travel purposes is influenced by the family attribute H, the personal attribute I, the historical activity record H and the time variable t, whereinAnd D h Representing a dummy variable, β, related to the family property H, the person property I and the historical activity record H H 、β I 、β h And beta t Utility parameters representing family attributes, personal attributes, historical activity records, and time variables. The expression is as follows:
the site utility of traveler n is given by:
it is related to the current location, the selected location and the selected activity type. TD ηl′l Indicating that the distance from the current location l to the selected location l' is in the part of the segment η, the linear segments comprising 0-1km, 1-2km, 2-5km, 5-15km and more than 15 km. Beta is a η A parameter representing a corresponding distance segment. Delta r′l′ Indicating the attractiveness of an activity r' at a location l, expressed in terms of the number of pois (points of interest) of the corresponding type. When delta r′l′ When equal to 0, ω r′l′ =1;δ r′l′ >0,ω r′l′ 0. M represents a large real number (set 999 in the present invention) to penalize locations that are not as attractive to the corresponding activity.
The effect of the travel pattern m' at nest j is the sum of the following two parts
TT m′ll′t And TC m′ll′t Represents the time and cost of travel m 'from the current location l to the selected location l' at time t, beta TT,m′ And beta TC,m′ A utility parameter representing Time (Travel Time) and Cost (Travel Cost). D c And D r′ Representing dummy variables related to the family attribute H, the personal attribute I, the historical travel pattern record c, and the new activity type r'. Beta is a H 、β I 、β c And the utility parameters represent family attributes, personal attributes and historical travel mode records.
Finally, computing the option-specific utility function requires starting from the last phase K for all a K ,v K (a K )=u K (a K ) And calculating an expected utility function V' k . For stage K ∈ { K-1, K-2, …, 2, 1}, u is calculated k (a k ) And matching to the possible state variables of the next stage according to the state transition matrix, so that the utility function expression specific to the option is as follows:
until the first stage.
And S6, estimating parameters of various instant utility functions in the step S5 by using known data, generating a trip activity chain of the traveler according to the obtained expected utility function, and verifying the aggregate result of activity type time arrangement, trip mode selection and place selection by using the observation data.
The present embodiment will be further explained by taking the data of the questionnaire for the travel of the residents in Chongqing city as an example. The data includes 158665 travel records for 30422 families, 72170 travelers, where the travel objectives include work, school, pickup, shopping, dining, social contact, home return and other entertainment activities, the travel modes are integrated as walking, riding, car, bus and subway, and the locations include 467 traffic districts divided by main cities of Chongqing, as shown in FIG. 2. And obtaining complete instant utility functions by using standard maximum likelihood estimation, and further calculating the specific utility function and the expected utility function of each stage option. And calculates each stage k, each state variable x k Next each option a k The probability is selected under the condition, the maximum option is selected as a simulation result, the calculation is carried out from the first stage to the last stage, and finally the whole day complete travel activity chain of all travelers under the given instant utility is obtained. And (4) comparing the counting result obtained by aggregating all the characteristics of the activity chain obtained by simulation, including activity type time arrangement, travel mode and place with the observed result. Fig. 3(a) - (h) show distribution diagrams of departure times of eight travel activities, fig. 4 shows comparison result diagrams of travel mode allocation rates, and fig. 5(a) - (c) show simulation results, observation results and absolute values of the simulation results and the observation results selected by each traffic cell location. In summary, the model based on the nested dynamic discrete selection has better performance in the aspect of travel activity chain generation because the correlation between prospective selection behavior and travel mode selection of a traveler can be captured.
Claims (6)
1. A method for generating a travel activity chain based on nested dynamic discrete selection is characterized by comprising the following steps:
s1, defining parameters of the dynamic discrete selection model according to the properties of the Markov chain, and constructing an all-day trip activity chain model of the traveler;
s2, defining a state transition matrix, and establishing a relation between the state-selection pairs of the continuous stages;
s3, on the premise of rational traveler and utility maximization decision criteria, deducing an expected utility function of the traveler in each stage;
s4, defining the distribution of the disturbance items to obtain a closed-form nested dynamic discrete selection model;
s5, defining an instant utility function according to the selection characteristics of the trip purpose, the trip location and the trip mode, and calculating an option specific utility function;
and S6, estimating parameters of various instant utility functions in the step S5 by using known data, generating a trip activity chain of the traveler according to the obtained expected utility function, and verifying the aggregate result of activity type time arrangement, trip mode selection and place selection by using the observation data.
2. A method for generating a chain of travel activities based on nested dynamic discrete selection according to claim 1, wherein in step S1, the parameters of the dynamic discrete selection model include time interval t, current location/, current activity type r, historical record of activity h, historical record of travel patterns c, and time interval τ for cumulative execution of current activities; for the state variable x of the traveler at each stage k k Then x k =(t k ,l k ,r k ,h k ,c k ,τ k );
The traveler needs to make some travel activity choices at each stage k to form a whole-day travel activity chain: t 'represents two choices of stay or trip, r' represents trip purpose choice, l 'represents trip place choice, and m' represents trip mode choice;
r 'if traveler chooses to stay' k =r k ,l′ k =l k ,m′ k =m stay ;
R 'if the traveler selects to go' k Is the next activity to be performed,/' k Is the location of the next campaign, m' k Representing a trip mode;
if with a k Indicating a selection made at stage k, then a k =(t′ k ,r′ k ,l′ k ,m′ k ) (ii) a Let the selection set be influenced by state variables and be denoted C kj (x k ) Then a is a k ∈C kj (x k );
The socioeconomic attribute and the travel attribute set of the traveler are represented by y, including a family attribute H, a personal attribute I, a travel distance TD, a travel time TT, and a travel cost TC.
3. A method for generating a traveling activity chain based on nested dynamic discrete selection according to claim 2, wherein in said step S2, if the position and activity type of the (k + 1) th stage corresponds to the selection of the (k) th stage, the link between the state-selection pairs of the successive stages is expressed as: f (l) k+1 |x k ,a k )=l′ k And f (r) k+1 |x k ,a k )=r′ k ;
the two time variables t and tau need to be updated respectively according to the stay or trip selection, when stay, t and tau both go with the time, then f (tau) k+1 |x k ,a k )=τ k +1 and f (t) k+1 |x k ,a k )=t k +1;
When choosing to go outThe accumulated time period variable is zeroed, f (τ) k+1 |x k ,a k ) When 0, the time period variable needs to be added with the travel time.
4. A method for generating a travel activity chain based on nested dynamic discrete selection according to claim 2, wherein in step S3, the order of ∈ is used k To represent the perturbation term, then ε k =(ε k (a k1 ),ε k (a k2 ),…,ε k (a kS ) In which epsilon) k (a kj )When indicated at stage k, corresponds to option a kj S represents the number of options at this stage;
setting up a stage utility function V containing disturbance term when a traveler selects a certain option at each stage k (x k ,y,ε k ) The expression of (a) is: traveler's current immediate utility function u k (x k ,y,a k ) The sum of the reduced future utility function and the perturbation term;
then a utility function V 'is expected' k (x k Y) is:
5. A method for generating a travel activity chain based on nested dynamic discrete selection according to claim 2, wherein in said step S4, four nests are defined to cover all selections, denoted by j;
setting to classify the belonged nests according to the travel modes in the options, including the staying nest j stay Car nest j auto Public transport nest j transit And a non-motorized nest j non-motorized Where the set of four nests is denoted by Ω, then j ∈ Ω, and the selection set in nest j is C kj (x k ),a k ∈C kj (x k );
Option a for at nest j k In other words, the perturbation term is split according to the following formula:
ε k (a k )=ε kj +ε kj (a k )
wherein epsilon kj Option a for all in-nest j k Are said to be equal, epsilon kj (a k ) Is an option-specific perturbation term and is specific to all a k ∈C kj (x k ) To say, epsilon kj (a k ) Independent from each other and obey the same Gumbel distribution;
for expected utility function V' k (x k Y), the maximum term in nest j is first selected and then compared between different nests, then:
according to Gumbel distribution properties, we obtain:
wherein the content of the first and second substances,
v k (a k ) The utility function which represents the option specificity is obtained by adding the instant utility function and the reduced future utility function; θ ∈ (0, 1) represents a reduction factor for future utility.
6. A method for generating a travel activity chain based on nested dynamic discrete selection according to claim 2, wherein in step S5, the immediate utility function u of stage k is k (x k ,y,a k ) Is composed of four functions including stay selection utilityPurpose selection utilityLocation selection utilityAnd travel mode selection utility
finally, the option-specific utility function is calculated starting from the last phase K, for all a K ,v K (a K )=u K (a K ) And calculating an expected utility function V' k (x k Y); for phase K ∈ { K-1, K-2, …, 2, 1}, u is calculated k (a k ) Option specific utility functionUntil the first stage.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210711903.5A CN115017720A (en) | 2022-06-22 | 2022-06-22 | Travel activity chain generation method based on nested dynamic discrete selection |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210711903.5A CN115017720A (en) | 2022-06-22 | 2022-06-22 | Travel activity chain generation method based on nested dynamic discrete selection |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115017720A true CN115017720A (en) | 2022-09-06 |
Family
ID=83077093
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210711903.5A Pending CN115017720A (en) | 2022-06-22 | 2022-06-22 | Travel activity chain generation method based on nested dynamic discrete selection |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115017720A (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110161258A1 (en) * | 2009-12-29 | 2011-06-30 | Nikovski Daniel N | Method for Converting Dynamical Systems with Continuous States into Markov Decision Processes with Discrete States |
CN106951999A (en) * | 2017-03-29 | 2017-07-14 | 北京航空航天大学 | The modeling of a kind of travel modal and the moment Combination selection that sets out and analysis method |
CN113656746A (en) * | 2021-07-21 | 2021-11-16 | 东南大学 | Travel mode chain selection method considering group heterogeneity under dynamic structure |
-
2022
- 2022-06-22 CN CN202210711903.5A patent/CN115017720A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110161258A1 (en) * | 2009-12-29 | 2011-06-30 | Nikovski Daniel N | Method for Converting Dynamical Systems with Continuous States into Markov Decision Processes with Discrete States |
CN106951999A (en) * | 2017-03-29 | 2017-07-14 | 北京航空航天大学 | The modeling of a kind of travel modal and the moment Combination selection that sets out and analysis method |
CN113656746A (en) * | 2021-07-21 | 2021-11-16 | 东南大学 | Travel mode chain selection method considering group heterogeneity under dynamic structure |
Non-Patent Citations (1)
Title |
---|
刘玲玲;刘魏巍;董洁霜;: "基于交叉巢式logit模型的交通出行方式选择研究", 物流工程与管理, no. 06, 15 June 2017 (2017-06-15) * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Bi et al. | Daily tourism volume forecasting for tourist attractions | |
Sáez et al. | Hybrid adaptive predictive control for the multi-vehicle dynamic pick-up and delivery problem based on genetic algorithms and fuzzy clustering | |
Huang et al. | A static bike repositioning model in a hub-and-spoke network framework | |
Wang et al. | Collaborative multiple centers fresh logistics distribution network optimization with resource sharing and temperature control constraints | |
Lei et al. | Inferring temporal motifs for travel pattern analysis using large scale smart card data | |
CN115062873B (en) | Traffic travel mode prediction method and device, storage medium and electronic device | |
Qin et al. | Reinforcement learning for ridesharing: A survey | |
Zou et al. | Passenger flow prediction using smart card data from connected bus system based on interpretable xgboost | |
JPH10240799A (en) | Support device for new store opening evaluation | |
Wang et al. | Collaborative multidepot electric vehicle routing problem with time windows and shared charging stations | |
CN115545758A (en) | Method and system for self-adaptive incremental site selection of urban service facilities | |
Qu et al. | How many vehicles do we need? Fleet sizing for shared autonomous vehicles with ridesharing | |
CN110490365B (en) | Method for predicting network car booking order quantity based on multi-source data fusion | |
CN114066503A (en) | Method for forecasting taxi demand based on virtual service area construction | |
Zhang et al. | A dynamic shared bikes rebalancing method based on demand prediction | |
Xi et al. | Hmdrl: Hierarchical mixed deep reinforcement learning to balance vehicle supply and demand | |
CN113656746B (en) | Travel mode chain selection method considering group heterogeneity under dynamic structure | |
CN110459050A (en) | A kind of short-term bus passenger flow prediction technique based on hybrid decision tree | |
Nasiboglu | Dijkstra solution algorithm considering fuzzy accessibility degree for patch optimization problem | |
CN111723871B (en) | Estimation method for real-time carriage full load rate of bus | |
CN116681197B (en) | Logistics line searching method and system based on big data | |
CN115017720A (en) | Travel activity chain generation method based on nested dynamic discrete selection | |
Yu et al. | Combining travel behavior in metro passenger flow prediction: A smart explainable Stacking-Catboost algorithm | |
Daisy | Microsimulation of activity participation, tour complexity, and mode choice within an activity-based travel demand model system | |
Li et al. | Spatiotemporal charging demand models for electric vehicles considering user strategies |
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 |