CN103116702A - Bicycle-mode traveling selection forecasting method based on activity chain mode - Google Patents

Abstract

The invention discloses a travel prediction method based on an activity chain mode of choosing a bicycle mode. The prediction method includes the following steps: conducting a data survey on the travel situation of residents and sorting out the survey results; Select the mode, and perform variable virtualization and coding operations on the travel mode; input the relevant variables in the activity chain mode into the multi-logit model, and calculate the co-evolution logit model; perform iterative calculations on the calculated co-evolution logit model results, and Record the selection results of the two travel modes; conduct statistics and analysis on the selection results of the two travel modes, and compare and analyze the prediction accuracy. The invention accurately predicts the proportion of traveling by bicycle through the statistics and analysis of urban residents' choice of travel vehicles, and then provides scientific and reasonable guidance for urban traffic planning and policy formulation.

Description

A travel prediction method for choosing bicycle mode based on activity chain model

technical field

The invention relates to the technical field of traffic demand prediction and traffic planning, in particular to a travel prediction method based on an activity chain model for selecting a bicycle mode.

Background technique

At present, with the acceleration of my country's urbanization process, urban space expansion, urban population size and urban road motorization level have all changed accordingly, especially in the urban passenger transport system. In the rapid urbanization, the urban passenger transport system has gradually exposed a series of negative problems such as traffic congestion, energy consumption, and air pollution.

How to optimize the urban passenger transportation system is the key to alleviating urban traffic congestion and energy saving and emission reduction, and bicycles have the advantages of flexibility, no pollution, no energy consumption, and less road resource occupation in short-distance travel, so bicycles can be used as urban passenger transportation It is part of the system and solves a series of problems in the current urban passenger transport system. However, in the past ten years, the sharing rate of bicycles as a mode of transportation has dropped significantly, and on the contrary, the sharing rate of individual motorized transportation modes has increased rapidly.

Therefore, through a series of studies on the demand for bicycle transportation, the analysis of the differences in the intentions of travelers to choose bicycles and the causes will help predict the future development trend of bicycle transportation and help to formulate effective and targeted transportation strategies. Demand management policies to guide travelers to use bicycles reasonably. However, the existing research only considered the individual's preference for bicycle use in a single trip, and did not consider the influence of residents' daily activity patterns on the choice of bicycle travel. In the activity-based traffic demand forecasting theory, it is believed that people's travel mode choice is affected by the characteristics of travel activities, for example, people prefer to choose a more flexible travel mode in the multi-destination activity mode, and the travel mode is also affected by the preferred means of transportation. Impact.

Therefore, it is necessary to expand the research and analysis unit from "single trip" to the level of "activity-activity chain mode", and predict the demand for bicycle travel in cities from the perspective of the interaction between bicycle travel and activity chain mode travel. Provide scientific and reasonable guidance for the formulation of urban transportation planning and policies.

Therefore, based on the above-mentioned problems, the present invention provides a method for predicting trips based on the activity chain mode of selecting a bicycle mode.

Contents of the invention

Purpose of the invention: The present invention provides a method for predicting travel by bicycle based on the activity chain model, which can predict the demand for travel by bicycle in cities, and provide scientific and reasonable guidance for urban traffic planning and policy formulation.

Technical solution: The present invention provides a method for predicting travel based on the activity chain mode of selecting a bicycle mode, and the prediction method includes the following steps:

Step (1) Carry out a data survey on the travel situation of residents, organize and count the survey results.

Step (2) Extract the survey results of residents' one-day travel data, bicycle travel and activity chain patterns, and perform variable dummy and coding operations on them.

Step (3) Input the relevant variables in the activity chain mode into the multinomial logit model, input the relevant variables in the bicycle travel into the binomial logit model, and analyze the interaction between the bicycle travel and the activity chain mode, and calculate As a result of the model, a co-evolutionary logit model is obtained.

Step (4) Perform iterative calculation on the calculated co-evolutionary logit model results, and record the selection results of bicycle travel and activity chain mode. When the travel modes of all travelers are selected, the iterative calculation ends.

Step (5) Statistically and analyze the selection results of bicycle travel and activity chain mode, and compare and analyze the prediction accuracy.

In the step (1), conducting a data survey on the travel situation of the residents and sorting out and counting the survey results includes the following steps,

Step (1-1) Divide traffic districts, adopt random sampling household survey questionnaires, and distribute questionnaires according to the population ratio of the districts;

Steps (1-2) determine the analysis variables;

Steps (1-3) collect variable data.

From the steps (1-1) to (1-3), firstly define the main activities in the sample, the mode of activity chain and the representation method, and then use random sample home visits to conduct questionnaire surveys, distribute questionnaires according to the proportion of the population of the community, and the content includes travel Individual characteristics, family characteristics, travel activities on typical working days, and which mode of transportation to choose for travel. Individual attributes include traveler gender, age, and occupation, etc. Family characteristics include family structure, family income, etc. Travel attributes include land characteristics, travel distance, etc. wait.

The activity chain mode in the step (2) is divided into, according to the number of activities, a simple activity chain of one activity, a complex activity chain of two or more activities; according to the purpose of travel, it is divided into a survival activity chain for the purpose of travel, Non-survival activity chains for leisure and entertainment purposes and mixed activity chains with two travel purposes; variable virtual and coding operations in the activity chain mode, hwh is a simple survival activity chain without other stays, hwhwh is a simple survival activity chain, including travel and stay based on home, hoh is a chain of survival activities, no other stays, hohoh is a chain of non-survival activities, returning home multiple times, hwh+o is a chain of simple travel, including chains of non-survival activities, of which bicycle The mode of behavior, using bicycles and not using bicycles, and finally discarding the activity chain mode samples with a sample size of less than 1% of the survey data to obtain effective samples.

In the step (3), it is assumed that the traveler has two modes of bicycle travel (D1) and activity chain mode (D2), (Di∈D, i=1, 2), each of which contains several options and the items are mutually Independently, assuming that the option with the greatest utility is selected, based on the Logit model, the formula for calculating the probability of option d being selected is:

${\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; u</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \}</span><span\; class="notranslate">\; ]</span>}{{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; d</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; u</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \}</span><span\; class="notranslate">\; ]</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{<span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

Among them, Pt(d) is the probability of selecting option d at time t, E{Ut(d)} is the expected utility of option d at time t, and t is the iteration period;

a. There is an interaction between D1 and D2 in the bicycle travel mode and the activity chain mode, that is, the result of one choice affects the utility of the other option, and the calculation formulas of the two utility functions are:

$\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; u</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \}</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; r</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&beta;</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; +</span>\underset{{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; \&prime;</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&beta;</span>}}_{{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; \&prime;</span>}}\underset{\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}_{{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; \&prime;</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span>$

$<span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{<span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

Among them, Xr is the attribute r of the option d, Xr' is the attribute r' of the option d after the given state S, and S is the selection mode Dj, The state of , β is the parameter of the variable X;

b. Pts is the calculation formula for the occurrence probability of state S at time t:

${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; S</span>}}{\mathrm{<span\; class="notranslate">\; \&Pi;</span>}}{\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

c. The formula for calculating the initial probability of the reciprocal of the number of options in the selection mode Di is:

${\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{<span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; .</span>$

In the step (4), the effectiveness of each option item in each selection mode at time t is affected by the result of the selection mode at time t-1, and at the end of each iteration cycle, the uncertainty of each selection mode Di depends on the size of the entropy The calculation formula is:

${\mathrm{<span\; class="notranslate">\; h</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; log</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \}</span><span\; class="notranslate">\; ,</span><span\; class="notranslate">\; \&ForAll;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$

Among them, the adjustment coefficient θi is introduced into the entropy function to make the entropy values of each decision branch equal at the initial moment, and the selection mode item with a smaller entropy value θiHt(Di) has less uncertainty, so the selection mode item is determined first and then in subsequent iterations The process will not change.

Compared with prior art, the beneficial effect of the present invention is:

A kind of travel prediction method based on the activity chain mode of the selected bicycle mode of the present invention, the prediction method has increased the defect that the travel demand prediction of the selected bicycle mode does not consider the activity chain mode in the past, through the selection of urban residents' travel vehicles under the activity chain mode Statistics and analysis can accurately predict the proportion of travel by bicycle, and then provide scientific and reasonable guidance for urban transportation planning and policy formulation.

Description of drawings

Fig. 1 is the schematic flow chart of carrying out urban bicycle traffic demand prediction of the embodiment of the present invention;

Fig. 2a is a schematic diagram of the sequence analysis results of choosing a bicycle to travel in an embodiment of the present invention;

Fig. 2b is a schematic diagram of the analysis result of the travel sequence of the activity chain mode according to the embodiment of the present invention.

Detailed ways

Below in conjunction with specific embodiment, a kind of selection bicycle mode trip prediction method based on activity chain pattern of the present invention is described in detail:

As shown in Figure 1, a method for predicting travel based on the activity chain mode of selecting a bicycle mode, the method for predicting comprises the following steps:

Step (1) Carry out a data survey on the travel situation of residents, organize and count the survey results.

Step (2) Extract the data survey results of residents' daily travel, bicycle travel and activity chain mode, and perform variable virtualization and coding operations on them.

Step (3) Input the relevant variables in the activity chain mode into the multinomial logit model, input the relevant variables in the bicycle travel into the binomial logit model, and analyze the interaction between the bicycle travel and the activity chain mode, and calculate As a result of the model, a co-evolutionary logit model is obtained.

Step (4) Perform iterative calculation on the calculated co-evolutionary logit model results, and record the selection results of bicycle travel and activity chain mode. When the travel modes of all travelers are selected, the iterative calculation ends.

Step (5) Statistically and analyze the selection results of bicycle travel and activity chain mode, and compare and analyze the prediction accuracy.

In step (1), conduct a data survey on the travel situation of residents, sort out and calculate the survey results, including the following steps,

Step (1-1) Divide traffic districts, adopt random sampling household survey questionnaires, and distribute questionnaires according to the proportion of the population of the districts;

Steps (1-2) determine the analysis variables;

Steps (1-3) collect variable data.

Among them, from step (1-1) to step (1-3), firstly define the main activities in the sample, the mode of activity chain and the representation method, and then use random sample home visits to conduct questionnaire surveys, distribute questionnaires according to the proportion of the population of the community, and the content includes travel Individual characteristics, family characteristics, travel activities on typical working days, and which mode of transportation to choose for travel. Individual attributes include traveler gender, age, and occupation, etc. Family characteristics include family structure, family income, etc. Travel attributes include land characteristics, travel distance, etc. wait.

The activity chain mode in step (2) is divided into simple activity chains with one activity and complex activity chains with two or more activities according to the number of activities; according to the purpose of travel, it is divided into survival activity chains for travel and leisure activity chains. Non-survival activity chains for entertainment purposes and mixed activity chains with two travel purposes; variable virtual and coding operations in the activity chain mode, hwh is a simple survival activity chain without other stays, hwhwh is a simple survival activity chain, Including home-based travel and round-trip stays, hoh is a survival activity chain, no other stays, hohoh is a non-survival activity chain, returning home many times, hwh+o is a simple trip, including non-survival activity chains, in which the bicycle way out Behavior, using bicycles and not using bicycles, and finally discarding the activity chain model samples with a sample size of less than 1% in the survey data to obtain effective samples.

In step (3), it is assumed that the traveler has two modes of bicycle travel (D1) and activity chain mode (D2), (Di∈D, i=1, 2), each of which contains several options and is independent of each other. Assuming that the option with the greatest utility is selected, based on the Logit model, the formula for calculating the probability of option d being selected is:

${\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; u</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \}</span><span\; class="notranslate">\; ]</span>}{{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; d</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; u</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \}</span><span\; class="notranslate">\; ]</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{<span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

Among them, Pt(d) is the probability of selecting option d at time t, E{Ut(d)} is the expected utility of option d at time t, and t is the iteration period;

a. There is an interaction between D1 and D2 in the bicycle travel mode and the activity chain mode, that is, the result of one choice affects the utility of the other option, and the calculation formulas of the two utility functions are:

$\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; u</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \}</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; r</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&beta;</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; +</span>\underset{{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; \&prime;</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&beta;</span>}}_{{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; \&prime;</span>}}\underset{\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}_{{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; \&prime;</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; )</span>$

$<span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{<span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

的状态，β为变量X的参数；Among them, Xr is the attribute r of the option d, Xr' is the attribute r' of the option d after the given state S, and S is the selection mode Dj,

The state of , β is the parameter of the variable X;

b. Pts is the calculation formula for the occurrence probability of state S at time t:

${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; S</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; S</span>}}{\mathrm{<span\; class="notranslate">\; \&Pi;</span>}}{\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

c. The formula for calculating the initial probability of the reciprocal of the number of options in the selection mode Di is:

${\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{<span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; .</span>$

In step (4), the utility of each option item in each selection mode at time t is affected by the result of the selection mode at time t-1, and at the end of each iteration cycle, the uncertainty of each selection mode Di depends on the entropy calculation formula for:

${\mathrm{<span\; class="notranslate">\; h</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; log</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \}</span><span\; class="notranslate">\; ,</span><span\; class="notranslate">\; \&ForAll;</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$

Among them, the adjustment coefficient θi is introduced into the entropy function to make the entropy values of each decision branch equal at the initial moment, and the selection mode item with a smaller entropy value θiHt(Di) has less uncertainty, so the selection mode item is determined first and then in subsequent iterations The process will not change.

Example 1

Step (1) is used to predict the travel demand of urban residents in Bengbu City who choose bicycles. Bengbu City is located in the northern part of Anhui Province, with an urban area of 601.5km2 and a population of 914,300 at the end of 2006. Bengbu City is a typical group city. Including the central group, the northern group and the eastern group, it is divided into 98 traffic small groups. Random sample household questionnaire survey is used, and questionnaires are distributed according to the population ratio of the district. The content includes the individual characteristics of travelers, family characteristics, typical working day travel activities and methods. Selection, individual attributes mainly include the traveler’s gender, age, occupation, education level, etc., family characteristics mainly include family structure and size, means of transportation, family income, etc., travel attributes include land characteristics, travel distance, etc., and survey data analysis involves The descriptive statistics of the obtained variables are shown in Table 1.

Carry out step (2), extract the travel activity chain patterns of Bengbu residents, discard the activity chain samples whose sample size is less than 1% in the travel survey data of Bengbu City, and obtain 5632 effective samples, and carry out the variable dummy sum of the two travel modes Coding operation, hwh is a simple survival activity chain mode, no other stops, hwhwh is a simple survival activity chain mode, including home-based travel and round trip stays, hoh is a survival activity chain mode, no other stays, hohoh is a non-survival type Activity chain mode, returning home multiple times, hwh+o is a simple travel mode, including non-survival activity chain mode, using bicycles and not using bicycles, this embodiment only considers the above 5 types of activity chain modes and 2 types of bicycle usage, As shown in table 2,

It can be seen from Table 2 that 37.2% of travelers use bicycles in their daily activities, and the proportion of hwh-type activities is as high as 44.0%. Hohoh-type activities and hwh+h-type activities are less, accounting for 3.7% and 1.6% respectively. Patterns (hwh and hwhwh) accounted for 80% of the total, and complex activity chain patterns (hwhwh, hohoh, hwh+o) accounted for 40.8% of the total.

Carry out step (3), calculate the selection order between the bicycle use and the activity chain mode of each sample according to the logit model, 54 samples need 3 iterations to complete all selections, 4100 samples need 4 iterations, 1478 samples It takes 5 iterations, and an average of 4.25 iterations are needed to complete all selections, indicating that there is an interaction between bicycle use and activity patterns, and the decision-making process is complex. The decision-making sequence of individual travelers in different activity chain modes is shown in Table 3.

It can be seen from Table 3 that an average of 65% of travelers first determine the activity chain mode and then choose the travel mode, only 35% of the travelers first decide whether to use bicycles and then determine the activity chain mode. In the non-survival activity chain mode (hoh , hohoh), nearly half (45.7%) of the travelers first decide whether to use bicycles for travel, which is significantly higher than the survival activity chain mode.

As shown in Figure 2a, it can be seen that 76.33% of the bicycle users first determined the activity chain mode, and only 23.67% of the bicycle users first determined the travel mode, while among the non-bicycle users, the proportion of the first travel mode selection was 39.91% %, much higher than that of bicycle users.

As shown in Figure 2b, it can be seen that only 25.69% of the individuals in the group with priority choice of travel mode choose bicycle travel, and 74.31% of individuals do not choose bicycle. 42.55%, significantly higher than the group that chooses the preferred travel mode.

The prediction accuracy of the binomial Logit model for the use of bicycles by individual travelers after the interaction between the two is shown in Table 4.

From 73.8% to 81.8%, the prediction accuracy without using a bicycle improved from 78.5% to 86.0%. The overall prediction accuracy of the binomial Logit model increased from 76.7% to 84.4%.

The above is only a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements can also be made, and these improvements should also be regarded as the present invention. scope of protection.

Claims (6)

1. the selection bicycle mode based on activity chain pattern Forecasting Methodology of going on a journey, it is characterized in that: the method comprises the following steps,

Step (1) is carried out data survey and is arranged, counts investigation result the resident trip situation;

Step (2) is extracted the data survey result of resident's trip on the one, and the bicycle mode is gone on a journey and the activity chain pattern, and it is carried out the virtual and encoding operation of variable;

Step (3) inputs to the correlated variables in the activity chain pattern in multinomial logit model, correlated variables during the bicycle mode is gone on a journey inputs in binomial logit model, and the bicycle mode is gone on a journey and the activity chain pattern is analyzed after mutual, calculate model result, obtain coevolution logit model;

Step (4) is carried out interative computation to the coevolution logit model result that calculates, and records that the bicycle mode is gone on a journey and the selection result of activity chain pattern, when the trip mode of all travelers select complete after, the finishing iteration computing;

Step (5) respectively the bicycle mode is gone on a journey and the selection result of activity chain pattern is carried out statistics and analysis, and precision of prediction is analyzed.

2. a kind of selection bicycle mode based on the activity chain pattern according to claim 1 Forecasting Methodology of going on a journey is characterized in that:

In described step (1), the resident trip situation carried out data survey and arranged, count investigation result, comprising the following steps,

Step (1-1) is divided traffic zone, adopts random sampling visit to the parents of schoolchildren or young workers survey, provides questionnaire in residential quarter population ratio;

Step (1-2) is determined situational variables;

Step (1-3) gathers variable data.

3. a kind of selection bicycle mode based on the activity chain pattern according to claim 2 Forecasting Methodology of going on a journey is characterized in that:

Described step (1-1) is to step (1-3), at first define main activities, activity chain pattern and method for expressing in sample, then adopt the random sampling visit to the parents of schoolchildren or young workers to carry out survey, provide questionnaire in residential quarter population ratio, content comprises traveler personal feature, family's feature, the trip activity of exemplary operation day and selects which kind of mode of transportation trip, individual attribute comprises traveler sex, age and occupation, family's feature comprises family structure, family income, and the trip attribute comprises soil feature, trip distance.

4. a kind of selection bicycle mode based on the activity chain pattern according to claim 1 Forecasting Methodology of going on a journey is characterized in that:

Activity chain pattern in described step (2) is divided into by movable number, once movable simple activities chain, twice and above movable complicated activity chain; Be divided into the survival-type activity chain take trip as purpose, with the non-survival-type activity chain of amusement and recreation purpose with contain the mixed type activity chain of two kinds of trip purposes by the trip purpose; Virtual and the encoding operation of the variable of activity chain pattern, hwh is simple survival-type activity chain, without other stop, hwhwh is simple survival-type activity chain, comprises based on the round stop of the trip of family, hoh for survival the type activity chain, without other stop, hohoh is non-survival-type activity chain, repeatedly returns home, hwh+o is for simply going on a journey, comprise non-survival-type activity chain, wherein the bicycle mode is gone on a journey and is, use bicycle and do not use bicycle, cast out at last enquiry data kind sample size less than 1% activity chain sample, obtain effective sample.

5. a kind of selection bicycle mode based on the activity chain pattern according to claim 1 Forecasting Methodology of going on a journey is characterized in that:

The hypothesis traveler has the bicycle mode to go on a journey (D1) and two kinds of activity chain patterns (D2) in described step (3), (Di ∈ D, i=1,2), respectively comprise some options and every between separate, the options of supposing the effectiveness maximum is selected, and based on the Logit model, the computing formula of the selected probability of options d is:

$P^t\left(d\right)=\frac{\exp \left[E\right\{U^{t-1}\left(d\right)\left\}\right]}{{\mathrm{\&Sigma;}}_{d^{\&prime;}\&Element;D_i}\exp \left[E\right\{U^{t-1}\left(d^{\&prime;}\right)\left\}\right]},\&ForAll;d\&Element;D_i,{\&ForAll;D}_i\&Element;D---\left(1\right)$

Wherein, Pt (d) chooses the probability of options d, E{Ut (d) for moment t } be the expected utility of moment t options d, t is iteration cycle;

A, bicycle mode go on a journey and the activity chain pattern in have reciprocation between D1 and D2, namely a kind of selection result exerts an influence to the effectiveness of another kind of options, two kinds of utility function computing formula are:

$E\left\{U^t\left(d\right)\right\}=\underset{r}{\mathrm{\&Sigma;}}{\mathrm{\&beta;}}_r{X}_r+\underset{r^{\&prime;}}{\mathrm{\&Sigma;}}{\mathrm{\&beta;}}_{r^{\&prime;}}\underset{s\&Element;D_j}{\mathrm{\&Sigma;}}{P}_s^t{X}_{r^{\&prime;}}\left(d\right|s)$

$\&ForAll;d\&Element;D_i,{\&ForAll;D}_i\&Element;D,j\&NotEqual;i---\left(2\right)$

Wherein, Xr is options d attribute r, and Xr ' is options d attribute r ' after given state S, and S is preference pattern Dj,

State, β is the parameter of variable X;

B, Pts for moment t state S probability of happening computing formula are:

$P_S^t=\underset{d\&Element;S}{\mathrm{\&Pi;}}{P}^t\left(d\right),S\&Element;D_j,j\&NotEqual;i---\left(3\right)$

Select the reciprocal initial probability calculation formula of item number to be in c, preference pattern Di:

$P^0\left(d\right)=\frac{1}{\left|D_i\right|},\&ForAll;d\&Element;D_i,{\&ForAll;D}_i\&Element;D---\left(4\right).$

6. a kind of selection bicycle mode based on the activity chain pattern according to claim 1 Forecasting Methodology of going on a journey is characterized in that:

In described step (4), in each preference pattern of t, each options effectiveness is subject to the t-1 impact of preference pattern result constantly constantly, and each iteration cycle is last, and the uncertainty of each preference pattern Di depends on that the big or small computing formula of entropy is:

$H^t\left(D_i\right)=-\underset{d\&Element;D_i}{\mathrm{\&Sigma;}}{P}^t\left(d\right)\&times;{\log }_2\left\{P^t\left(d\right)\right\},\&ForAll;D_i\&Element;D---\left(5\right)$

Wherein, introduce to adjust coefficient θ i in entropy function each decision-making branch entropy of initial time is equated, the preference pattern item correspondence of less entropy θ iHt (Di) is less has uncertainty, and at first this preference pattern item is determined and no longer changed in the successive iterations process.

Images