CN112179370A

CN112179370A - Continuous optimal route planning method based on dynamic road network

Info

Publication number: CN112179370A
Application number: CN202011233351.9A
Authority: CN
Inventors: 商烁; 陈力思; 王皓; 杜峰; 任平; 张绍林
Original assignee: Thinvent Digital Technology Co Ltd
Current assignee: Thinvent Digital Technology Co Ltd
Priority date: 2020-11-06
Filing date: 2020-11-06
Publication date: 2021-01-05
Anticipated expiration: 2040-11-06
Also published as: CN112179370B

Abstract

The invention discloses a continuous optimal route planning method based on a dynamic road network, which gives a group of travel request flows Q_nFirstly, generating a high-quality initial route combination pi by using initial route search_nThe combination includes an initial route specified for each request in the travel request stream. Next, a batch optimization process is performed to optimize each of the initial routes in the initial route combination. Finally, combining the optimized routes II_nAnd adding the global route combination pi. And returning pi as a real-time planning result after processing a batch of travel requests. By the mode, the fact that the newly planned route influences the future traffic state is considered by the initial route searching method, namely the planned route increases the traffic flow of some road sections so as to influence the actual traffic time of the road sections to search the optimal route combination for the continuous formation request, the algorithm efficiency is improved, and the problem of line congestion is solved.

Description

Continuous optimal route planning method based on dynamic road network

Technical Field

The invention relates to the field of transportation, in particular to a continuous optimal route planning method based on a dynamic road network.

Background

With the rise of applications based on location services, path planning services have become an indispensable part of life. Route planning and trip recommendation have led researchers to extensive research in recent years. The goal of these studies is to make an optimal route for a single trip based on current traffic conditions. It is worth noting that with the increasing frequency of path planning service usage, a large number of users are likely to distribute travel requests densely within extremely short time intervals, particularly during high peak periods such as commute time, thereby creating a continuous stream of travel requests. In this new scenario, the need to implement path planning for the travel request stream becomes more urgent. There has been relevant research aimed at sequentially making individual optimal routes for individual trips in a trip request. However, when planning a route for a travel request stream, planning an individually optimal route with an individually optimal final goal based only on current traffic conditions may lead to traffic congestion. More rational route planning should take into account that previously planned routes will have an impact on future traffic conditions, since they will increase the traffic flow on the road sections in the road network.

Optimal route planning has been widely researched in a plurality of different application scenarios, existing route planning based on global optimization is not oriented to real-time traffic conditions, how to process continuous travel request streams and how to quantify influences of the travel request streams on future traffic are not considered, and therefore, a route planning method oriented to single requests often causes potential traffic congestion.

Disclosure of Invention

The invention provides a continuous optimal route planning method based on a dynamic road network for solving the problems, which is realized by the following technical scheme:

a continuous optimal route planning method based on a dynamic road network comprises the following steps:

s1, acquiring dynamic road network information;

s2, planning an initial route for each newly arrived travel request by utilizing an initial route searching algorithm to obtain an initial route set;

s3, periodically carrying out batch optimization processing on the initial route planned before the current time node at fixed time intervals, and adding the initial route into an optimal route set;

s4, updating road section labels corresponding to all the routes in the route set, issuing the optimized optimal route set and emptying the initial route set;

and S5, outputting the optimized optimal route set in the current period and performing route planning in the next period by using the steps.

The scheme has the advantages that the initial route planned before the current time node is periodically subjected to batch optimization processing at fixed time intervals, more factors influencing traffic states can be considered, and potential traffic jam is avoided.

Further, the initial route search algorithm specifically includes:

s21, initializing the information of each vertex in the road network, and creating a vertex set through which a planned route passes;

s22, setting a priority queue, adding the starting point of the travel request issued at the current moment in the travel request flow into the queue, and selecting the top point of the head of the queue from the starting point to expand the top point in the priority queue;

and S23, when the vertex expands to the stroke request end point, reversely generating a route from the end point according to the precursor vertex records of each vertex, and returning the route vertex set as an initial route result.

The further scheme has the advantages that by using the priority queue, the time complexity of the scheme/algorithm can be reduced, and the operation efficiency is improved. The head of line element is a vertex object that has the property that there is a minimum expected transit time to reach the end point through the vertex. The expansion can be finished heuristically and quickly by selecting the head element, namely the vertex, for expansion each time, and a proper initial route can be found quickly.

Further, the process of apex expansion is as follows:

expanding from any stroke starting point in the stroke request stream to an adjacent vertex connected with the stroke starting point, and selecting a head of line element adjacent to the stroke starting point;

continuing to expand to the adjacent vertex connected with the selected head element by taking the selected head element as a starting point;

when the vertex after expansion is a stroke end point, reversely generating a route from the end point according to the record of the front-driving vertex of each vertex;

when the selected vertex is not the travel end point, the traffic flow of each route sent from the point is calculated, and the elapsed time from the starting point of the section of the route to the end point of the section of the route is updated.

The advantage of the above further solution is that the apex expansion ends when it reaches the end of the stroke. The precursor node information of each corresponding vertex object in the road network accurately records a vertex connected with the precursor node information, and the current vertex object can be reached at the fastest speed through the vertex. Continuously backtracking the precursor nodes of the destination object from the destination object; and searching for the precursor node of the precursor node until the starting point of the travel is found, so that an initial route from the starting point to the end point is generated. This route is the optimal route we have modeled, i.e. at the departure time, it has the smallest expected transit time.

Further, the method for updating the elapsed time from the starting point of the road segment to the end point of the road segment comprises the following steps:

checking whether the end point of the segment can be reached through the selected vertex:

if the time required for reaching and satisfying is shorter than the time required for reaching from other vertices, the time from the starting point to the ending point, the time for passing through the selected route, and the information on the preceding vertices are updated.

The method has the advantage that the precursor vertex information recorded by each vertex can be continuously updated through the expansion of the dynamic road network. Specifically, the previous vertex that reached the point most quickly is continuously updated. The updated vertex is rejoined into the priority queue for updating the next vertex to which it is connected. The information that needs to be updated can be quickly calculated and these updates are necessary. Through the continuous expansion of the network, a high-quality initial route can be quickly and accurately found.

Further, the elapsed time of the vehicle from the start point of any link to the end point of the link at any time is expressed as:

where e is the selected road segment, t is the time at which the vehicle passed the road segment e, C_eIs the capacity of the road section e;

t (e, T) is the passing time of the road section e at the moment T, and f (e, T) is the traffic flow of the road section e at the current moment.

The beneficial effect of the further scheme is that the influence of different properties (such as length, width, speed limit and the like) of each road section on the road section passing time is reasonably considered. Of particular concern is the effect of the predicted minimum transit time for the road segment and the real-time traffic flow on the road segment on the actual transit time for the road segment, consistent with real-life scenarios. The scheme has better expandability for different road networks with different parameter values.

Further, the batch optimization processing method comprises the following steps:

s31, selecting one line from the initial route set one by one as the current initial route to be optimized, and updating the road section label according to the defined published route;

s32, according to the updated road section label, re-searching a shortest new route from the starting point to the end point of the travel, and executing the exchange operation of the new route and the initial route;

s33, checking the validity of the exchange operation according to the pre-checking strategy;

and S34, repeating the steps S31 to S33, and issuing the route set when no optimizable new route exists for any one route in the initial route set.

The further scheme has the advantages that unnecessary traffic jam can be avoided by optimizing the planned route set by considering the potential influence of the planned route set on future traffic. With the road segment labels, this potential impact can be quantified.

Further, the pre-check strategy is to calculate an upper bound rate of a total transit time reduction rate of changing all routes by any one route, and is specifically represented as:

UB (pi) is the upper bound rate of the total transit time reduction rate of all the routes changed by any route;

II is an optimal route combination which is planned in the journey request flow;

Π_nan initial route set currently being optimized is obtained;

TT(Π,Π_n) The total passing time of all routes comprises the planned optimal route combination and the currently optimized initial route set;

TT({Π,Π_n\\ pi }) is the total transit time for all routes after the initial route is taken out and the effect of the initial route on other trips is eliminated.

The beneficial effect of the above further scheme is that the route to be optimized with smaller lifting space is filtered through the pre-checking strategy, and the operation efficiency of the algorithm can be further improved.

Further, the effectiveness of the exchange strategy is expressed by the following calculation formula:

therein, II_nSwap (π, π') represents the total transit time for all trips after the swap;

e is a constant.

Further, the number of the swap operations in the current time period is a limited number of operations, and the number of the maximum effective swap operations is represented as:

wherein TT is total transit time of initial line combination planned currently, TT_mThe minimum transit time in the ideal state.

The beneficial effect of the further scheme is that the exchange operation of the route with smaller lifting space can be filtered by adopting a smaller lifting factor E, and only the exchange operation of the route with larger lifting space for the global transit time is adopted. Therefore, the global transit time can be greatly reduced by each effective exchange, the exchange times are reduced, and the efficiency of the algorithm is greatly improved.

Drawings

The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention. In the drawings:

fig. 1 is a schematic flow chart of a continuous optimal route planning method based on a dynamic road network according to the present invention.

FIG. 2 is a diagram of experimental data of the impact of the number of trip requests | Q | of different algorithms on different road networks, wherein:

FIG. 2(a) shows CPU running time on NY road network for different numbers of travel requests, four algorithms, IND, SBP SA, SBP and SBP DA.

In FIG. 2(b), the global transit time of the route set planned by the four algorithms of IND, SBP SA, SBP and SBP DA on the NY road network under different numbers of travel requests.

FIG. 2(c) shows CPU running time on TG nets for four algorithms, IND, SBP SA, SBP and SBP DA, with different numbers of travel requests.

In FIG. 2(d), the global transit time of the route set planned by the four algorithms IND, SBP SA, SBP and SBP DA on the TG road network under different numbers of travel requests.

Fig. 3 is a schematic diagram of experimental data of influence of the arrival rate of the journey request on different road networks under different algorithms, wherein:

FIG. 3(a) illustrates the CPU running time consumed by the SBP algorithm to plan route sets on NY road network and TG road network, respectively, at different levels of travel request arrival rates.

And (b) in FIG. 3, under different degrees of travel request arrival rates, the SBP algorithm respectively plans the global transit time of the route set on the NY road network and the TG road network.

Fig. 4 is a schematic diagram of experimental data of parameters α & β on different road networks, wherein:

in FIG. 4(a), different parameters alpha are set, and the SBP algorithm is used for planning the CPU running time consumed by the route set on the NY road network and the TG road network respectively.

In FIG. 4(b), different parameters alpha are set, and the SBP algorithm respectively plans the global transit time of the route set on the NY road network and the TG road network.

FIG. 4(c) sets different parameters β, the CPU running time consumed by the SBP algorithm to plan the route sets on NY road network and TG road network respectively.

In FIG. 4(d), different parameters beta are set, and the SBP algorithm respectively plans the global transit time of the route set on the NY road network and the TG road network. .

Fig. 5 is a schematic diagram of experimental data of the influence of batch optimization time interval T on different road networks, wherein:

in FIG. 5(a), different batch optimization time intervals T are set, and the CPU running time consumed by the SBP algorithm for planning route sets on the NY road network and the TG road network is respectively set.

And (b) setting different batch optimization time intervals T, and setting the global transit time of the route set planned by the SBP algorithm on the NY road network and the TG road network respectively. .

FIG. 6 is a diagram of experimental data of the influence of parameters E and pre-inspection strategies on different road networks, wherein:

and in the step (a) of FIG. 6, different parameters are set to form the CPU running time consumed by the SBP algorithm for planning the route set on the NY road network and the TG road network respectively.

And in the step (b) of FIG. 6, setting different parameters E and the SBP algorithm to respectively set the global transit time of the planned route set on the NY road network and the TG road network.

FIG. 6(c) shows CPU operation time consumed by routing a route set on the NY road network by using two algorithms of SBP and SBP PC without pre-check feature under different numbers of travel requests.

Detailed Description

Hereinafter, the term "comprising" or "may include" used in various embodiments of the present invention indicates the presence of the invented function, operation or element, and does not limit the addition of one or more functions, operations or elements. Furthermore, as used in various embodiments of the present invention, the terms "comprises," "comprising," "includes," "including," "has," "having" and their derivatives are intended to mean that the specified features, numbers, steps, operations, elements, components, or combinations of the foregoing, are only meant to indicate that a particular feature, number, step, operation, element, component, or combination of the foregoing, and should not be construed as first excluding the existence of, or adding to the possibility of, one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing.

In various embodiments of the invention, the expression "or" at least one of a or/and B "includes any or all combinations of the words listed simultaneously. For example, the expression "a or B" or "at least one of a or/and B" may include a, may include B, or may include both a and B.

Expressions (such as "first", "second", and the like) used in various embodiments of the present invention may modify various constituent elements in various embodiments, but may not limit the respective constituent elements. For example, the above description does not limit the order and/or importance of the elements described. The foregoing description is for the purpose of distinguishing one element from another. For example, the first user device and the second user device indicate different user devices, although both are user devices. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of various embodiments of the present invention.

It should be noted that: if it is described that one constituent element is "connected" to another constituent element, the first constituent element may be directly connected to the second constituent element, and a third constituent element may be "connected" between the first constituent element and the second constituent element. In contrast, when one constituent element is "directly connected" to another constituent element, it is understood that there is no third constituent element between the first constituent element and the second constituent element.

The terminology used in the various embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the various embodiments of the invention. As used herein, the singular forms are intended to include the plural forms as well, unless the context clearly indicates otherwise. Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which various embodiments of the present invention belong. The terms (such as those defined in commonly used dictionaries) should be interpreted as having a meaning that is consistent with their contextual meaning in the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein in various embodiments of the present invention.

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to examples and accompanying drawings, and the exemplary embodiments and descriptions thereof are only used for explaining the present invention and are not meant to limit the present invention.

Example 1

A continuous optimal route planning method based on dynamic road network, as shown in fig. 1, includes the following steps:

s1, acquiring dynamic road network information;

specifically, a dynamic road network G ═ (V, E) is composed of a set of vertices V and a set of edges E, where

The vertices represent the connection points of the links, and the connecting edges represent the specific links. Each side e (v)_i,v_j) Connecting two vertexes v_iAnd v_jV herein is_i,v_jE.g. V. For any section e, use C_eTo represent the vehicle capacity of the road section, T_m(e) Representing the minimum transit time on the section e, i.e. the time when no other vehicles are passing, in this embodiment, the transit time of each section is dynamically changed, and the real-time transit time of a section is the minimum transit time T_m(e) And the real-time dynamic traffic flow on the road section is in a proportional relation.

specifically, the initial route search algorithm may be expressed as:

in this embodiment, the dynamic traffic flow of each road section is composed of two parts, one part is the traffic flow of an unsolicited vehicle, i.e., a vehicle not using a route planning system, and is directly used as input data, which is not a research category; some are vehicles that request a traffic flow for vehicles, i.e., vehicles that use a planning system, that travel on a planned route with increased traffic flow over the traversed road segment. The traffic volume generated by the vehicle requested for segment e at time t is the number of vehicles using our planning system that are still traveling on segment e at time t. The traffic flow of each road section of the road network is dynamically changed, and the traffic time is also dynamically changed along with the traffic flow, namely, the shortest path from the starting point to the end point is searched in the dynamically changed road network.

To enable the calculation of traffic flow within the system, a series of link labels L are maintained on each link_e＝(l₁，l₂...l_i) The link labels record time information of a route passing through the modified link and the link, and each link label l_i＝{t_a，t_bRecord a specific request vehicle information of entering section e, including starting time t_aAnd the time t of leaving the road section_b. Requesting the traffic volume generated by the vehicle, i.e. the link label L, for the link e at the time t_eT in the set_a≤t≤t_bThe number of tags in the label. When no travel request exists, the road section label set maintained by each road section is an empty set, and then when a batch of new forming request sets are not processed completely, the road section labels are dynamically updated, and the actual passing time of each road section at the current moment is updated.

Each vertex V ∈ V contains the following record information from the starting point V_sPrecise transit time t to this point_sLower bound t of transit time from this point to the point of emphasis_dThe minimum time et from the starting point to the point and a precursor vertex information pred, after initializing the information of each vertex, setting a route set pi as an empty set.

the vertex object inside the priority queue PQ according to t at that point_s+t_dIs arranged from small to large, wherein the head-of-line element has the smallest t_s+t_dA point of value. In the process of line searching, head element v is selected from priority queue PQ each time and adjacent street vertex connected with head element v is searched, and each selected vertex v also has minimum t_s+t_dBased on such vertex dilation policy, the vertex of the smallest transit time may be generated.

When the selected vertex v is the end of travel v_dAnd reversely utilizing all the precursor vertexes to generate a route pi, and taking the route pi as a result to return a value.

When the selected vertex v is not the end of travel v_dThen calculate each route e from vertex vFor the end point v ' of the route e (v, v '), it is checked whether the end point can be reached by the vertex v, and the time taken to reach the end point is calculated, and if the time of arrival is shorter than the time of arrival of each of the other vertices at the end point v ', the vertex information corresponding to the vertex v ', that is, the time t from the start point to the end point v ', is updated_sWhen the vertex v 'is not in the priority queue PQ, the vertex v' is added to the priority queue. When the set of priority queues is empty or has expanded to the end of the journey request, the route search algorithm terminates

S3, periodically performing batch optimization processing on the initial route planned before the current time node at fixed time intervals, and adding the initial route into the optimal route set, which can be specifically expressed as:

in the scheme, an initial route set pi_nIncluding the most recently arrived journey request Q_nInitial course specified, other than Π_nThat is, the global route combination Π includes all routes planned before the current time, and the process of optimization requires the reuse of the initial route set Π_nFor the initial route set, it may specifically be an initial route set specified by a travel request stream arriving in a time period, including routes planned by the global road network in the same time period before the current time.

in each optimization process, selecting any route pi epsilon_nInputting the route planning, which considers the planned initial route other than the route as a known route, into the traffic information stream, predicting the future traffic state again and trying to reassign a new route pi' to the journey request q_iFurther considering the potential influence of the initial route currently planned on the future traffic condition, a new road section label set is updated

It is made up by using original road section label

And estimated time information of the initial route being planned. Handle

As known traffic information, an initial route search algorithm is performed for the journey q_iAssigning a new route of pi'_i。

In order to check the validity of an exchange operation, the link labels of all the links must be updated to calculate the total transit time reduction rate of all the trips, which is very time-consuming, therefore, the scheme proposes a pre-checking strategy to further improve the efficiency of the algorithm, specifically, an upper bound UB of the total transit time reduction rate of all the trips is defined, the specific calculation mode is,

wherein pi is the selected initial route to be optimized, and UB (pi) is the upper bound rate of the total transit time reduction rate of all the routes changed by the selected route to be optimized;

Π_nan initial route set currently being optimized is obtained;

TT(Π，Π_n) Is composed ofThe total passing time of all routes comprises the planned optimal route combination and the currently optimized initial route set;

TT({Π，Π_n\\ pi }) is the total transit time for all routes after the initial route is taken out and the effect of the initial route on other trips is eliminated.

Note that TT ({ Π, Π) here_n\π})≠TT(Π，Π_n) ET (pi, t), ET (pi, t) is the predicted transit time for calculating a route pi, i.e. is the sum of the predicted transit times for each section in the route pi. The expression of the formula is that after pi is eliminated, the total passing time of all the routes is assumed to be absent, wherein not only the passing time of the route pi itself is considered, but also the influence of the route pi on other routes is considered to be eliminated, and the passing time of other routes is influenced to a certain extent due to the fact that the route pi causes other traffic flows.

To check the validity of the swap operation, it is provided that the swap operation can reduce the sum of all formed identities by a ratio of at least 1+ ∈, i.e.

This exchange operation is effected artificially and is used to update the original initial path, of the type Π_nSwap (π, π') represents the total transit time for all trips after the swap, e is a fixed constant. A very small constant value e is used to exclude invalid operations, i.e. operations that reduce the transit time by swapping operations only to a small extent. In addition, the setting of the constant e also ensures that the number of swap operations is limited, in practice the number of swap operations and e are inversely proportional, in particular, the number of most efficient swap operations can be expressed as,

wherein TT is total transit time of initial line combination planned currently, TT_mIs the minimum transit time in the ideal state。

When the journey requests Q_nWherein the batch algorithm is terminated when there is a valid operation that can no longer produce a better planning result. At this time, for any route pi ∈ pi_nNo effective switching operation can be found.

During the optimization process, pi for each initial route_nIt is checked whether it can be replaced by a new route pi'. It is assumed that there is no valid swap operation in this round robin. For each route pi ∈ pi_nThe validity of the swap operation is pre-checked by computing a lifting upper bound UB (π). If the swap operation is potentially valid, a new road segment label L' is generated. With L 'as an input, the initial route search method is performed to generate a new route pi'. If the swap operation is finally determined to be valid, this swap operation will be used and the flag set to true to record that the current cycle still has a valid swap operation. When one cycle is finished, all pi epsilon_nNo valid swap operation is detected and the algorithm terminates. That is to say that no effective operation exists which can produce better results

and S5, outputting the optimized optimal route set in the current cycle and performing route planning in the next cycle by using the steps S2-S4.

Experimental study

Experimental studies using two road network datasets: san Joaquin County Road Network (TG Road Network)¹And the New York Road Network (NY Road Network)²Storing the road network graph by using the adjacency list. For each link, a vehicle capacity C with a value of 20-100 is assigned to each link based on the length of the link_e. Minimum transit time T per road section_m(e) Is randomly generated and is between 5 and 10 minutes. To simulate an unsolicited vehicle, assume an average traffic flow of unsolicited vehicles on each road segment e

Is 0.4 XC_eWith real-time unsolicited vehicles dynamically changing periodically about average traffic flow

To

Given a minimum unsolicited traffic flow

And calculating the lower bound of the passing time between every two vertexes in advance as the static traffic flow of all road sections, wherein the heuristic value H (v) in the lower initial route searching algorithm_i,v_j) Randomly sampling a starting point-an end point to simulate a travel request in a region with a dense road network, and giving a starting time q of a first travel₁T, departure time q of the following ith trip_iT departure time q of last trip request_i-1Small amplitude increases or remains unchanged on the basis of t. The rate at which travel requests arrive is set as shown in Table 1

TABLE 1 Rate settings data for journey request arrivals

	TG road network	NY road network
			Number of journey requests \| Q	4,000-20,000\|default 4,000	2,000-10,000\|default 2,000
Parameter e	0.0002-0.0010\|default 0.001	0.0002-0.0010\|default 0.001
			Optimizing interval T	1s-5s\|default 2s	1s-5s\|default 2s
Rate of arrival of travel	20/s-100/s\|default 50/s	20/s-100/s\|default 50/s
			α	1-5\|default 2	1-5\|default 2
β	1-3\|default 2	1-3\|default 2

The real-time traffic flow f (e, t) on the section e at the time t is calculated according to the following formula,

f(e,t)＝f′(e,t)+f″(e,t)

where f' (e, t) represents the traffic volume of the unsolicited vehicle outside the system and f "(e, t) represents the traffic volume of the solicited vehicle within the system. To simulate an unsolicited vehicle, assume that f' (e, t) is dynamically changing periodically over time.

In order to evaluate the result quality of the algorithm, a person-based search algorithm (Ind algorithm) is additionally implemented. Specifically, given a set of travel request streams Q, for each travel request Q e Q, the Ind algorithm generates an optimal personal optimal route based on the traffic conditions at the departure time of a single travel, consistent with existing research, and the performance of the exact algorithm is not studied because it is time consuming, requiring at least one day of time at default settings. The SBP algorithm consists of two parts: initial route search and batch optimization process. The proposed initial route search reasonably considers the self-induction idea and the network dynamics, and in order to verify the superiority, the SBP & ltSA & gt algorithm and the SBP & ltDA & gt algorithm are respectively implemented. The SBP-SA algorithm is an SBP algorithm that does not adopt a self-perception concept for initial route search, that is, it does not consider an additional influence of a route planned by a planning system along with a traffic flow. The SBP-DA algorithm is an SBP algorithm that does not consider network dynamics, and is directed to a static traffic state snapshot, i.e., a static traffic state corresponding to a departure time of a trip, in an initial route search process. CPU running time and total pass time TT (Π) of all the trips are used as evaluation indexes of algorithm efficiency and algorithm accuracy. Unless specifically stated, the experimental results are the average results of more than 20 independent experiments. All algorithms were run on the windows10 platform using Java programming using Intel (R) core (TM) i5-9300 HPprocessor (2.40GHz) and 16GB of memory. Specific default parameter settings are shown in table 1.

Experimental results as shown in fig. 2, the performance of the proposed algorithm was investigated at different trip request numbers | Q | under default parameter settings. Intuitively, the larger the number of travel requests | Q |, the total transit time of all the travel routes increases. In addition, a larger | Q | also results in greater computational consumption and increased CPU run time. Figure 2 shows the performance of the proposed method on TG road network and NY road network. Unexpectedly, as the number of trip requests | Q | increases, the required CPU time increases. Of these algorithms, the personal Ind based algorithm requires the least CPU time because it is once planned and not subject to the proposed batch optimization process. It has also been found that the total transit time TT (Π) of all routes in the global route combination generated by the Ind algorithm is greater than other algorithms, particularly where the number of route requests is greater (e.g., SBP algorithms that can reduce transit time by over 40% when | Q | -10,000 in the NY road network and NY | Q | -20,000 in the TG road network compared to the Ind algorithm).

SBP SA and SBP DA performed poorly compared to SBP algorithms. Since they would perform a relatively large number of switching operations in the optimization process, which is very time consuming. Specifically, SBP SA does not take into account that an already planned route also adds additional traffic to the road traffic, and therefore its traffic prediction during the initial route search phase is not reliable. For the SBP-DA algorithm, the path planning is carried out by taking a snapshot of the traffic state at the departure time of a journey. Therefore, the initial routes specified by the SBP SA algorithm and SBP DA algorithm tend to be of low quality, requiring extensive swapping operations to be performed during batch optimization to adjust these low quality initial routes, thereby resulting in more computational expense to improve the quality of the results. It is also observed from fig. 2(b) and 2(d) that the total transit times of all the routes of the global route combination generated by the three algorithms are very close, and that the SBP algorithm always generates a smaller transit time TT (Π). These experimental results demonstrate that by adopting the self-perception idea and considering network dynamics, the initial route search can effectively generate a high-quality processing route, and such an initial route can reduce the number of switching operations in the subsequent optimization processing.

Fig. 3 shows the algorithmic behavior of the SBP algorithm when the trip request arrival rate changes. The larger the arrival rate of the travel request is, the more traffic flows are poured into each road section of the road network in unit time. In this case, the increase in the amount of traffic per road section becomes more significant. Thus, more additional traffic is generated on each road segment. In fig. 3, it can be seen that the CPU time and the total transit time TT (Π) have a significant tendency to increase, both in the NY road network and in the TG road network. TT (Π) increases because as traffic flow increases, the transit time per road segment increases accordingly. On TG networks, the total transit time for all the itineraries planned is much greater than on NY networks, since TG networks have a greater number of itinerary requests generated in their default settings.

Fig. 4 shows the algorithm behavior of the SBP algorithm when varying the parameters α and β (cf. equation 1) in the transit time function. Intuitively, the larger alpha or the smaller beta, the larger the passage time of each road segment increases, and thus traffic congestion may be aggravated. In fig. 4(a) and 4(b), when α is changed from 1 to 5, both the CPU time and the total transit time for all trips tend to increase. In contrast, fig. 4(d), the total transit time has a tendency to decrease as β changes from 1 to 3. It is noted that, as shown in fig. 4(c), when β is changed, the CPU time does not have a particularly significant tendency to change, because the CPU time, i.e., the required calculation consumption, is not directly related to the value of β.

Fig. 5 shows the behavior of the SBP algorithm when the optimization time interval T varies. A larger T represents an increased latency per request, but the planning results tend to be more optimal because more of the route being planned is taken into account in influencing future traffic conditions. In this case, the total transit time TT (Π) of all the trips is greatly reduced by the exchanging operation. In fig. 5(a), CPU time increases with increasing T because more travel requests need to be processed in one optimization process. In fig. 5(b), it can be seen that TT (Π) has a slight tendency to decrease. These results demonstrate that a suitable optimization interval can more accurately predict future traffic conditions by reusing the already planned initial route, thereby avoiding traffic congestion.

The SBP algorithm performance when the parameter e is changed from 0.0002 to 0.001 was studied, considering the impact of the parameter e pre-checking strategy on the algorithm performance. The total number of active swap operations during the optimization process is inversely proportional to the value of the parameter e. Larger e means that the higher the threshold of valid operations, the fewer the number of valid operations and hence the CPU time will be reduced. However, a larger ∈ also results in a global route combination ∈ with a larger transit time. In fig. 6(a), when the parameter e is changed from 0.0002 to 0.001, the CPU time tends to decrease. In contrast, the total transit time for all trips has a slightly increasing trend, as shown in fig. 6 (b). When the epsilon changes from 0.0002 to 0.0004, the reduction amplitude of the required CPU time is the largest, and in the NY road network and the TG road network, the reduction amplitude of the required CPU time respectively exceeds 36 percent and 19 percent. It was also found that 0.001 is very suitable for the choice of the value of e, and it can be seen from the figure that the combined performance is best when e is 0.001. The above experimental results show that by using a suitable parameter e, the performance of the SBP algorithm can be greatly improved. The SBP algorithm without the pre-check strategy is denoted by "SBP PC", and in fig. 6(c), it is observed that the SBP algorithm always requires less CPU run time than the SBP PC algorithm, and this experimental result shows that the proposed pre-check strategy can further improve the algorithm efficiency.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are merely exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims

1. A continuous optimal route planning method based on a dynamic road network is characterized by comprising the following steps:

s1, acquiring dynamic road network information and a travel request stream;

s2, planning an initial route for each newly arrived travel request by using an initial route searching method to obtain an initial route set;

s3, periodically carrying out batch optimization processing on the planned initial route before the current time node at fixed time intervals, and adding the initial route into an optimal route set;

2. The method according to claim 1, wherein the initial route searching method comprises:

3. The method according to claim 2, wherein the vertex dilation comprises:

when the selected vertex is not the travel end point, the traffic flow of each route sent from the point is calculated, and the elapsed time from the starting point of the road section to the end point of the road section is updated.

4. The method according to claim 3, wherein the method for updating the elapsed time from the start point of the road segment to the end point of the road segment comprises:

and checking whether the end point of the link can be reached through the selected vertex, and if the end point of the link can be reached and the time is less than the time from other vertexes, updating the time from the starting point to the end point, the time of passing through the selected link and the information of the precursor vertexes of the link.

5. The method according to claim 4, wherein said dynamic road network-based continuous optimal route planning method,

the time taken from the starting point of the road segment to the end point of the road segment is expressed as:

where e is the selected road segment, t is the time at which the vehicle passed the road segment e, C_eIs the vehicle capacity of the section e, T (e, T) is the transit time of the section e at the time T, T_m(e) F (e, t) is the real-time traffic flow on the road section e at the moment t, and alpha and beta are set constants.

6. The method according to claim 5, wherein the batch optimization processing method comprises:

s31, selecting one line from the initial route set one by one as the current initial route to be optimized, and updating the road section label according to the published route at the current moment;

7. The method according to claim 6, wherein the pre-check strategy is to calculate an upper bound rate of a total transit time reduction rate of any one route for changing all routes, specifically represented as:

Π_nan initial route set currently being optimized is obtained;

TT({Π,Π_n\ pi }) is the total transit time of all routes after the initial route to be optimized is taken out and its impact on other trips is eliminated.

8. The method according to claim 6, wherein the effectiveness of said switching strategy is expressed by the following calculation formula:

therein, II_nSwap (π, π') represents the total transit time for all trips after the swap; pi' is a new shortest route generated by the selected route pi to be optimized, and epsilon is a constant.

9. The method according to claim 6, wherein the maximum number of switching operations is expressed as:

wherein TT is the total passing time of the initial line combination planned currently,

TT_mthe minimum transit time in the ideal state.