CN105006149B - Traffic estimates Dynamic iterations method - Google Patents
Traffic estimates Dynamic iterations method Download PDFInfo
- Publication number
- CN105006149B CN105006149B CN201510406126.3A CN201510406126A CN105006149B CN 105006149 B CN105006149 B CN 105006149B CN 201510406126 A CN201510406126 A CN 201510406126A CN 105006149 B CN105006149 B CN 105006149B
- Authority
- CN
- China
- Prior art keywords
- section
- variance
- vehicle
- hourage
- dynamic iterations
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Landscapes
- Traffic Control Systems (AREA)
Abstract
Estimate Dynamic iterations method the invention discloses traffic, this method utilizes existing charge record data, the operation conditions of road is reappeared by Dynamic iterations method, including recorded for every OD, calculating obtains the shortest path that the vehicle is travelled in road network；Driving states of the inverting vehicle on each section of the shortest path, obtain the flow and speed at each section each moment based at the uniform velocity assuming；Dynamic iterations inverting is carried out to all vehicle registrations；When calculating obtained variance f (x) more than or equal to setting variance threshold values, Dynamic iterations are carried out by variance equalization adjustment variance；Obtain final inversion result.Time Continuous is made each tracing point of each iteration be adjusted on space-time by the present invention by Dynamic iterations, if so that it is guaranteed that the reasonability and accuracy of reproduction road operation conditions result, can help to monitor road running situation in real time.
Description
Technical field
The present invention relates to technical field of intelligent traffic, more particularly to one kind according to history charge data to highway state
The Dynamic iterations method estimated.
Background technology
With the fast development of China's economy, the development of highway communication technology and the raising of the efficiency of management become heavy to closing
Will.Freeway traffic system is typical complication system, and the complexity of the system is based on the network structure of road.Have
Effect ground carries out the basis that accurate measurement is Improving Expressway operation management level and service quality to traffic route situation.
The topological structure of transportation network and the traffic parameter of road can reflect the performance of transportation network to a certain extent,
But then to be needed history in transportation network topological structure and transportation network from the performance of more fully angle estimator transportation network
And current traffic flow conditions be combined.The performance metric of transportation network mainly contains the current effect of entirety of overall road network
Rate, the travel pattern of each part (such as import and export, section), road passage capability are with expanding that volume of traffic increase is shown
Malleability and environmental pollution according to arithmetic for real-time traffic flow situation and the measurement of power consumption state etc..
At present, it is to understand road grid traffic situation, generally using the average speed method of vehicle, i.e., is to change point according to 15 minutes
Time window (when window), window distribution when being carried out to every track, then statistics calculating is carried out to the data in each window, handed over
Logical situation.This existing method belongs to discreet static method, the actual speed difference between each section has spatially been obliterated, in the time
Upper use discrete time slot, reduces time-varying precision, also fails to accurately to conclude each data (vehicle OD) and should have on statistics is calculated
Space-time position point, therefore, the accuracy estimated using average speed method road network traffic is not high.
The content of the invention
In order to overcome the above-mentioned deficiencies of the prior art, the present invention is based on traffic flow model knowledge, according to crossing vehicles passing in and out
Data inquire into (inverting) traffic on road network at that time, a kind of traffic estimation Dynamic iterations method is specifically provided, by when
Between serialization, the iterative calculation linked makes each tracing point during each iteration be adjusted on space-time, global excellent by setting
Change target and strict constraints, it is ensured that estimated result reappears the reasonability and accuracy of road operation conditions, can help reality
When monitor road running situation.
The technical scheme that the present invention is provided is：
A kind of traffic estimates Dynamic iterations method, and this method utilizes existing charge record data, passes through Dynamic iterations
Method reappears the operation conditions of road, specifically includes following steps：
1) beginning and end OD records all in driving recording, including road network are taken out；
2) for every OD record, calculating obtains the shortest path that vehicle is travelled in road network, is made up of multiple sections, remembered
Make { Section1,Section2,…,Sectionn}；
3) based at the uniform velocity it is assumed that calculating obtains vehicle into { t at the time of each section1,t2,…,tn, inverting vehicle exists
Driving states on each section of the shortest path, obtain the flow and speed at each section each moment based at the uniform velocity assuming；
4) Dynamic iterations inverting is carried out to all vehicle registrations；
For the OD vehicle registrations of system-wide net, variance f (x) is calculated according to formula 1, judges whether variance f (x) meets setting
Variance threshold values；
F (x)=min (∑siD(Ti),if i∈Snon-free) (formula 1)
In formula 1, min () represents that optimization aim is minimum value；Snon-freeIt is the atom section of all non-free streams；D
(Ti) it is variance calculation formula D (x)=E (x2)-E(X)2, wherein, x is the average value of the hourage in the section, and X is the section
Hourage set；E (X) is the average value of the hourage in the section；I is the numbering in section；TiI-th of atom section
The hourage vector of upper all OD records；
When calculating obtained variance f (x) less than the variance threshold values set, terminate Dynamic iterations inverting, into step 5)；
When calculating obtained variance f (x) more than or equal to setting variance threshold values, circulated as follows：
41) each section in road network is directed to, in the expectation that the speed for obtaining the section is calculated according to weight distribution principle
Value：
In formula 2, mtiIt is the intermediate value of the hourage in i-th of section；tijThe trip on i-th of section is recorded for j-th of OD
The row time；TjFor the hourage vector of OD records all on j-th of atom section；The vehicle journeys time is longer, is calculating
Shi Quanchong is smaller；
42) for every OD record and step 41) in the obtained expectation intermediate value of section speed, according to re-allocation principle,
By variance equalization adjustment variance, make tijTowards with mtiThe small direction of subtractive be iterated：
tij=tij+alpha*(mti-tij) (formula 3)
In formula 3, tijThe hourage on i-th of section is recorded for j-th of OD；mtiIt is the hourage in i-th of section
Intermediate value；Alpha is coefficient, is determined by experience or actual conditions, and span is 0.01~0.1；
5) after cyclic process terminates, the hourage being assigned in each section is recorded according to each OD, inverting is obtained
Including the final speed in each section and link flow as a result,.
Estimate Dynamic iterations method for above-mentioned traffic, further,
Step 4) variance threshold values are set and are adjusted with specific reference to the vehicle number in section or different sections.At this
In inventive embodiments, the variance threshold values are set as 200.
Step 41) the weight distribution principle refers to the shorter car of stroke, its hourage is more reliable, calculates section speed
The weight assigned during the expectation intermediate value of degree is bigger.
Step 42) variance equalization is specifically in each section of its approach when vehicle OD recorded into last iteration
The hourage of distribution equalizes with the desired variance in section.
Compared with prior art, the beneficial effects of the invention are as follows：
Estimated (inverting) for historical traffic road conditions, prior art uses vehicle average speed method, this method mostly
For discreet static method, i.e., by changing within 15 minutes a point time window, window distribution when every track is carried out, then to number in each window
According to being counted, therefore, this method has spatially obliterated the actual speed difference between each section, when using discrete in time
Section, reduces time-varying precision, also fails to accurately conclude each due space-time position of data (vehicle OD) on statistics is calculated
Point.
The present invention is based on traffic flow model knowledge, and traffic shape on road network at that time is inquired into according to the data of crossing vehicles passing in and out
Condition, specifically provides a kind of traffic estimation Dynamic iterations method, the principle of Dynamic iterations method is by Time Continuous, linkage
Ground is iterated calculating, each tracing point of each iteration is adjusted on space-time, setting global optimization target and strict constraint
Condition, so that it is guaranteed that the reasonability and accuracy of reproduction road operation conditions result, can help to monitor road running situation in real time.
On this basis, further forecast analysis future trend situation, assists decision-making.
Brief description of the drawings
Fig. 1 estimates the FB(flow block) of Dynamic iterations method for the traffic that the present invention is provided.
The FB(flow block) for the Dynamic iterations inversion algorithm specific steps that Fig. 2 provides for the present invention.
Fig. 3 is the road network and the schematic diagram of shortest path map in the embodiment of the present invention,
Wherein, the section in 1-circled portion is the vehicle shortest path in embodiment.
Embodiment
Below in conjunction with the accompanying drawings, the present invention, the model of but do not limit the invention in any way are further described by embodiment
Enclose.
The present invention provides a kind of Dynamic iterations method estimated for traffic, and this method utilizes existing charge record
Data, the operation conditions of road is reappeared by Dynamic iterations method, can provide help for monitoring road running situation in real time, herein
On the basis of, can further forecast analysis road it is following move towards situation, play a part of aid decision.
The Dynamic iterations method estimated for traffic specifically includes following steps：
1) driving recording is taken out；
Take out beginning and end (OD) records all in road network；
2) for every OD record, the shortest path that vehicle is travelled in road network is calculated；
For every OD record, the shortest path in road network is calculated according to its starting and terminal point, is made up of multiple sections, is remembered
Make { Section1,Section2,…,Sectionn}；
3) based at the uniform velocity it is assumed that driving states of the inverting vehicle on each section of the shortest path, are obtained based at the uniform velocity false
If each section each moment flow and speed；
According to its starting and terminal point time and at the uniform velocity travel it is assumed that calculate obtain vehicle into each section at the time of
{t1,t2,…,tn, then the flow Flow (Section at respective stretch corresponding momenti,ti) increase by 1, the respective stretch corresponding moment
Speed Speed (Sectioni,ti) statistical sample amount increase by 1；
4) Dynamic iterations inverting is carried out to all vehicle registrations；
For it is all record carry out steps 3) it is described calculating finish after, by formula calculate obtain variance f (x)：
F (x)=min (∑siD(Ti),if i∈Snon-free) (formula 1)
In formula 1, min () represents that optimization aim is minimum value；Snon-freeIt is the atom section of all non-free streams；D
(Ti) it is variance calculation formula D (x)=E (x2)-E(X)2, wherein, x is the average value of the hourage in the section, and X is the section
Hourage set；E (X) is the average value of the hourage in the section；I is the numbering in section；TiI-th of atom section
The hourage vector of upper all OD records；
For the OD vehicle registrations of system-wide net, variance f (x) is calculated according to formula, judges whether variance f (x) meets setting
Threshold value；Threshold value is rule of thumb set and adjusted with practical application；Specifically, threshold value is according to the vehicle number or different in section
There are different selections in section.When calculating obtained variance f (x) less than given threshold, terminate Dynamic iterations inverting, into step
5)；When calculating obtained variance f (x) more than or equal to given threshold, circulated as follows：
41) each section in road network is directed to, (i.e. the shorter car of stroke, its hourage gets over according to weight distribution principle
Reliably, the weight assigned when calculating intermediate value is bigger), calculate the expectation intermediate value for the speed for obtaining the section：
In formula 2, mtiIt is the intermediate value of the hourage in i-th of section；tijThe trip on i-th of section is recorded for j-th of OD
The row time；TjFor the hourage vector of OD records all on j-th of atom section；The vehicle journeys time is longer, is calculating
Shi Quanchong is smaller；
42) for the expectation intermediate value of the section speed obtained in every OD record and previous step, it is according to re-allocation principle
The hourage distributed when vehicle OD is recorded into last iteration in each section of its approach is balanced with the desired variance in section
Change.Variance equalization, which refers to adjust, allows tij to be iterated towards the direction small with mti subtractive：
tij=tij+alpha*(mti-tij) (formula 3)
In formula 3, tijThe hourage on i-th of section is recorded for j-th of OD；mtiIt is the hourage in i-th of section
Intermediate value；Alpha is coefficient, is determined by experience or actual conditions, and span is 0.01~0.1；
5) after cyclic process terminates, the hourage being assigned in each section is recorded according to each OD, inverting is obtained
Including the final speed in each section and link flow as a result,.
The present invention is further described below by example.
Example 1：
It is complete documentation of the vehicle into and out road network that the advantage of charge record data, which is mainly reflected in charge record, interior
Appearance contains vehicle license plate, vehicle, vehicle and enters and leaves the important informations such as time and the website of road network.Simply, it is false
If all vehicles are at the uniform velocity travelled in road network, the shortest path traveling of all vehicle selection origin-to-destinations, then known vehicle
Into, leave website and the time of road network, then can extrapolate the position of the speed and any time vehicle of vehicle in road network
Put, i.e. residing section.The flow and average speed in each section can further be counted.Consequently, it is possible to just can be by road
The charge data reproduction road operation conditions of mouth vehicles passing in and out.
Fig. 3 is a part for Anhui Province's road network, collects the charge record of the part road network, performs following operation：
1) for each vehicle registration, the shortest path in road network is calculated according to its beginning and end；
Shortest path is made up of multiple sections, i.e. { Section1,Section2,…Sectionn, the shortest path has
N section.The method for calculating the shortest path in road network according to beginning and end at present can use dijkstra's algorithm (Di Jie
Si Tela algorithms) acquisition is calculated, as shown in figure 3, the section in circled portion is the shortest path of the vehicle.
2) according to the time of its beginning and end and at the uniform velocity travel it is assumed that to should be in shortest path n section
{Section1,Section2,…Sectionn, at the time of calculating vehicle into each section, it is denoted as { t1,t2,…tn}；Enter one
Step calculates the time for obtaining the car in each section；
Chronomere's precision is minute；In the present embodiment, t1=3min；
Illustrate code as follows：
3) after all record calculating are finished, then into iterative inversion step, Fig. 2 is that Dynamic iterations inversion algorithm is specifically walked
Rapid FB(flow block), it is specific to perform following operation：
31) variance f (x) is obtained by variance calculation formula (equation 1 above)；If variance f (x) value is less than or equal to setting
Threshold value then end operation；
Illustrate code as follows：
In the present embodiment, given threshold is 200, and the value that calculating obtains variance f (x) is 243, and variance f (x) is more than setting threshold
Value, into following circulation, including step 32a)~32b)：
Each section in road network 32a) is directed to, in the expectation that the speed for obtaining the section is calculated according to weight distribution principle
Value；
Weight distribution principle is the shorter car of stroke, and its hourage is more reliable, is calculating the weight assigned during intermediate value
It is bigger.The expectation intermediate value for the speed for obtaining the section is calculated by above-mentioned formula 2.In the present embodiment, calculating obtains first section
The expectation intermediate value of speed be 2min；
32b) for the section speed intermediate value obtained in every OD record and previous step, according to re-allocation principle by vehicle OD
When recording last iteration its institute by way of the distribution of each section the desired variance equalization in hourage and section；
Variance equalization adjusts particular by above-mentioned formula 3, allow travelling of j-th OD record on i-th of section when
Between tij be iterated towards the direction small with the intermediate value mti of the hourage in i-th of section subtractive；In the present embodiment, tij=
3+0.1* (2-3)=2.7.
33) the OD vehicle registrations of system-wide net are directed to, variance f (x) is calculated according to formula 1；See whether f (x) is met less than threshold
The condition of value.It is 190 that this step of the present embodiment, which calculates and obtains f (x), and less than threshold value 200, circulation terminates；
Illustrate code as follows：
34) after cyclic process terminates, the hourage tij being assigned in each section is recorded according to each OD, is obtained
Final speed and link flow.
According to obtained final speed and link flow, the real-time condition in more real road network can be obtained, is
Further analysis traffic conditions provide sound assurance.
It should be noted that the purpose for publicizing and implementing example is that help further understands the present invention, but the skill of this area
Art personnel are appreciated that：Do not departing from the present invention and spirit and scope of the appended claims, various substitutions and modifications are all
It is possible.Therefore, the present invention should not be limited to embodiment disclosure of that, and the scope of protection of present invention is with claim
The scope that book is defined is defined.
Claims (5)
1. a kind of traffic estimates Dynamic iterations method, this method utilizes existing charge record data, by Dynamic iterations side
Method reappears the operation conditions of road, specifically includes following steps：
1) beginning and end OD records all in driving recording, including road network are taken out；
2) for every OD record, calculating obtains the shortest path that vehicle is travelled in road network, is made up of, is denoted as multiple sections
{Section1,Section2,…,Sectionn}；
3) based at the uniform velocity it is assumed that calculating obtains vehicle into { t at the time of each section1,t2,…,tn, inverting vehicle at this most
Driving states on each section of short path, obtain the flow and speed at each section each moment based at the uniform velocity assuming；
4) Dynamic iterations inverting is carried out to all vehicle registrations；
For the OD vehicle registrations of system-wide net, variance f (x) is calculated according to formula 1, judges whether variance f (x) meets the variance of setting
Threshold value；
F (x)=min (∑siD(Ti),if i∈Snon-free) (formula 1)
In formula 1, min () represents that optimization aim is minimum value；Snon-freeIt is the atom section of all non-free streams；D(Ti) be
Variance calculation formula D (x)=E (x2)-E(X)2, wherein, x is the average value of the hourage in the section, and X is the travelling in the section
The set of time；E (X) is the average value of the hourage in the section；I is the numbering in section；TiOwn on i-th of atom section
OD record hourage vector；
When calculating obtained variance f (x) less than the variance threshold values set, terminate Dynamic iterations inverting, into step 5)；Work as meter
When obtained variance f (x) is more than or equal to setting variance threshold values, circulated as follows：
41) each section in road network is directed to, the expectation intermediate value for the speed for obtaining the section is calculated according to weight distribution principle：
In formula 2, mtiIt is the intermediate value of the hourage in i-th of section；tijWhen being that j-th of OD records the travelling on i-th of section
Between；TjFor the hourage vector of OD records all on j-th of atom section；The vehicle journeys time is longer, is weighed when calculating
Weight is smaller；
42) for every OD record and step 41) in the obtained expectation intermediate value of section speed, according to re-allocation principle, pass through
Variance equalization adjustment variance, makes tijTowards with mtiThe small direction of subtractive be iterated：
tij=tij+alpha*(mti-tij) (formula 3)
In formula 3, tijThe hourage on i-th of section is recorded for j-th of OD；mtiIn the hourage for being i-th of section
Value；Alpha is coefficient, is determined by experience or actual conditions, and span is 0.01~0.1；
5) after cyclic process terminates, the hourage being assigned in each section is recorded according to each OD, inversion result is obtained,
Including the final speed in each section and link flow.
2. traffic as claimed in claim 1 estimates Dynamic iterations method, it is characterized in that, step 4) variance threshold values are specific
Set and adjusted according to the vehicle number in section or different sections.
3. traffic as claimed in claim 2 estimates Dynamic iterations method, it is characterized in that, the variance threshold values are set as 200.
4. traffic as claimed in claim 1 estimates Dynamic iterations method, it is characterized in that, step 41) the weight distribution original
The shorter car of stroke is then referred to, its hourage is more reliable, the weight assigned during the expectation intermediate value for calculating section speed is bigger.
5. traffic as claimed in claim 1 estimates Dynamic iterations method, it is characterized in that, step 42) the variance equalization
The hourage distributed when vehicle OD specifically being recorded into last iteration in each section of its approach and the desired side in section
Difference equalization.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510406126.3A CN105006149B (en) | 2015-07-10 | 2015-07-10 | Traffic estimates Dynamic iterations method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510406126.3A CN105006149B (en) | 2015-07-10 | 2015-07-10 | Traffic estimates Dynamic iterations method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105006149A CN105006149A (en) | 2015-10-28 |
| CN105006149B true CN105006149B (en) | 2017-07-21 |
Family
ID=54378802
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510406126.3A Expired - Fee Related CN105006149B (en) | 2015-07-10 | 2015-07-10 | Traffic estimates Dynamic iterations method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105006149B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107993436A (en) * | 2017-11-22 | 2018-05-04 | 思建科技有限公司 | A kind of road condition predicting method and system based on OBD |
| CN110322054B (en) * | 2019-06-14 | 2023-04-28 | 中交第一公路勘察设计研究院有限公司 | Optimized layout method of road section traffic monitor |
| CN110617834B (en) * | 2019-10-31 | 2021-02-26 | 电子科技大学 | Shortest path planning method under Gaussian process road network |
| CN112837542B (en) * | 2020-12-30 | 2022-04-08 | 北京掌行通信息技术有限公司 | Method and device for counting traffic volume of highway section, storage medium and terminal |
| CN113223293B (en) * | 2021-05-06 | 2023-08-04 | 杭州海康威视数字技术股份有限公司 | Road network simulation model construction method and device and electronic equipment |
| CN113954868B (en) * | 2021-10-08 | 2023-04-25 | 南京航空航天大学 | Lane-level path planning method and system based on space-time traffic model |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4847772A (en) * | 1987-02-17 | 1989-07-11 | Regents Of The University Of Minnesota | Vehicle detection through image processing for traffic surveillance and control |
| CN101510357B (en) * | 2009-03-26 | 2011-05-11 | 美慧信息科技（上海）有限公司 | Method for detecting traffic state based on mobile phone signal data |
| WO2011126215A2 (en) * | 2010-04-09 | 2011-10-13 | 고려대학교 산학협력단 | Traffic flow control and dynamic path providing system linked with real-time traffic network structure control based on bidirectional communication function-combined vehicle navigation, and method thereof |
| CN103050009B (en) * | 2013-01-21 | 2015-02-18 | 北京世纪高通科技有限公司 | Method, device and system for providing dynamic traffic information |
-
2015
- 2015-07-10 CN CN201510406126.3A patent/CN105006149B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN105006149A (en) | 2015-10-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105006149B (en) | Traffic estimates Dynamic iterations method | |
| CN102521989B (en) | Dynamic-data-driven highway-exit flow-quantity predicting method | |
| Cortes et al. | General-purpose methodology for estimating link travel time with multiple-point detection of traffic | |
| US7953544B2 (en) | Method and structure for vehicular traffic prediction with link interactions | |
| CN103903430B (en) | Dynamic fusion type travel time predicting method with multi-source and isomorphic data adopted | |
| Singh et al. | Estimation of traffic densities for multilane roadways using a markov model approach | |
| CN107563566B (en) | Inter-bus-station operation time interval prediction method based on support vector machine | |
| CN109272157A (en) | A kind of freeway traffic flow parameter prediction method and system based on gate neural network | |
| CN107085943B (en) | Short-term prediction method and system for road travel time | |
| CN109410587B (en) | Macroscopic traffic flow parameter estimation method for urban expressway | |
| CN106652441A (en) | Urban road traffic condition prediction method based on spatial-temporal data | |
| CN107305742A (en) | Method and apparatus for determining E.T.A | |
| CN109686091B (en) | Traffic flow filling algorithm based on multi-source data fusion | |
| CN102629418A (en) | Fuzzy kalman filtering-based traffic flow parameter prediction method | |
| CN103092076B (en) | Motor train unit braking procedure multi-model self-adapting PID controls | |
| CN108417032B (en) | Analysis and prediction method for roadside parking demand in urban central area | |
| CN105046956A (en) | Traffic flow simulating and predicting method based on turning probability | |
| CN111583628B (en) | Road network heavy truck traffic flow prediction method based on data quality control | |
| CN106899306A (en) | A kind of track of vehicle line data compression method of holding moving characteristic | |
| CN110930693B (en) | Online short-term traffic flow prediction method for road section | |
| CN112633602B (en) | Traffic congestion index prediction method and device based on GIS map information | |
| CN113947899B (en) | Queuing service time dynamic estimation method under low-permeability track data | |
| Mittal et al. | Network flow relations and travel time reliability in a connected environment | |
| CN107123265A (en) | A kind of traffic status of express way method of estimation based on parallel computation | |
| CN110796876A (en) | Road section vehicle total number estimation method based on Kalman filtering |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170721 Termination date: 20210710 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |