Predictive Method for Traffic Flow of Elevator Systems Based on Neural Networks

Huang, Min; Xu, Lin; Wang, Jianhui; Gu, Shen

doi:10.1109/icca.2003.1595109

Cited by 4 publications

(4 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The long-term statistics can also be used to estimate a forecasting model that tries to capture the dynamics of the traffic and to forecast the future based on recent observations in the short-term statistics. There are several learning and forecasting methods (e.g., Siikonen 1997, Powell et al 2000, Huang et al 2003, Xiang et al 2005, Luo et al 2005, Yan et al 2006, Imrak andÖzkirim 2006, Jianru et al 2007). …”

Section: Theoretical Foundationmentioning

confidence: 99%

The Elevator Trip Origin-Destination Matrix Estimation Problem

Kuusinen¹,

Sorsa²,

Siikonen³

2015

Transportation Science

View full text Add to dashboard Cite

This dissertation studies people flow in buildings, especially the process of how passengers arrive at elevator lobbies, estimation of elevator passenger traffic, and human behaviour and decision making in evacuations. The arrival process is studied by taking into account, for the first time, that passengers do not always arrive and use elevators individually but rather in batches. The results suggest that the common assumption that individual arrivals follow a Poisson distribution may not hold when the proportion of batch arrivals is large.To estimate the elevator passenger traffic in a building, new mathematical models and algorithms are developed. The new methods are based on mathematical optimization, namely, linear programming, integer least squares and constraint programming. The results from numerical experiments show that the new approaches satisfy real-time elevator control requirements. In addition, randomized algorithms result in better quality passenger traffic statistics than traditional deterministic algorithms. The dissertation presents also an experimental evacuation study. The results show that people may not be able to select the fastest exit route and that cooperation may slow down the evacuation.The new estimation models and algorithms presented in this dissertation enable better elevator control and some of them are already being implemented by KONE Corporation. The results also give new insights into the process of how passengers arrive at the elevator lobbies and use elevators and into human behaviour in evacuation situations, which affect elevator and building safety planning. Tiivistelmä Tässä väitöskirjassa tutkitaan rakennusten henkilöliikennettä, erityisesti henkilöiden saapumisprosessia hissiauloihin, hisseillä tapahtuvan henkilöliikenteen estimointia sekä henkilöiden käyttäytymistä ja päätöksentekoa rakennuksen evakuoinnissa. Saapumisprosessia tutkitaan ottaen ensimmäistä kertaa huomioon, että henkilöt eivät aina saavu ja käytä hissiä yksin vaan isommissa joukoissa. Tulokset osoittavat, että yleinen olettamus yksittäisten saapumisten Poisson jakautuneisuudesta ei välttämättä päde, kun joukkosaapumisten osuus on suuri.Rakennuksen henkilöliikenteen estimoimiseksi väitöskirjassa kehitetään matemaattisia malleja ja algoritmeja, jotka perustuvat lineaariseen optimointiin, pienimmän neliösumman kokonaislukuoptimointiin ja rajoiteoptimointiin. Numeeristen testien tulokset osoittavat, että menetelmät toteuttavat todellisen hissiohjauksen reaaliaikavaatimukset. Lisäksi satunnaistetut algoritmit tuottavat laadultaan parempia henkilöliikennetilastoja kuin perinteiset deterministiset algoritmit. Väitöskirjassa esitetään myös kokeellinen evakuointitutkimus, jonka tulokset osoittavat, että ihmiset eivät välttämättä kykene valitsemaan nopeinta poistumisreittiä ja että evakuoitavien pyrkimys yhteistyöhön voi hidastaa poistumista.Väitöskirjassa kehitetyt henkilöliikenteen estimointimallit ja -algoritmit mahdollistavat hissien tehokkaamman ohjaamisen ja osa menetelmistä on jo tuotannossa KONE...

show abstract

Section: Theoretical Foundationmentioning

confidence: 99%

The Elevator Trip Origin-Destination Matrix Estimation Problem

Kuusinen¹,

Sorsa²,

Siikonen³

2015

Transportation Science

View full text Add to dashboard Cite

show abstract

“…Since constraints (2)- (4) are only imposed on low-level decision variables, , the formulation above can be further converted into 1 (9) where (10) From (9) and (10), a two-level optimization framework can be naturally derived. Given a high-level decision variable , i.e., , the low-level optimizes the single car dispatching problems (10) for individual cars and provide the optimized performance to the high-level for the calculation of the objective function value in (9).…”

Section: A Two-level Optimization Frameworkmentioning

confidence: 99%

“…Given a high-level decision variable , i.e., , the low-level optimizes the single car dispatching problems (10) for individual cars and provide the optimized performance to the high-level for the calculation of the objective function value in (9). Based on the evaluation from the low-level, the high-level is then to optimize passenger-to-car assignment (9).…”

Section: A Two-level Optimization Frameworkmentioning

confidence: 99%

“…The latest advancements in sensor technology and information technology further open up the possibility to collect future traffic information within a certain time window [3]- [6]. Beyond such information, statistical data and/or statistical forecasts have been exploited to create virtual future passenger traffic [7]- [9]. More information enables the development of elevator systems with better performance.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Optimization of Group Elevator Scheduling With Advance Information

Sun

Zhao

Luh

2010

IEEE Trans. Automat. Sci. Eng.

View full text Add to dashboard Cite

Abstract-Group elevator scheduling has received considerable attention due to its importance to transportation efficiency for mid-rise and high-rise buildings. One important trend to improve elevator systems is to collect advance traffic information. Nevertheless, it remains a challenge to develop new scheduling methods which can effectively utilize such information. This paper is to solve the group elevator scheduling problem with advance traffic information. This problem is difficult due to various traffic patterns, complicated car dynamics, and combinatorial explosion of the search space. A two-level formulation is developed with passenger-to-car assignment at the high-level and single car dispatching that is innovatively formulated as passenger-to-trip assignment at the low-level. Detailed car dynamics are embedded in simulation models for performance evaluation. Taking advantage of advance information, a new door action control method is suggested to increase the flexibility of elevators. In view of the hierarchical problem structure, a two-level optimization framework is established. Key problem characteristics are exploited to develop an effective trip-based heuristic for single car dispatching, and a hybrid nested partitions and genetic algorithm method for passenger-to-car assignment which can be extended to solve a generic class of sequential decision problems. Numerical results demonstrate solution quality, computational efficiency, benefit of advance information and the new door action control method, and values of new features in our hybrid method. Q. Note to Practitioners-This

show abstract

Estimation Of Passenger Waiting Time In Elevator Systems With Artificial Neural Network

Dursun¹

2010

Intelligent Automation & Soft Computing

View full text Add to dashboard Cite

Predictive Method for Traffic Flow of Elevator Systems Based on Neural Networks

Cited by 4 publications

References 7 publications

The Elevator Trip Origin-Destination Matrix Estimation Problem

The Elevator Trip Origin-Destination Matrix Estimation Problem

Optimization of Group Elevator Scheduling With Advance Information

Estimation Of Passenger Waiting Time In Elevator Systems With Artificial Neural Network

Contact Info

Product

Resources

About