Two years ago, I have discussed the way to forecast the traffic condition with Dr. Luan in TRC. I was surprised by the forecast way referred by Luan. But, at that time, we do not have better solution to this problem. In this paper, I will review the forecast methods, redefine the aims of traffic forecast and state one idea before implementing.

Review the forecast methods

Most forecast methods applied by traffic agency to measure the congestion are based on the historical congestion events. Forecasts based on the times and place of traffic congests in history are very coarse and inaccurate, if  no further investigation whether conditions and causes for these traffic congest reexisted again. The difference among these methods implementing this idea exists in the statistic range, like the congestions happened minutes, hours, weeks, months and seasons ago. But, this is not enough. Rear-view mirror only tells u what happened behind, but not tells u what you will encounter in the future, no matter how the mirror is big. In reality, most traffic forecast application summed or average the information on a corridor of routes, at the traffic peak period. That is to say, any such systems that tell you what will happen is looking at the past – a rear-view mirror, big or small.

Redefine the aims of traffic forecast

Usually, Traffic forecast involves forecasts of Annual Average Daily Traffic (AADT), Design Hour Volumes (DHV) and Directional Design Hour Volumes (DDHV). In this paper traffic forecast talks about the possible traffic congestion.

Road users often want to take the direct route with least resistance and best level of services, like smooth driving, comfort, lowest cost and so on. Travelling time still plays an important role among these factors in the process of choosing the route, because safety, comfort and others can be improved in the term of car itself in some extend, but speed limit (by the technology or the traffic rules) exist. What is really annoying are those unexpected delays (usually due to a traffic incident, crash or load spill). We don’t know what is happening or how long it will take before unimpeded passage again! People are scared about uncertainty which resulted in unexpected delays. Of course, these unexpected delays and the derived fear can be compromised by technology and equipment of radio broadcast, telephone, and in-car navigation device.

Traffic congestion will caused by traffic accident and immoderate traffic flow. So the way to forecast the traffic congestion will reflect the trend of possible traffic accident and immoderate traffic flow.

New idea: Ways and data source to traffic forecast

In the previous part, the possible causes to traffic congestion were stated.

In this part, some notes will be elaborated. Then one essential term for substantial traffic forecast was also stated here.

Notes:

预测：交通预测必须能够去预测，而不是展示信息，哪怕这种展示是何等实时或近实时。

Essential term:

交通预测的最大问题，在于其有效性。而且，由于道路本身的特性，其accessibility不是由其Euler 距离来衡量的。比如说，通过某条道路的通行时间，可以通过（道路长度/平均通行速度）来衡量。但其可访问性，是依据道路的出入口来展现的。道路作为一个因此，交通预测的有效性，必须通过得到预测信息后，有能力根据预测信息做出有效的反馈来衡量。比如，你已经进入了一条道路，此时你得到预测信息说前方会堵塞，但是你已经没有选择对这一预测信息做出反馈，你依旧只能前行。这次的预测是失败的。同样，尽管前方有更多的路况，但由于道路等级、路面承重、速度、车型等限制，你没有其他选择。此时，这种预测同样对你而言是失败的。

交通预测的失误率必须低，否则可能引起更大的麻烦。如同地震预测一样，错误的交通预测可能引起原本不会产生的交通阻塞。一旦预测失误，Road Users将会及时转移到相邻的其他道路，导致相邻道路的交通负载突然急剧增加，从而引起原本交通阻塞。而原来被“咒”的道路却有很多通行容量没有实现。

交通阻塞的原因？

交通事故

交通量过大，引起平均速度十分低