Author
László Ketskeméty
Mini Project ID
BMEMPLOAD3
Description
The case: Illegal overweighting of heavy goods vehicles is a major problem throughout Europe. Heavy goods vehicles, buses and coaches transporting goods in Europe must comply with certain rules on weights and dimensions for road safety reasons, and to avoid damage to roads, bridges and tunnels. Besides monitoring the vehicles using WIM (Weighting In-motion) or the OBW (On Board Weighting) systems, these data would feed into a central European database to monitor the total European road network, forecast deterioration, plan maintenance / renovation, and the ideal allocating of central resources.
The below solution to the problem must be explained in writing:
- The existing WIM network should be extended in such a way the extended network can form a representative sample of an appropriate size in the road section population.
- The set of road sections need to be separated into homogeneous parts by clustering based on the road section database.
- A representative sample then can be formed by stratified sampling, while the existing stations shall be considered as part of the sample. New measuring point installations must be planned for the “missing” road sections.
Sector
VET
Data
N/A
Model
You would like to argue in favour of the solution presented, however you have to present the arguments that you based decision on, otherwise they seem unfounded. When presenting the solution, the following arguments have to be made:
- There are several possible solutions to the weighting and monitoring problem described in the description. According to one, all existing stations should all be included in the planned sample. This solution can only be implemented if there are no high-density areas in the network. In such cases, only a sample with a large number of items could ensure representativeness. If there are high-density areas, the measuring point elements to be included in the sample must be selected at random from that part. Unselected points can be used to check the estimate.
- There are several techniques for clustering. Either dynamic or hierarchical techniques can be good. It is essential to choose a good metric function. In addition to the speed of execution, which algorithm reproduces an expert separation (the training set) with the highest accuracy may be decisive in the selection.
- By stratified sampling of clusters, it is expedient to select the sampling points so that the ratios in the sample correspond to the ratios between the clusters. The appropriate number of elements from the clusters should be formed by simple random selection.
Calculation
N/A
Review M3S2 – Measuring the on-board weighting of heavy vehicles.
You must be logged in to post a review.