The Algorithm Design And The Implementation Of The Hot Box Detection System

The main topic of this paper is the design of algorithm and software on the information processing of the Infrared Hotbox Detection System III (abbreviation as HBDS-III). The information processing of HBDS-III includes four main modules: Counting of Axles and Cars, Temperature Computation, Rolling and Sliding Bearing Distinguishment, Hotbox Forecast. This paper focuses on three of them, only the temperature computation is excluded.A new algorithm of Rolling and Sliding Bearing Distinguishment is provided according to the characteristics of HBDS-III in this paper. The algorithm is based on the characteristic distance after the pretreatment of abnormal temperature wave. It can deal with abnormal wave effectively, including Peak, Tub, Broad, Narrow, Shifting, and so on. The algorithm is standardized and more effective than ever for Rolling and Sliding Bearing Distinguishment.Fuzzy Logic is used in the algorithm of Hotbox Forecast. Fuzzy rules are based on the experience of experts. Hotbox Forecast is realized by the fuzzy inference according to the rules. The new algorithm and software need less parameter to be modified than ever. Hotbox Forecast by fuzzy inference is more accordant to the human judgement.Comparing with automatic Chinese words segmentation, the algorithm of Counting of Axles and Cars is based on the maximum and second-maximum matching. To handle with the lost and excessive of axle distance data, the creation of overlapped axle distance table is needed, and the experience of distribution of axle distance by data statistics is also very important. But there are no good ideas about it when the original axle distance tables are not complete enough.The module of Rolling and Sliding Bearing Detection has been used in the HBDS-III. The software of Hotbox Detection needs more testing, and is currently tested in the laboratory. Counting of Axles and Cars algorithm is accomplished already, and need more particular testing.