We developed a new diagnostic method by using a laser beam. This method is as follows: A tooth surface is irradiated by the zonal laser beam from an oblique direction, and then the irradiated laser beam line is shifted along the surface of the tooth according to gear rotation. If the damage on the irradiated tooth surface exists, the voltage proportional to laser reflection increases. We developed the method to predict and make the reflection benchmark on the normal condition according to the gear surface. To make the benchmark of the diagnosis, the three dimensional basic-data map (x: irradiated angle, y: irradiated distance, z: reflection intensity) was created by measuring the gear only whose material, heat treatment, and roughness were same as the targeted gear. By using the equations of tooth profile and fillet curves calculated from the specifications of the targeted gear, the distance and angle relations between the laser sensor and the tooth surface can be derived. By using the three dimensional basic-data map, the benchmark can be created. The measured reflection data of the non-damage gear agreed well with the benchmark, therefore we can diagnose the various specification gears, if the targeted gear’s material, heat treatment, and roughness are same. Finally, by using the benchmark which was made by our developed method, we proposed a novel diagnosis method. The procedure of the method is as follows:
1) The benchmark is made from the targeted gear’s specifications.
2) To take into account the fluctuation of the benchmark line influenced by the roughness on the gear surface, normal condition area of the reflected data is defined in the range between −0.05 V and +0.05 V of the benchmark line.
3) The normal condition area and measured data is compared, if the measured data is deviated from the normal condition area, there is defined as the abnormal area possible to be damaged.
To confirm the validity of this diagnosis method, the measured value of the damage area with caliper directly and calculated value from the method as mentioned above. The errors of the area and the location were within 20 %. Therefore, the effectiveness of the method using the benchmark data can be confirmed.
