Car Impeller Quality Inspection Machine With Machine Vision Image Processing

Car Impeller Quality Inspection Machine With Machine Vision Image Processing

With the development of the automobile industry, the performance of the engine is gradually improved, and the requirements for the stability and reliability of the water pump of the engine cooling system are getting higher and higher. The impeller is an important part of the water pump, which transfers the mechanical energy to the liquid, increases the pressure energy and kinetic energy of the liquid, and realizes cooling and heat exchange. The impeller speed in the automobile water pump is high, the operating conditions change greatly, and the quality requirements of the impeller are quite similar high.

Our advantages

Pic:Two different models of impellers, very similar in appearance:

The customer has successfully produced impellers for many pump manufacturers. There are dozens of models produced on site at the same time, and the process involves the situation that multiple models share equipment. The appearance and ruler pair of multiple types of impellers are very close, and it is easy to be mixed. As below, there are two different types of impellers, which are very similar in appearance.

If the water pump manufacturer installs the wrong type of impeller in the water pump, it will cause the water pump to work abnormally and directly affect the cooling system of the automobile engine. In addition, the impeller will also appear defective products in the production process, such as iron core cracks, time out of the error range, etc., it is very necessary to classify and screen the impeller. At this stage, the customer uses manual screening and classification to determine whether the products are qualified, but the work intensity is high, the accuracy is poor, and the efficiency is low. Therefore, an automated online classification and screening system is urgently needed.

Insepticon principle

The system is combined with an automated mechanism to perform online detection, and the average detection time of each sample does not exceed 1 second. The PC-based system is adopted, the imaging hardware adopts industrial camera and lens, ring light source, and the processing software adopts the self-developed KVIS image processing software. The equipment first arranges the impellers in a queue through the conveyor belt and guide rails, and then pushes the impellers into the visual inspection area through the material distribution cylinder for photo detection. When the detection is completed, the /0 signal is output to the motion control system to execute the reject/pass action.

Since there are many product models on site, and there are unknown models that will be added in the future, in order to meet the adaptive requirements of the system, the system uses machine learning algorithms for feature extraction and classification. Through sampling learning, the accuracy rate increases with the sample size, fully satisfying customers required requirements.