1. Introduction to the background of the sand control test

With the emergence of electronic computers in the middle of the twentieth century, science and technology have developed rapidly. Artificial intelligence has been applied to many areas of our work and life. With the development of research, artificial intelligence will affect our work and life more deeply. The main content of artificial intelligence research is: knowledge representation, machine learning and knowledge acquisition, intelligent robots, computer vision, automatic reasoning and search methods, natural language Comprehension, knowledge processing systems and self-sending programming, and computer vision is an important field of artificial intelligence.

Computer vision is to allow computers to have the same observation and recognition capabilities as human eyes. Furthermore, it is to use cameras and computers to replace human eyes to identify, track and measure targets, and further image processing. Computer vision mainly uses computers to simulate human visual functions, extract information from objective images, process and understand them, and finally apply them to actual detection, measurement and control. The biggest characteristics of computer vision technology are fast speed, large amount of information, and multiple functions.

At present, computer vision is mainly used in security cameras, traffic cameras, unmanned driving, unmanned aerial vehicles, finance, medical care, etc. We can also use computer vision technology in engine sand prevention tests. The hazards caused by the dust in the air being sucked into the engine are: large particles of sand and gravel can damage the blades, small particles of sand and dust will abrade the blades, deposit on the shaft surface and destroy the balance of the rotor, and in severe cases, it will also cause the engine to be damaged. Shutdown in the air caused a major accident.

2. Sand control test and image processing technology

Take the high-speed camera recording particle movement as an example to introduce the sand control test process. First, use a high-speed camera to record the movement of the particles in the flow channel test piece, and then perform image processing and analysis on the video to obtain the particle's movement trajectory, incident velocity, exit velocity, collision incident angle and exit angle movement characteristics (for the future Further study the motion law of the gas and solid two-phase flow in the pipeline to provide reference), and then adjust the particle motion model and parameter adjustment, parameter estimation or parameter optimization to minimize the error between the simulated output value of the model and the actual observation value, which is a follow-up prevention Sand provides a numerical basis for optimizing the flow path and improving the sand control performance of the engine.

The anti-sand test device is composed of a steady flow section, a test section, a drainage section and a high-speed camera. The high-speed camera is used to take pictures of the test section. In order to facilitate the later image processing work, reduce the influence of background noise. When using high-speed cameras, some optical measurement problems need to be solved: first, the light source needs to use a high-power light source with a certain light-gathering ability to improve the imaging effect of sand and dust; second, in order to prevent the background plexiglass from affecting the light Background noise caused by the reflection of the background, it is necessary to paste a black paper with a certain light-absorbing ability on the background plexiglass to reduce the reflection of the ground.

The Phantom V411 high-speed camera is used for the measurement of the sand control test. Its pixel is 1 million pixels, the highest resolution is 1280*800, the shooting rate is 4200 frames per second, and the image format is Cine, Cine Compressed and CineRaw. The analysis process of particle movement characteristics is as follows:

1. Import: Video files shot by high-speed cameras

2. Recognition: Recognize particles based on the difference method between consecutive adjacent frames, and perform continuous tracking to obtain their motion trajectory

3. Add collision boundary: collision boundary (automatic recognition and manual recognition)

4. Calculate characteristic parameters: position, velocity (change in displacement), acceleration (change in velocity), collision incident angle and exit angle

The key technologies that need to be used in the sand control test are: particle identification, particle tracking, and particle motion characteristic parameters. Automatic target recognition technologies include: classic statistical pattern recognition methods, knowledge-based automatic target recognition methods, model-based automatic target recognition methods, automatic target recognition methods based on multi-sensor information fusion, targets based on artificial neural networks and expert systems Automatic identification method. The common features of automatic target recognition technology include complexity, aspect ratio and compactness, and other features include corners, moments and textures. The basic methods of motion detection include inter-frame difference method, background difference method and optical flow method. Moving target tracking algorithms include correlation-based tracking method, feature-based tracking method and model-based tracking method. The key technologies that need to be used in the sand control test are: particle identification, particle tracking, and particle motion characteristic parameters. Automatic target recognition technologies include: classic statistical pattern recognition methods, knowledge-based automatic target recognition methods, model-based automatic target recognition methods, automatic target recognition methods based on multi-sensor information fusion, targets based on artificial neural networks and expert systems Automatic identification method. The common features of automatic target recognition technology include complexity, aspect ratio and compactness, and other features include corners, moments and textures. The basic methods of motion detection include inter-frame difference method, background difference method and optical flow method. Moving target tracking algorithms include correlation-based tracking method, feature-based tracking method and model-based tracking method.

3. Image analysis conclusions and prospects

Judging from the results of the sand control test, the following conclusions can be drawn from the analysis:

1. As the air flow rate in the pipeline increases, the particle separation efficiency is correspondingly improved. This is because the inertial force component of the particles on the air streamline increases, and the particles obtain sufficient inertial force to be thrown into the clearing channel.

2. The velocity angle of the inlet particles has a great influence on the separation effect. The larger the velocity angle of the particle inlet, the more the particles collide with the wall and the smaller the probability of being separated.

In summary, the effect of gas-solid separation in bifurcated pipelines depends on many factors, including:

The size and density of solid particles

The elastic modulus of the inner wall material of the bifurcated pipe

The position, velocity and direction of the solid particles entering the pipeline

The air velocity in the pipe and the accelerated movement of the carried particles are one of the most important factors affecting its inertial force and separation effect.

When the particles do not collide at low concentrations, as the volume component of the added solid particles increases, the collision between particles can not be ignored, leading to many unpredictable results, which greatly affect the separation effect.