Self learning-based coal rock recognition method

Abstract

The invention discloses a self learning-based coal rock recognition method. The method comprises steps: a high-layer structure feature matrix D is firstly learnt from auxiliary data in an offline mode, wherein the auxiliary data are non-labeled non-coal rock natural images and are acquired more easily; the coal rock images are represented as a linear combination of a plurality of high-level characteristic atoms, and the coefficient of the linear combination forms a new eigenvector as the eigenvector of the coal rock image; the extracted coal rock eigenvector is then used to train a classifier;and in a recognition process, an eigenvector of an unknown category of a coal rock image is extracted and inputted to the classifier which completes training, and the category of the coal rock imageis finally outputted. The method uses non-labeled non-coal rock natural images which are acquired easily as training samples, the overhead of marking a large amount of coal rock samples is saved, theextracted coal rock eigenvector has strong discriminability and robustness, and good recognition effects are achieved.

Description

technical field [0001] The invention relates to a coal-rock identification method based on self-study, which belongs to the field of coal-rock identification. Background technique [0002] Coal and rock identification is to use a method to automatically identify coal or rocks. In the coal production process, coal and rock identification technology can be widely used in drum coal mining, tunneling, caving mining, raw coal gangue and other production links. It is of great significance to reduce the labor intensity of workers, improve the working environment, and realize the safe and efficient production of coal mines. [0003] There are many coal and rock identification methods in practical application, including natural gamma ray detection method, radar detection method, stress pick method, infrared detection method, active power monitoring method, vibration detection method, sound detection method, dust detection method, memory detection method, etc. Cutting method, etc., b...

AI Technical Summary

Problems solved by technology

[0004] 1. It is necessary to install various sensors on the existing equipment to obtain information, resulting in complex structure and high cost of the device;

[0005] 2. Equipment such as shearer drums and roadheaders are subject to complex forces, severe vibrations, severe wear, and large dust during the production process. It is difficult to deploy sensors, which may easily lead to damage to mechanical components, sensors, and electrical circuits, and poor device reliability;

[0006] 3. For different types of mechanical equipment, there is a big difference in the optimal type of sensor and the selection of signal pickup points, which requires personalized customization and poor adaptability of the system

[0007] In order to solve the above problems, the image technology-based coal rock recognition method has been paid attention to and researched continuously, and has achieved certain results, including the coal rock recognition method based on wavelet transform, which uses wavelet base to extract coal rock image features for classification and recognition; a The coal and rock recognition method based on dictionary learning uses the dictionary learning method to extract the characteristics of coal and rock images, which can obtain the basis function through learning, and has strong applicability; a coal and rock recognition method based on the asymmetric generalized Gaussian model in the wavelet domain , use the improved relative entropy similarity measure to realize the extraction of coal and rock features, etc. In these coal and rock recognition technologies, the extraction of coal and rock image features is the key factor to obtain a satisfactory recognition rate, and high-quality expression requires the training process There are sufficient samples, but the number of training samples is too small. On the one hand, it cannot truly reflect the characteristic distribution of the coal and rock sample data set, so the coal and rock image features obtained by training are weak in expressive ability, and cannot achieve the ideal recognition effect; on the other hand, There are too few training samples, and the training of the recognition model is easy to overfit, resulting in low generalization performance. However, it is time-consuming and laborious to obtain a large number of labeled coal rock image samples in practical applications.

In the existing semi-supervised learning, when the number of existing labeled training samples is insufficient, it is necessary to add a large number of unlabeled sample images of the same type to assist training on the basis of the training sample data set with class labels, saving the cost of labeled samples. overhead, but for sample images that are difficult to obtain such as coal and rock, semi-supervised learning still has certain limitations; and in transfer learning, it is required that the source domain samples and the target domain have certain common characteristics, and there are certain limitations in actual operation. Difficulties

Images