Research on Machine Tool Condition Recognition and

Man-machine Effectiveness Evaluation

Abstract

With the rapid development of science and technology, mechanical equipments are widely used in all walks of life, how to accurately monitor the current health status and the    control performance of equipment, how to evaluate the level of operation of workers, and calculating the cost of production are particularly important for managers to manage the industry scientifically. In this paper,take an ordinary lathe as an example, three methods are used to analyze and process the vibration signal of the lathe to realize the state identification and calculate the time of each working condition of the lathe. According to the real-time vibration signal of the lathe, we can achieve a series of analysis about the operators and the lathe.

In this paper, the signal acquisition object is a certain type of lathe, the research contents include:

(1) To optimize the auditory perception feature set of signals under different operating conditions based on neural network and principal component analysis. Signals of each working condition is processed by the ZCPA auditory model to obtain the auditory spectrum, and then the auditory perception feature is extracted from the auditory spectrum. The weights and thresholds of neural connections in neural networks are optimized by the genetic algorithm, the auditory perception feature set is used as input of the neural network. Through nonlinear mapping of neural networks, a new feature set is obtained. After the new feature set is analyzed by principal component analysis, the projection data of the new feature set in the first principal component direction can be classified.

(2) Using the genetic algorithm to optimize the ZCPA auditory model, the adaptability of the auditory model to various state signals is improved, and the difference between the auditory spectrum generated by each state data is increased. At the same time, through the improvement of the correlation formula, the signal can be recognized in two aspects of shape and amplitude, and the recognition rate is improved.

(3) Status recognition of lathe working process signals based on auditory significant model and information entropy. The vibration signal is processed by the auditory significant model to obtain the global significant graph. According to the fluctuation trend of the global significant graph the impact point of the vibration signal is found. Then, according to the information entropy characteristic of the signal before and after the impact point, the impact point is judged whether it is the switching point of the adjacent state. State recognition is achieved by extracting switching points.

(4) According to the information entropy of the vibration signal, the current health status of the equipment can be determined. According to the calculation of vibration signal significant map energy parameters, fluctuation characteristics, annoyance and other parameters, the workers' level of operation can be judged. Calculating the cost of production according to the time of each state of the lathe.

Key words: ZCPA auditory model; neural network; genetic algorithm; principal component analysis; information entropy; auditory significant model

以上就是部分论文写作范文，如写好论文需要论文查重的小伙伴，可以点击知网论文查重，然后选择相应的论文查重系统。

PS：同学们要先了解自己学校是什么检测系统哦！