Failure analysis of mould hydraulic servo vibration system
Failure analysis of mould hydraulic servo vibration system
It is one of the key technologies in continuous casting to control the mold vibration according to the set waveform. The mold vibration device driven by electro-hydraulic servo system can adjust the amplitude, frequency, waveform and other vibration parameters in real time according to the continuous casting process, which is convenient and efficient. It is the most advanced driving mode at present. This kind of full hydraulic electro-hydraulic servo non sinusoidal oscillation mould is used in the production line of billet caster in a steel plant, and the effect is very good. However, because the electro-hydraulic servo vibration system works near the mold, with high temperature, much dust and small space, it is difficult for technicians to check and test in close range. Therefore, compared with the traditional mechanical vibration system, it has some disadvantages, such as high failure rate, large fault concealment and long fault recovery time, which have a certain impact on production, Therefore, the manufacturer proposes to establish the condition monitoring and fault diagnosis system of mould hydraulic vibration system to ensure the production.
(2) On line condition monitoring of mould hydraulic servo vibration system
The mold hydraulic vibration control system of a steel plant is composed of DYNAFLEX vibration PC, main controller (MCU), slave controller (5C), hydraulic servo system, etc. the working principle is: the continuous casting process parameters are collected by DYNAFLEX vibration PC, and the vibration parameters obtained by calculation are transmitted from the vibration PC to the main controller (MCU) in a cycle of 5S, The master controller then assigns these parameters to the slave controller (casting flow controller 5C), and the slave controller transmits the latest parameters to the servo control unit every 1ms. At the same time, the stroke of the vibration hydraulic cylinder is fed back to the servo amplifier pl6 through the displacement sensor, and the error signal is obtained by comparing with the command signal output from the slave controller, Then the current negative feedback amplifier module on the servo amplifier board pl6 amplifies the power and drives the electro-hydraulic proportional servo valve cylinder to form a closed-loop control system.
In order to monitor the operating parameters of the hydraulic vibration system, the corresponding sensors must be set in the appropriate position to sample the operating parameters of the system. After field investigation and analysis, the distribution scheme of sampling points for condition monitoring data as shown in figure s is designed. The dotted line in figure s is the equipment added to the original system. The principle of signal type selection is sampling the signal voltage of the original system and sampling the signal current and transmission voltage of the new equipment. All signal sampling is completed by data acquisition card after photoelectric isolation, and enters into the fault diagnosis computer, so that the work site and fault diagnosis system can be reliably isolated without mutual influence, and normal production can be guaranteed. For the original system signals (Bosch amplifier pl6 input signal, valve core displacement signal, hydraulic cylinder piston displacement signal), from the voltage point to the isolator, it is converted to the range suitable for the data acquisition card, and then stored in the data base through data acquisition, software filtering and other preprocessing. In addition, in order to improve the accuracy of fault diagnosis, a pressure sensor is added in the second chamber of the original system hydraulic cylinder to sample and analyze the working pressure. The field pressure signal is converted into current and transmitted to the electric control room. It is converted into the range suitable for the data acquisition card through the isolator, and then stored in the database through data acquisition, software filtering and other preprocessing.
After preprocessing, the sampled data is directly displayed on the monitoring screen with real-time waveform and refreshed every 5S. The operator can preliminarily judge the working condition of the system according to the tracking situation of the curve on the screen. When a monitored parameter exceeds the predetermined effective zone, the software can also automatically detect it and immediately change the color of the parameter curve and flash an alarm, Prompt operators to pay attention to or intervene as soon as possible, so as to avoid major failure and affect production.
(3) Establishment of sample database and development of fault diagnosis expert system
① The sample database is established. Some normal and faulty vibration valve controlled cylinders of the steel plant are used to simulate various working conditions in the laboratory. The system parameter curves under various conditions are obtained through the data acquisition system, which are used to form standard samples and form a sample database for the diagnosis expert system as reference standards and training teachers' samples. In addition, the mathematical model of mold vibration electro-hydraulic servo system is established, as shown in Figure t. On the basis of theoretical analysis, different fault phenomena can be simulated by changing the parameters of the relevant links in the mathematical model, and then simulated by digital simulation software, more simulation parameter curves of the system under normal and fault conditions can be obtained, so as to further expand the sample database and make the diagnosis range of the expert system wider.
② The neural network of fault diagnosis expert system can fit the nonlinear function in a large range. The causality of hydraulic servo system fault is a typical nonlinear function. Therefore, based on the mature BP neural network, the fault diagnosis expert system of mold vibration servo system is established. It consists of an input layer, an output layer and an intermediate layer between the input and output layers, forming a three-level diagnosis network with multiple inputs and multiple outputs. Before the fault diagnosis expert system is put into online use, the sample data obtained from the test and simulation are taken as teachers and input into the neural network for training. When the network reaches convergence within the specified tolerance range, the weights and thresholds can be loaded into the fault diagnosis system as the network standard parameters. At this time, the expert system has the ability of on-line real-time diagnosis, which can judge the on-site monitoring parameters with the neural network that has been learned, and identify whether there is a fault in the system. In order to improve the accuracy and universality of fault diagnosis, the expert system will continue to add new samples in the process of using, and repeat training and learning regularly.