Study On The Status Assessment Of Continuous Casters And The Fault Trend Prediction Based On Pulling Force

The segment of continuous caster is the core component of continuous caster,and its operation status will directly affect the quality of slab.The running process data of the segment directly reflects its running state.How to use the running process data to evaluate the state of the segment and predict the fault trend is the current research hotspot.In this paper,based on the time series data of tension and straightening force,the method of condition evaluation,fault location and fault prediction of the whole segment of arc slab caster is studied.The main research contents are as follows,First,a preprocessing method for time series of tension and straightening force is proposed.Firstly,aiming at the problems of abnormity and fluctuation in the time series of total tension and straightening force,the data are cleaned and smoothed,and the effects of traditional wavelet denoising and SG smoothing algorithm are compared.Then,aiming at the problem of multi-mode and multi working conditions in the time series of total tension and straightening force,the variation law of the operation data of total tension and straightening force under various working conditions of continuous caster is analyzed,and a mathematical model is proposed for sequence segmentation.Finally,the time domain features are extracted and the feature parameter set is constructed.Second,in this paper,the state evaluation method of the whole sector of continuous caster under two different working conditions is proposed.For the normal casting condition,because the change trend of the two stream total tension and straightening force data is relatively stable and has strong correlation,a fan-shaped whole section condition evaluation method based on AE characteristic curve is proposed.The method uses AE model and sliding window to calculate the eigenvalue of the total tension and straightening force data set,obtains the eigenvalue curve,sets the eigenvalue threshold,and puts forward the total tension and straightening force variation evaluation method.The method of judging the direction of urbanization.Aiming at the problem of small number of abnormal samples and imbalance between the number of abnormal samples and the number of normal samples at the beginning of casting,a two classification model of SVDD is constructed,and the time-domain features are extracted as the input of the model;the model parameters are optimized by using WOA algorithm.The experimental results show that the above method can effectively identify the abnormal state of the whole segment of the continuous caster under various working conditions.Third,a fault location model based on fuzzy clustering and a fault prediction model based on SVR are proposed.Firstly,the time domain features of each sector segment are extracted to obtain the feature pattern matrix,and then the fault location is carried out by combining the fuzzy clustering model and statistical rules.Finally,the sliding window is used to process the time series of the fault sector sectional tension force,and the training and testing samples are constructed,and the linear kernel function support vector machine is used to predict the fault.The experimental results show that the method can effectively identify the fault segment and predict the fault.