ORP Prediction Based On Dynamic Data Driven

The technology of bio-oxidation gold extraction is the main smelting method for refractory gold ores in the world today,and it has been paid more and more attention because of its advantages such as low investment,simple operation and light environmental pollution.Its role is to use bacteria to wrap the gold particles containing sulfur,arsenic and iron and other elements of mineral impurities oxidation decomposition,so that the gold particles in heavy ore exposed,and thus improve the cyanide process of gold extraction rate.In the process of biological oxidation pretreatment,the biological redox potential is the embodiment of the process and rate of bacterial oxidized ore,and is an important index of process operation in the actual production process,which has a certain relationship with the final gold extraction rate of ore.Therefore,it is of great significance to predict the ORP of the next time in the process of the process of biological oxidation pretreatment and improve the gold extraction rate.Biological oxidation pretreatment has the characteristics of nonlinearity,hysteresis,dynamic and time-varying,and complex production environment and sensor interference,the actual measurement of ORP data has a great deal of non-stationary and volatility;this poses a huge challenge to ORP’s precision prediction.Aiming at the problem of randomness and volatility of ORP,a ORP prediction method based on wavelet analysis and phase space reconstruction is proposed.In view of the dynamic and time-varying problems of biological oxidation pretreatment,Integrating Dynamic Data Driven Application Systems to improve LSSVR and establish a ADAPT_LSSVR dynamic predictive model with self-correcting and updating capability.In order to further improve the accuracy of ORP prediction,this paper improves the wolf swarm algorithm and optimizes the model parameters,and the feedback correction method is used to revise the model predictive output.The main contents of this paper are:1)The background of the thesis research,the development status of the ORP Prediction Research and the current situation of dynamic Data driving research are introduced.2)The types of refractory gold ores,the biological oxidation mechanism of the bio-oxidation gold extraction process,the redox potential,the factors affecting the redox potential and the actual production process of biological oxidation are introduced.3)The paper introduces the theory of wavelet analysis and phase space reconstruction in the process of data preprocessing,expounds the principle and derivation process of the support vector machine,improves the lssvr,and puts forward the ADAPT_LSSVR dynamic prediction model.The prediction process of ADAPT_LSSVR dynamic Prediction model is given.4)The traditional Wolf Swarm algorithm’s thought and calculation process are summarized,the analysis of wolf swarm algorithm wandering behavior,siege behavior and other shortcomings.The learning strategy of teaching and learning is integrated into the Wolf Colony algorithm,and the Wolf Swarm algorithm’s wolf wandering behavior,the fierce Wolf siege behavior and the step size setting are improved.Then the standard test function is used to verify the performance of the improved Wolf swarm algorithm.5)The ORP Dynamic prediction system is developed,and the working process of ORP Prediction system is introduced.Based on the selection of optimization algorithm,the improvement of the Wolf Group algorithm and the filtering,a series of simulation experiments are designed to verify the validity of the ORP precision prediction method.