A method for estimating parameters of cobalt removal process in hydrometallurgy of zinc based on UKF theory

Abstract

The present invention discloses a method for estimating parameters of a cobalt removal process in a hydrometallurgical zinc smelting process based on UKF theory. The parameter estimation described in the method includes the following steps: 1) collecting data and preprocessing data; 2) establishing a model; 3) constituting a model parameter estimation algorithm for a cobalt removal process; 4) performing simulation to find the estimated optimal parameters; 5) according to the discretized state equation and measurement equation, adopting a proportional correction symmetric sampling strategy to obtain a UKF recursive formula for estimating unknown parameters; 6) verifying the validity of the parameter estimation model through numerical calculation and comparison with the corresponding output data in the actual production process. The present invention can accurately estimate the unknown parameters in the cobalt removal process model and predict the ion concentration. The prediction accuracy is high and the method is effective.

Description

Technical Field

[0001] The invention relates to the technical field of cobalt removal process in hydrometallurgy, in particular to the field of cobalt removal process parameter estimation, and specifically to a method for estimating cobalt removal process parameters in hydrometallurgy based on UKF theory. Background Art

[0002] Zinc and its compounds are important and irreplaceable non-ferrous metal resources in modern industry, and are widely used in electroplating, replacement smelting, alloys, anti-corrosion and pharmaceutical industries. Hydrometallurgy is the main production method of zinc, accounting for more than 80% of the world's zinc production. In addition to the main metal element zinc, zinc minerals also contain other impurity metal elements. During the leaching process, these impurity metal ions will precipitate into the leachate along with zinc ions under the action of strong acid. Their presence will not only reduce the quality of the product, but also cause a significant decrease in the current efficiency of the electrolysis process. In severe cases, it may even cause plate burning, threatening production safety. Therefore, the solution must be purified before electrolysis to reduce the concentration of these impurity ions to within the range required by the process indicators, so as to provide a high-purity metal electrolyte for the subsequent electrolysis process.

[0003] The process of removing cobalt from arsenic salt is an important part of hydrometallurgy. Cobalt ions are not only the most difficult to remove, but also the most harmful impurities in the electrolysis process. Its concentration, as a key indicator of the purification process, not only determines the consumption of zinc powder, but also directly affects the electrolysis efficiency and product quality. The cobalt ions in the solution are mainly removed by adding zinc powder for replacement, but due to the complex process mechanism and many influencing factors, especially the impurity ion concentration cannot be detected online, which makes the process optimization and control difficult and the consumption of zinc powder high.

[0004] The main production equipment for the cobalt removal process is a continuous stirred tank reactor (CSTR). However, a single reactor cannot directly reduce the excess cobalt ions in the leachate to within the process requirements, so it is necessary to remove cobalt in four reactors cascaded from high to low. The outlet ion concentration of the previous reactor is the inlet ion concentration of the next reactor. The cobalt ion concentration at the outlet of each reactor needs to be reduced in turn as required to ensure that the cobalt ion content in the zinc sulfate solution is reduced to within the process index requirements after leaving the last reactor.

[0005] There are other impurity ions such as nickel ions and copper ions in the cobalt removal process, which cannot be ignored. In particular, copper ions act as a catalyst in the process, creating conditions for the smooth removal of cobalt and nickel reactions.

[0006] In view of the current problems in the detection and prediction of impurity ion concentration in the impurity removal process, it is necessary to conduct in-depth research on the process and reaction mechanism of cobalt removal. According to the characteristics of differences, reflux and mutual influence of cascade reactors, the influence of other impurity ions in the solution on cobalt ions is considered, and a dynamic synergistic CSTR (Synergistic Continuously Stirred Tank Reactor, SCSTR) model is established to describe the cooperative relationship of cascade reactors.

[0007] In addition to the layout of the model structure, the parameter estimation of the cobalt removal kinetic model is also an important aspect that needs to be considered. The electrochemical reaction of various ions in the cobalt removal process is a highly nonlinear process. The established SCSTR model has strong nonlinearity, which leads to great difficulties in determining the model parameters. Therefore, establishing a collaborative SCSTR model of the cobalt removal process and estimating its unknown parameters to predict the reactor outlet ion concentration is of great significance for adjusting the blindness of operating parameters and improving the cobalt removal efficiency. Summary of the invention

[0008] The purpose of the present invention is to propose a method for estimating the parameters of the cobalt removal process in hydrometallurgy based on UKF theory, to effectively estimate the model parameters of the cobalt removal process, to improve the accuracy of the estimation of the model parameters of the cobalt removal process, to achieve the purpose of predicting the outlet ion concentration, and to solve the problem that the existing prediction model for the outlet ion concentration of the cobalt removal process contains unknown parameters, which makes the prediction model unable to be used normally.

[0009] To achieve the above object, the present invention provides a method for estimating parameters of a hydrometallurgical zinc smelting cobalt removal process based on UKF theory, the method comprising the following steps:

[0010] 1) Collect sample data;

[0011] Preprocess sample data;

[0012] 2) Establishing a dynamic model of the cobalt removal process, i.e., a set of nonlinear differential equations;

[0013] 3) parameter estimation algorithm constituting the dynamic model of the cobalt removal process;

[0014] 4) using the iterative Euler method in the numerical integration method to perform a discretized numerical solution on each state of the dynamic model of the cobalt removal process to obtain a discretized state equation and a measurement equation;

[0015] 5) Based on the discretized state equation and measurement equation, a proportionally corrected symmetric sampling strategy is adopted to obtain the UKF recursive formula for estimating unknown parameters;

[0016] 6) Use the UKF recursive formula to calculate the state filter value x a(k), thereby obtaining the estimated values ​​of the unknown parameters in the dynamic model of the cobalt removal process. The effectiveness of the dynamic model of the cobalt removal process is verified by comparing the estimated values ​​of the unknown parameters with the corresponding output data in the actual production process.

[0017] The collected sample data mainly consists of inlet solution flow, underflow solution flow, reaction volume, waste acid flow, arsenic salt flow, reactor inlet ion concentration, zinc powder dosage, redox potential, reactor temperature, and reactor outlet ion concentration;

[0018] The preprocessing of sample data comprises the following steps:

[0019] 1) Use the 3σ criterion statistical discrimination method to process abnormal data;

[0020] 2) Use cubic spline interpolation and polynomial fitting methods to fill in the missing values ​​of field data;

[0021] The statistical discrimination method of the 3σ criterion has the following steps:

[0022] The sample data is Y = {y1,y2,…,y n}, the mean is The deviation is Then the standard deviation is obtained according to the Bessel formula:

[0023]

[0024] If a sample data y i Deviation of |e i |>3σ, the data is considered unreasonable and should be removed from the sample data; the comparison is repeated in this way until all sample data are tested;

[0025] The method of cubic spline interpolation and polynomial fitting, the steps are as follows:

[0026] Take the mean of the two methods of cubic spline interpolation and polynomial fitting to fill;

[0027] After processing abnormal data and field data with some missing values, an initial sample set of unknown parameter estimates is obtained.

[0028] The dynamic model of the cobalt removal process, i.e., a set of nonlinear differential equations, is established as follows:

[0029]

[0030]

[0031] Where: Q in , Q underare the liquid flow after copper removal and the underflow flow after recovery respectively; V is the reaction volume; is the concentration of impurity ions in the solution after copper removal; the superscript i (i = 1, 2, ..., 4) represents the reactor # i; the subscript j represents the type of impurity ions j = 1: cobalt ions; j = 2: nickel ions, such as represents the concentration of impurity ion j in reactor #i; is the ORP value, which represents the redox capacity of the solution and is controlled by the amount of zinc powder added; is the frequency factor; E e is the standard activation energy; is the factor affecting the cathode activation energy due to the change in electrode potential; F is the Faraday constant; e eq,j is the equilibrium potential; and is an approximation of the redox potential and the mixing potential; R is the ideal gas constant; T c is the reaction temperature; i is the mass-reaction surface coefficient; is the mass of zinc powder used to replace impurity ion j in reactor #i.

[0032] Parameter estimation algorithm of the dynamic model of the cobalt removal process:

[0033] The cobalt ion concentration at each reactor outlet is used as the state variable to estimate the unknown parameters. The state variable is selected as The control variable is in represents the zinc powder addition rate of each reactor, and the output variable is the same as the state variable; the unknown parameter χ=[χ 1 ,x 2 ,x 3 ,x 4 ] as the state variable, and establish the augmented state equation and observation equation:

[0034]

[0035] y a (t) = H a x a (t)+v(t);

[0036] In the formula, is a nonlinear function; w(t) and v(t) are both zero-mean white noise sequences, and their covariance matrices are Q and R respectively;

[0037]

[0038] therefore:

[0039]

[0040] The iterative Euler method in the numerical integration method is used to perform a discretized numerical solution on each state of the dynamic model of the cobalt removal process to obtain a discretized state equation and a measurement equation. The steps are as follows:

[0041]

[0042] χ 1 (k+1)=χ 1 (k)+dχ 1 (k) step_length

[0043] χ 2 (k+1)=χ 2 (k)+dχ 2 (k) step_length

[0044] χ 3 (k+1)=χ 3 (k)+dχ 3 (k) step_length

[0045] χ 4 (k+1)=χ 4 (k)+dχ 4 (k) step_length;

[0046] In the formula, step_length is the step length used, and its value is 0.001. The discrete state equation and measurement equation obtained by numerical solution method are:

[0047] x a (k+1)=f a (x a (k),u(k),w(k),k)

[0048] y a (k) = H a x a (k)+v(k).

[0049] The proportionally corrected symmetric sampling strategy includes the following:

[0050] 1) Sigma sampling points and weight calculation:

[0051]

[0052]

[0053] In the formula, {φ i}(i=0,1,…,2n) is a set of Sigma sampling points obtained; and is the mean and variance; λ=α 2 (n+κ)-n, κ is the proportional factor, α is the scaling factor, which is used to control the distribution state of the sampling points; and are the corresponding mean weights and variance weights; β reflects the high-order characteristics of state history information;

[0054] 2) Initialize state estimation and estimated variance:

[0055]

[0056] In the formula, and is the initial state estimate and estimated variance of the filter;

[0057] 3) Time update:

[0058]

[0059] In the formula, From the sampling point through the nonlinear state function f a (·) delivered; and is the mean and variance of the one-step forecast;

[0060] 4) Measurement update:

[0061]

[0062] In the formula, {ζ i}(i=0,1,…,2n) is the 2n+1 Sigma sampling points obtained; From these sampling points through H a The transmission is obtained; and One-step forecast means, variances, and cross-covariances for the measured variables:

[0063]

[0064] In the formula, is the measured value at time k+1; is the filter gain; and is the state estimate and estimated variance at time k+1.

[0065] When estimating unknown parameters, set the initial values ​​of relevant parameters: Q k =0.001*eye(8), R k=0.01*eye(4), α=0.01,κ=0,β=2;After multiple recursive calculations, each model parameter can converge to a certain value and estimate the unknown parameters;

[0066] The estimated unknown parameters are obtained through numerical calculation and compared with the corresponding output data in the actual production process to verify the effectiveness of the parameter estimation model.

[0067] The method for estimating process parameters of cobalt removal in hydrometallurgy of zinc based on UKF theory described in the present invention has the beneficial effect that: for complex industrial processes of cobalt removal, since there is a complex unknown functional relationship between the process data collected by the sensor and the outlet ion concentration of the reactor, the established mechanism model contains unknown parameters and has strong nonlinearity. The present invention fully considers the nonlinearity of the cobalt removal process and the dynamic characteristics of the process, and estimates the unknown parameters in the model through industrial process data based on UKF theory to improve the credibility of the model. Compared with the prior art, the method of the present invention is simple to implement and easy to operate. Using the cobalt removal process model and the estimated unknown parameters in the present invention, the predicted outlet ion concentration is close to the measured value, with higher accuracy, and can achieve the effect of efficient cobalt removal. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a flow chart;

[0069] Figure 2 Taking the collected cobalt ions at the reactor outlet as an example, the cubic spline interpolation and polynomial fitting methods are used to fill in the missing values ​​of the field data;

[0070] Figure 3 is the convergence distribution state of each model parameter of the reactor;

[0071] Figure 4 It is a comparison parameter between the estimated and measured cobalt ion concentrations at the outlet of each reactor. DETAILED DESCRIPTION

[0072] Example 1

[0073] The method for estimating parameters of the hydrometallurgical zinc removal process based on UKF theory is as follows: Figure 1 As shown, the method comprises the following steps:

[0074] 1) Collecting sample data; the collected sample data mainly consists of inlet solution flow, underflow solution flow, reaction volume, waste acid flow, arsenic salt flow, reactor inlet ion concentration, zinc powder dosage, redox potential, reactor temperature, and reactor outlet ion concentration;

[0075] During normal production, the inlet solution flow and underflow solution flow will not fluctuate too much, so they are considered constant; the reaction volume V is also constant; compared with the inlet solution flow and underflow solution flow, the waste acid flow and arsenic salt flow are both small, so they are ignored when establishing the equation;

[0076] The preprocessing of sample data includes the following:

[0077] 1) Use the 3σ criterion statistical discrimination method to process abnormal data;

[0078] 2) Use cubic spline interpolation and polynomial fitting methods to fill in the missing values ​​of field data;

[0079] The statistical discrimination method of the 3σ criterion has the following steps:

[0080] The sample data is Y = {y1,y2,…,y n}, the mean is The deviation is Then the standard deviation is obtained according to the Bessel formula:

[0081]

[0082] If a sample data y i Deviation of |e i |>3σ, the data is considered unreasonable and should be removed from the sample data; the comparison is repeated in this way until all sample data are tested;

[0083] The method of cubic spline interpolation and polynomial fitting, the steps are as follows:

[0084] Take the mean of the two methods of cubic spline interpolation and polynomial fitting to fill;

[0085] After processing the abnormal data and the field data with some missing values, the initial sample set of unknown parameter estimation values ​​is obtained;

[0086] 2) Establish a dynamic model of the cobalt removal process;

[0087]

[0088] Where: Q in , Q under are the liquid flow after copper removal and the underflow flow after recovery respectively; V is the reaction volume; is the concentration of impurity ions in the solution after copper removal; the superscript i (i = 1, 2, ..., 4) represents the reactor # i; the subscript j represents the type of impurity ions j = 1: cobalt ions; j = 2: nickel ions, such as represents the concentration of impurity ion j in reactor #i; is the ORP value, which represents the redox capacity of the solution and is controlled by the amount of zinc powder added; is the frequency factor; E e is the standard activation energy; is the factor affecting the cathode activation energy due to the change of electrode potential; F is the Faraday constant; e eq,j is the equilibrium potential; and is an approximation of the redox potential and the mixing potential; R is the ideal gas constant; T c is the reaction temperature; i is the mass-reaction surface coefficient; is the mass of zinc powder used to replace impurity ion j in reactor #i;

[0089] 3) An algorithm for estimating model parameters constituting the cobalt removal process;

[0090] Parameter estimation algorithm for the dynamic model of cobalt removal process:

[0091] The cobalt ion concentration at each reactor outlet is used as the state variable to estimate the unknown parameters. The state variable is selected as The control variable is in represents the zinc powder addition rate of each reactor, and the output variable is the same as the state variable; the unknown parameter χ=[χ 1 ,x 2 ,x 3 ,x 4 ] as the state variable, and establish the augmented state equation and observation equation:

[0092]

[0093] y a (t) = H a x a (t)+v(t);

[0094] In the formula, is a nonlinear function; w(t) and v(t) are both zero-mean white noise sequences, and their covariance matrices are Q and R respectively;

[0095]

[0096] therefore:

[0097]

[0098] 4) The iterative Euler method in the numerical integration method is used to perform a discretized numerical solution on each state of the dynamic model of the cobalt removal process to obtain a discretized state equation and a measurement equation. The steps are as follows:

[0099]

[0100] χ 1 (k+1)=χ 1 (k)+dχ 1 (k) step_length

[0101] χ 2 (k+1)=χ 2 (k)+dχ 2 (k) step_length

[0102] χ 3 (k+1)=χ 3 (k)+dχ 3 (k) step_length

[0103] χ 4 (k+1)=χ 4 (k)+dχ 4 (k) step_length;

[0104] In the formula, step_length is the step length used, and its value is 0.001. The discrete state equation and measurement equation obtained by numerical solution method are:

[0105] x a (k+1)=f a (x a (k),u(k),w(k),k)

[0106] y a (k) = H a x a (k)+v(k);

[0107] 5) Based on the discretized state equation and measurement equation, a proportionally corrected symmetric sampling strategy is adopted to obtain the UKF recursive formula for estimating unknown parameters;

[0108] The proportionally corrected symmetric sampling strategy includes the following:

[0109] ①Sigma sampling points and weight calculation:

[0110]

[0111]

[0112] In the formula, {φ i}(i=0,1,…,2n) is a set of Sigma sampling points obtained; and is the mean and variance; λ=α 2 (n+κ)-n, κ is the proportional factor, α is the scaling factor, which is used to control the distribution state of the sampling points; and are the corresponding mean weights and variance weights; β reflects the high-order characteristics of state history information;

[0113] ② Initialize state estimation and estimated variance:

[0114]

[0115] In the formula, and is the initial state estimate and estimated variance of the filter;

[0116] ③Time update:

[0117]

[0118] In the formula, From the sampling point through the nonlinear state function f a (·) delivered; and is the mean and variance of the one-step forecast;

[0119] ④Measurement update:

[0120]

[0121] In the formula, {ζ i}(i=0,1,…,2n) is the 2n+1 Sigma sampling points obtained; From these sampling points through H a The transmission is obtained; and One-step forecast means, variances, and cross-covariances for the measured variables:

[0122]

[0123] In the formula, is the measured value at time k+1; is the filter gain; and is the state estimate and estimated variance at time k+1;

[0124] 6) Use UKF recursive formula to calculate the state filter value x a(k), thereby obtaining the estimated values ​​of the unknown parameters in the dynamic model of the cobalt removal process. The effectiveness of the dynamic model of the cobalt removal process is verified by comparing the estimated values ​​of the unknown parameters with the corresponding output data in the actual production process.

[0125] The sample data are collected to obtain ion concentration data and data related to the ion concentration.

[0126] Four reactors are set up in the cobalt removal process. The reaction temperatures of the four reactors fluctuate within a small range. Therefore, the temperature is assumed to be constant during the modeling process and the heat balance equation is not considered.

[0127] When estimating unknown parameters, set the initial values ​​of relevant parameters: Q k =0.001*eye(8), R k =0.01*eye(4), α=0.01,κ=0,β=2;After multiple recursive calculations, the model parameters converged and the estimated unknown parameters were obtained.

[0128] The estimated unknown parameters are obtained through numerical calculation and compared with the corresponding output data in the actual production process to verify the effectiveness of the parameter estimation model.

[0129] Table 1 shows the statistical results of the mean relative error (MRE), maximum relative error (MAXRE), mean absolute error (MAE) and root mean square error (RMSE) of the four reactors. The results show that the parameters estimated by the augmented equation of state and UKF algorithm can be used in the established SCSTR model to predict the reactor outlet ion concentration in the cobalt removal process.

[0130] Table 1 Statistical results

[0131]

Claims

1. A method for estimating parameters of the hydrometallurgical zinc smelting and cobalt removal process based on UKF theory, characterized in that: The method comprises the following steps: 1) Collect sample data; Preprocess sample data; 2) Establish a dynamic model of the cobalt removal process, i.e., a set of nonlinear differential equations, as follows: Where: Q in , Q under are the liquid flow after copper removal and the underflow flow after recovery respectively; V is the reaction volume; is the concentration of impurity ions in the solution after copper removal; the superscript i (i = 1, 2, ..., 4) represents the reactor # i; the subscript j represents the type of impurity ions j = 1: cobalt ions; j = 2: nickel ions, such as represents the concentration of impurity ion j in reactor #i; is the ORP value, which represents the redox capacity of the solution and is controlled by the amount of zinc powder added; is the frequency factor; E e is the standard activation energy; is the factor affecting the cathode activation energy due to the change in electrode potential; F is the Faraday constant; e eq,j is the equilibrium potential; and is an approximation of the redox potential and the mixing potential; R is the ideal gas constant; T c is the reaction temperature; i is the mass-reaction surface coefficient; is the mass of zinc powder used to replace impurity ion j in reactor #i; 3) The parameter estimation algorithm constituting the dynamic model of the cobalt removal process is as follows: The cobalt ion concentration at each reactor outlet is used as the state variable to estimate the unknown parameters. The state variable is selected as The control variable is in represents the zinc powder addition rate of each reactor, and the output variable is the same as the state variable; the unknown parameter χ=[χ 1 ,x 2 ,x 3 ,x 4 ] as the state variable, and establish the augmented state equation and observation equation: y a (t)=H a x a (t)+v(t); In the formula, is a nonlinear function; w(t) and v(t) are both zero-mean white noise sequences, and their covariance matrices are Q and R respectively; therefore: 4) The iterative Euler method in the numerical integration method is used to perform a discretized numerical solution on each state of the dynamic model of the cobalt removal process to obtain a discretized state equation and a measurement equation. The steps are as follows: χ 1 (k+1)=χ 1 (k)+dχ 1 (k)·step_length χ 2 (k+1)=χ 2 (k)+dχ 2 (k)·step_length χ 3 (k+1)=χ 3 (k)+dχ 3 (k)·step_length χ 4 (k+1)=χ 4 (k)+dχ 4 (k)·step_length In the formula, step_length is the step length used, and its value is 0.

001. The discrete state equation and measurement equation obtained by numerical solution method are: x a (k+1)=f a (x a (k),u(k),w(k),k) y a (k)=H a x a (k)+v(k); 5) Based on the discretized state equation and measurement equation, a proportionally corrected symmetric sampling strategy is adopted to obtain the UKF recursive formula for estimating unknown parameters; 6) Use the UKF recursive formula to calculate the state filter value x a (k), thereby obtaining the estimated values ​​of the unknown parameters in the dynamic model of the cobalt removal process. The effectiveness of the dynamic model of the cobalt removal process is verified by comparing the estimated values ​​of the unknown parameters with the corresponding output data in the actual production process.

2. A method for estimating parameters of a zinc hydrometallurgy process for removing cobalt based on UKF theory as claimed in claim 1, characterized in that: The collected sample data mainly consists of inlet solution flow, underflow solution flow, reaction volume, waste acid flow, arsenic salt flow, reactor inlet ion concentration, zinc powder dosage, redox potential, reactor temperature, and reactor outlet ion concentration; The preprocessing of sample data comprises the following steps: 1) Use the 3σ criterion statistical discrimination method to process abnormal data; 2) Use cubic spline interpolation and polynomial fitting methods to fill in the missing values ​​of field data; The statistical discrimination method of the 3σ criterion has the following steps: The sample data is Y = {y1,y2,…,y n }, the mean is The deviation is Then the standard deviation is obtained according to the Bessel formula: If a sample data y i Deviation of |e i |>3σ, the data is considered unreasonable and should be removed from the sample data; the comparison is repeated in this way until all sample data are tested; The method of cubic spline interpolation and polynomial fitting, the steps are as follows: Take the mean of the two methods of cubic spline interpolation and polynomial fitting to fill; After processing abnormal data and field data with some missing values, an initial sample set of unknown parameter estimates is obtained.

3. A method for estimating parameters of a hydrometallurgical zinc smelting cobalt removal process based on UKF theory as claimed in claim 2, characterized in that: The proportionally corrected symmetric sampling strategy includes the following: 1) Sigma sampling points and weight calculation: In the formula, {φ i }(i=0,1,…,2n) is a set of Sigma sampling points obtained; and is the mean and variance; λ=α 2 (n+κ)-n, κ is the proportional factor, α is the scaling factor, which is used to control the distribution state of the sampling points; and are the corresponding mean weights and variance weights; β reflects the high-order characteristics of state history information; 2) Initialize state estimation and estimated variance: In the formula, and is the initial state estimate and estimated variance of the filter; 3) Time update: In the formula, From the sampling point through the nonlinear state function f a (·) delivered; and is the mean and variance of the one-step forecast; 4) Measurement update: In the formula, {ζ i }(i=0,1,…,2n) is the 2n+1 Sigma sampling points obtained; From these sampling points through H a The transmission is obtained; and One-step forecast means, variances, and cross-covariances for the measured variables: In the formula, is the measured value at time k+1; is the filter gain; and is the state estimate and estimated variance at time k+1.

4. A method for estimating parameters of a hydrometallurgical zinc smelting and cobalt removal process based on UKF theory as claimed in claim 3, characterized in that: When estimating unknown parameters, set the initial values ​​of relevant parameters: Q k =0.001*eye(8), R k =0.01*eye(4), α=0.01, κ=0, β=2; After multiple recursive calculations, the model parameters converged and the estimated unknown parameters were obtained.

5. A method for estimating parameters of a hydrometallurgical zinc smelting cobalt removal process based on UKF theory as claimed in claim 4, characterized in that: The estimated unknown parameters are obtained through numerical calculation and compared with the corresponding output data in the actual production process to verify the effectiveness of the parameter estimation model.

Images