A method and system for automatically controlling moisture content of sintering return powder

Through the multi-factor control method and neural network prediction model, automatic control of moisture in sintered powder in the smelter is realized, solving the problems of inaccurate moisture control and environmental pollution, and improving production efficiency and safety.

CN114265885BActive Publication Date: 2024-11-22CENT SOUTH UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111620431.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-28
Publication Date
2024-11-22
Estimated Expiration
2041-12-28

AI Technical Summary

Technical Problem

It is difficult to achieve accurate and stable moisture control of sintered refrigerant in smelters. Due to the unstable initial material size, temperature fluctuations, large hysteresis and manual operation, the moisture content of conveyor belt materials fluctuates greatly, workers have high labor intensity and serious environmental pollution.

Method used

The multi-factor control method and the moisture content prediction model based on the time series neural network are adopted, combined with the fuzzy control word table and the boxing method, the clean water flow is adjusted in real time to realize automatic control and stable management of the moisture of materials in the cooling cylinder.

Benefits of technology

It improves the accuracy and accuracy of moisture control, reduces workers' labor intensity, reduces dust pollution, improves sintering efficiency, and improves the operating environment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114265885B_ABST
    Figure CN114265885B_ABST
Patent Text Reader

Abstract

The present invention provides a fuzzy feedforward moisture automatic control method and system based on the assistance of a prediction model, the method comprising: acquiring production data of sintering return powder to establish a production database; using the feeding frequency and feed port temperature in the production database as input variables, using the clean water flow rate as an output variable, and establishing a fuzzy control word table between the input quantity and the output quantity; performing a fuzzy control word table query based on the currently collected feeding frequency and feed port temperature data to obtain the current fuzzy water addition amount; constructing a preferred historical data set based on the acquired production data, using the preferred historical data set to establish a moisture content prediction model based on a time series neural network, inputting the currently collected data into the prediction model to output a moisture content prediction value; dividing the deviation category between the moisture content prediction value and the preset moisture content value according to a boxing method; correcting the current fuzzy water addition amount online according to the deviation category to obtain the current target water addition amount.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of automatic monitoring and intelligent control of moisture content of ore sintering materials, and in particular proposes a method and system for automatic moisture control of sintering return powder materials. Background Art

[0002] The cooling process of the returned powder in the lead-zinc ore sintering process of the smelter is the latter process of the sintering process - recycling the sintering ingredients that do not meet the sintering process requirements. The process of recycling the returned powder requires buffering in the buffer bin, crushing in the crusher, cooling with water in the cooling cylinder, belt transmission and other processes to transport the returned powder back to the returned powder bin for re-batch. The present invention is applied to the process from cooling the returned powder with water in the cooling cylinder to belt transmission. Water is added to the feed end of the cooling cylinder, and the material is rotated and stirred by the cooling cylinder to achieve the purpose of cooling. At present, the smelter adopts the control method of artificial water addition, and there are two ways of adding water, mainly adding sludge (mineral processing waste liquid) and supplemented by adding clean water. The key points of process operation are: the purpose of the cooling cylinder is to cool and mix the materials from the powder return bin. The amount of feed is positively correlated with the frequency of the vibrating feeder. Different frequencies roughly correspond to different sludge valve openings. On-site workers adjust the sludge valve openings based on existing operating experience, and then visually inspect the moisture content of the material on the conveyor belt. If the moisture content is too high, reduce the amount of sludge added, and if the moisture content is too low, increase the amount of sludge added. However, the nature of the sludge is unstable, and the sludge pipe is prone to clogging, so it is necessary to add clean water to assist or directly use clean water instead of sludge. In addition, in order to avoid dust pollution and threats to the health of production workers as much as possible, a temporary emergency water spraying device is set on the conveyor belt. Once the workers notice that the moisture content is too low and dust will begin to rise, they will immediately open the backup water device and spray water directly on the conveyor belt to cover the dust. By analyzing the on-site process and the current operating status, the current water addition method mainly has the following difficulties or shortcomings:

[0003] 1) The size of the initial incoming material is unstable and the powder return bin is closed, making it difficult to grasp its mechanism. Even if the frequency of the vibrating feeder is fixed, the amount of material discharged cannot be accurately calculated, resulting in the inability to accurately control the amount of material discharged.

[0004] 2) Since the cooling cylinder needs to cool the incoming materials, the temperature of the incoming materials fluctuates greatly under different working conditions without any specific rules. The material feeding speed is fast, so the amount of cooling water is affected not only by the amount of incoming materials, but also by the temperature of the incoming materials. The current production method does not take this influencing factor into consideration.

[0005] 3) The nature of the sludge is unstable. Since the sludge is the waste liquid formed by mineral processing, it is a mixture of different mineral residues, and the proportion cannot be determined. Sometimes it will cause the sludge flow meter and valve to be blocked, making it impossible to achieve precise control.

[0006] 4) Workers visually observe the moisture content of materials on the conveyor belt based on their experience to determine the amount of sludge and clean water to be added and adjusted. Manual adjustment based on experience has the characteristics of slow response, long reaction cycle, poor adjustment effect and high labor intensity.

[0007] 5) It takes about 3 minutes for the returned powder to go from the inlet to the outlet of the cooling cylinder. It takes about 3 minutes to see the effect after adding sludge or clean water. This is a nonlinear, large lag, and slow time-varying process. It is very difficult to control the moisture value of the material at the outlet of the cooling cylinder to be stable within the required range.

[0008] 6) The manual valve control method of adding water is backward. When workers find that the moisture content of the conveyor belt material does not meet the requirements, the operator needs to go from the duty room to the site to manually adjust the valve. The operation is stopped when the moisture content of the conveyor belt material reaches the requirements. The backward water adding method increases the workload of the operators.

[0009] 7) Since the cooling cylinder is like a "black box" closed to the outside, the friction during material transmission is large, and it is difficult to install a detection device inside the cooling cylinder. Therefore, it is difficult to grasp the physical and chemical change mechanism inside. Only the operating parameters such as the inlet and outlet temperature of the material, the real-time moisture value of the material, etc. can be detected outside the cooling cylinder. The detection is not timely and accurate.

[0010] In summary, the cooling cylinder water adding system is affected by many factors, and it is difficult to achieve the ideal goal of stable and precise control of the moisture content of the conveyor belt material.

[0011] Therefore, the industry is in urgent need of a new method and system for controlling moisture in sintering return powder. Summary of the invention

[0012] The present invention provides a method and system for automatically controlling the moisture content of sintering return powder materials, which solves the problems of unstable moisture content and inaccurate control caused by manual water addition, and realizes target control, automatic control and stable control of the material cooling and mixing process in the cooling cylinder, so as to achieve the purpose of keeping the moisture content of the mixed conveyor belt materials within a stable range, avoids conveyor belt slippage caused by excessively high material moisture content and dust caused by excessively low material moisture content, improves sintering efficiency and reduces the labor intensity of workers.

[0013] A method for automatically controlling the moisture content of sintered powder return material comprises the following steps:

[0014] Obtaining production data of sintering return powder and establishing a production database; the production database includes the feeding frequency of the return powder, the temperature of the feed port, the bottom mud flow, the clean water flow, the spare water flow, the discharge port temperature and the real-time moisture content of the material on the conveyor belt;

[0015] A multi-factor control method is used to take the feeding frequency and feeding port temperature, which have the greatest impact on the moisture content of the material in the production database, as input variables, and the clean water flow rate as an output variable, to establish a fuzzy control word table between the input and output quantities;

[0016] According to the currently collected data of feeding frequency and feeding port temperature, the fuzzy control word table is queried to output the fuzzy water addition amount for this time;

[0017] According to the production data of the sintering return powder obtained, an optimal historical data set is constructed, and a moisture content prediction model is established using the optimal historical data set; the currently collected data is input into the prediction model, and the moisture content prediction value is output;

[0018] Divide the deviation categories of the moisture content prediction value and the preset moisture content value according to the binning method;

[0019] The fuzzy water addition amount output by the fuzzy control word table is corrected online according to the deviation category to obtain the target water addition amount.

[0020] Furthermore, a multi-factor control method is used to take the feeding frequency and feeding port temperature, which have the greatest impact on the moisture content of the material in the production database, as input variables, and the clean water flow rate as the output variable, to establish a fuzzy control word table between the input and output, specifically:

[0021] Assuming the sampled feed frequency is f and the feed inlet temperature is t, calculate the clean water output u:

[0022]

[0023]

[0024] in, and “·” are both fuzzy matrix operators, R represents the fuzzy relationship, u1 is the control output obtained by the first fuzzy relationship, NB u is the fuzzy set NB f , N.M. f or fuzzy set NB t , N.M. t When the mapping fuzzy set is established at the same time, if the membership function values ​​of the feed frequency f and the feed inlet temperature t correspond to the quantified level and take 1, and the rest are taken as 0, formula (2) is simplified to:

[0025]

[0026] In the formula, They are fuzzy sets NB f , N.M. f The membership degree of the i-th element of , i.e., the measured feeding frequency is at the i-th level; They are fuzzy sets NB t , N.M. t The membership degree of the jth element of , that is, the measured temperature is the jth level; similarly, the control quantities u2, u3, ... u n , then the control quantity is a fuzzy set u expressed as:

[0027] u=u1+u2+u3+……+u n (4)

[0028] The maximum membership method is used to convert the control quantity from fuzzy quantity to precise quantity, and the fuzzy control word table is obtained.

[0029] Furthermore, based on the production data of the sintering return powder obtained, an optimal historical data set is constructed, and a moisture content prediction model is established using the optimal historical data set; the currently collected data is input into the prediction model, and the moisture content prediction value is output, specifically:

[0030] The preferred historical data set includes: feeding frequency, bottom mud flow, clean water flow, feeding port temperature and material moisture content at the discharge port;

[0031] The moisture content prediction model is a moisture content prediction model based on a time series neural network; the currently collected data include sediment flow, clean water flow, feeding frequency, and feed port temperature as input variables of the prediction model, and the moisture content of the material at the discharge port is the output variable of the prediction model.

[0032] Furthermore, the moisture content prediction model based on time series neural network is constructed as follows: the preferred historical data set is divided into a training set, a validation set and a test set in a ratio of 6:2:2, the training set data of the preferred historical data set is fully utilized through appropriate offline training, and the model is obtained by training using the Levenberg Marquardt algorithm.

[0033] Furthermore, the deviation categories between the predicted moisture content value and the preset moisture content value are divided according to the binning method, specifically:

[0034] The historical data set M of the deviation between the predicted value of the moisture content at the outlet and the preset moisture content value during the mixing process is mapped to the moisture content deviation classification set N in the form of function mapping by the binning method, and the deviation category function y=f(x) corresponding to the moisture content deviation x of the mixed material is obtained, where x∈M;

[0035] Since the measurement range of the moisture meter is between 0% and 10%, assuming that when the difference between the predicted moisture content and the set value is less than -0.8%, -0.8% to -0.2%, -0.2% to 0.2%, 0.2% to 1%, and greater than 1%, the corresponding moisture content deviation categories are 1, 2, 3, 4, and 5 respectively.

[0036] Furthermore, the fuzzy water addition amount output by the fuzzy control word table is corrected online according to the deviation category to obtain the target water addition amount, including:

[0037] If the deviation category y is 1, the target water addition amount P = a × (1 + 15%);

[0038] If the deviation category y is 2, the target water addition amount P = a × (1 + 7%);

[0039] If the deviation category y is 3, the target water addition amount P = a;

[0040] If the deviation category y is 4, the target water addition amount P = a × (1-7%);

[0041] If the deviation category y is 5, the target water addition amount P = a × (1-15%);

[0042] Among them, P is the target water addition amount this time, and a is the fuzzy water addition amount this time output by the fuzzy control word table.

[0043] The present invention also provides an automatic moisture control system for sintering returned powder materials, comprising a processor and a memory connected to the processor, wherein the memory stores an automatic moisture control program for sintering returned powder materials, and when the automatic moisture control program for sintering returned powder materials is executed by the processor, the steps of the automatic moisture control method for sintering returned powder materials as described above are implemented.

[0044] Furthermore, the automatic moisture control system of the sintered return powder material also includes: a cooling cylinder, a clean water storage device, a bottom mud storage device, a vibrating feeder, a material conveyor belt, and an infrared thermometer for measuring the temperature of the cooling cylinder feed port and the discharge port;

[0045] The clean water storage device, the bottom mud storage device and the vibrating feeder are connected to the feeding end of the cooling cylinder;

[0046] The material conveyor belt is connected to the discharge end of the cooling cylinder;

[0047] A clean water electric valve and a clean water flow meter are provided in sequence on the pipeline between the clean water storage device and the cooling cylinder;

[0048] A manual valve and a sediment flow meter are provided in sequence on the pipeline between the sediment storage device and the cooling cylinder;

[0049] A moisture meter is also provided on the material conveyor belt.

[0050] Furthermore, the material conveyor belt is provided with a backup water device, and the backup water device is equipped with a backup water spraying device to avoid the occurrence of dust when the moisture content is low.

[0051] The present invention has the following beneficial effects:

[0052] 1. The present invention provides an automatic moisture control method for sintering return powder. Based on previous operating experience, a production database is established to solve the problem of no initial data in the control system, which has reference significance for the domestic sintering process water adding system of the same type. In addition, the present invention uses a neural network to build a moisture content prediction model, and then performs online correction on the fuzzy output according to the deviation between the predicted moisture content and the set value. While limiting the control output to a certain range in advance, it also greatly improves the control accuracy and precision, and greatly improves the control delay phenomenon caused by large time lag.

[0053] 2. The present invention provides a method for automatically controlling the moisture content of sintering return powder, which adopts the fuzzy feedforward-closed loop control principle and constructs and trains a moisture content prediction model based on a time series neural network based on historical data sets. When designing a fuzzy controller, the cooling cylinder feed temperature and the conveyor vibration frequency are used as inputs, and the clean water flow rate is used as the system output. The cooling cylinder feed temperature, conveyor vibration frequency, and clean water flow rate domains, fuzzy language variable word sets, and membership functions are determined. A multi-factor control method is used, with two variables as inputs, rather than taking the error and error change rate of a single variable as input.

[0054] 3. The present invention provides an automatic moisture control system for sintering return powder, which adopts the method of correcting the output value of the fuzzy controller by the output of the prediction model, greatly improving the large lag, nonlinearity, slow time variation, and strong coupling problems that cannot be solved independently by a single fuzzy control. After the moisture prediction value is introduced, the deviation from the set moisture value is obtained, and the amount of water added this time can be predicted in advance. The method of moisture deviation classification is adopted, so that the system can correct the amount of water added more accurately without frequent actions. The role of the prediction model is limited by the fuzzy control-dominated method, avoiding the system instability caused by the sudden change of the prediction value caused by the sudden change of the industrial environment. At the same time, it solves the problems of low moisture content of conveyor belt materials and dust, high temperature of materials at the discharge port and damage to equipment, and high moisture content of conveyor belt materials and slipping materials caused by inaccurate judgment, untimely adjustment, and unstable moisture caused by manual experience operation. It not only greatly reduces the labor intensity of workers and improves labor efficiency, but also improves the operating environment of operators and reduces the impact of dust on the surrounding environment.

[0055] In addition to the above-described purposes, features and advantages, the present invention has other purposes, features and advantages. The present invention will be further described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The drawings constituting a part of this application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0057] Figure 1 This is a flow chart of a method for automatically controlling the moisture content of sintering return powder according to the present invention;

[0058] Figure 2 It is a schematic diagram of the corrected fuzzy output of the prediction model of the present invention;

[0059] Figure 3 It is the training model response of the moisture content prediction model based on the time series neural network adopted by the present invention;

[0060] Figure 4 is a regression curve of a training model of a moisture content prediction model based on a time series neural network adopted by the present invention;

[0061] Figure 5 is an additional test response of the model of the moisture content prediction model based on the time series neural network adopted by the present invention;

[0062] Figure 6 The present invention is a fuzzy feedforward moisture content automatic control system diagram based on the assistance of a prediction model. DETAILED DESCRIPTION

[0063] The embodiments of the present invention are described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0064] The specific model of adding water to sintering return powder is as follows: after the sintering and roasting process is completed, it is crushed by a single shaft (one crusher) and a toothed roller (two crushers) and then screened. The screened material is a qualified sintered block and is sent to a blast furnace for smelting. The screened material passes through a corrugated roller crusher (three crushers) and a smooth roller crusher (four crushers) to make qualified return powder. In this sintering return powder water addition automatic control system, the return powder sent by the conveyor is cooled and mixed. The return powder includes lead ore, zinc ore, mixed ore, limestone, dust collection ash, blue powder, secondary zinc oxide, electric dust, etc. The NIR1 moisture meter is used to detect the moisture content of the mixed material online in real time at the outlet of the cooling cylinder. The moisture content relationship model is:

[0065]

[0066] In the above formula, M(t) is the moisture content of the material at the outlet of the cooling cylinder at time t, S(t) is the moisture content of the sludge, W(t) is the total mass of the sludge, D(t) is the moisture content of the sintering return powder raw material, F(t) is the mass of the sintering return powder raw material, and U(t) is the amount of water added.

[0067] It can be seen that the moisture content of the material at the discharge port is affected by many factors.

[0068] Based on this, Figure 1 and Figure 2 As shown, the present invention provides a method for automatically controlling the moisture content of sintering return powder, comprising the following steps:

[0069] Step 101, obtaining production data of sintering return powder and establishing a production database; the production database includes the feeding frequency of the return powder, the feed port temperature, the bottom mud flow, the clean water flow, the spare water flow, the discharge port temperature and the real-time moisture content of the material on the conveyor belt.

[0070] Step 102, adopting a multi-factor control method, taking the feeding frequency and the feeding port temperature which have the greatest impact on the moisture content of the material in the production database as input variables, taking the clean water flow rate as the output variable, and establishing a fuzzy control word table between the input and output quantities.

[0071] Step 103: perform a fuzzy control word table query based on the currently collected feed frequency and feed port temperature data, and output the fuzzy water addition amount for this time.

[0072] Step 104: construct an optimal historical data set based on the production data of the sintering return powder obtained, and use the optimal historical data set to establish a moisture content prediction model; input the currently collected data into the prediction model, and output the moisture content prediction value. The currently collected data includes feed inlet temperature, feed frequency, bottom mud flow rate, and clean water flow rate.

[0073] Step 105: Divide the deviation categories between the predicted moisture content value and the preset moisture content value according to the binning method.

[0074] Step 106: Correct the fuzzy water addition amount output by the fuzzy control word table online according to the deviation category to obtain the target water addition amount.

[0075] In step 101, the production database is specifically: we compare the actual moisture measured by the material outlet moisture meter with the set value, save the operating parameters with a deviation of less than 5%, and obtain an empirical operation database that can keep the moisture of the material at the outlet stable and accurate under various conditions. This includes the feed inlet temperature, feed frequency, sludge flow, clean water flow (water addition), the amount of standby water added, the moisture meter reading and the moisture setting value at that time. The feed frequency is determined by the vibration frequency of the vibrating feeder. Based on previous operating experience, the production database is established to solve the problem of no initial data in the control system, which is of reference significance to the domestic sintering process water addition system of the same type.

[0076] In step 102, unlike previous studies, this study does not directly use the error between the actual output value and the set value and the error change rate as fuzzy variables, but uses a multi-factor control method to take the feed frequency f and temperature t, which have the greatest impact on the moisture content of the material in the production process, as input variables, and the clean water flow u as the output variable, and selects the triangle membership function, and the domain quantification formula is:

[0077]

[0078] Among them, m and n represent the two endpoints of the discretized domain, and a and b represent the minimum and maximum values ​​of the actual physical quantity. According to the change range of the feed frequency (30HZ~42HZ), thirteen levels of domains are obtained after quantization: [-6, -5, -4, -3, -2, -1, 0, +1, +2, +3, +4, +5, +6]. The fuzzy set is {NB, NM, NS, O, PS, PM, PB}, and the frequency change fuzzy assignment is shown in Table 1:

[0079] Table 1 Fuzzy assignment table of feeding frequency f

[0080]

[0081] According to the change range of feed temperature (60℃~200℃), quantization results in fourteen levels of domain: [-6, -5, -4, -3, -2, -1, 0-, 0+, +1, +2, +3, +4, +5, +6]. The fuzzy set is {NB, NM, NS, NO, PO, PS, PM, PB}, and the fuzzy assignment of feed temperature change is shown in Table 2:

[0082] Table 2 Fuzzy assignment table of feed temperature t

[0083]

[0084] According to the change range of clean water output (0~9m 3 / h), after quantization, fifteen levels of domains are obtained: [-7, -6, -5, -4, -3, -2, -1, 0, +1, +2, +3, +4, +5, +6, +7]. The fuzzy set is {NB, NM, NS, O, PS, PM, PB}, and the fuzzy assignment of clean water flow change is shown in Table 3:

[0085] Table 3 Fuzzy value table of clean water flow u

[0086]

[0087] Based on expert experience, the fuzzy rules are described in the form of "IF A THEN B" (only four examples are listed):

[0088] (1)if t=NB or NM and f=NB or NM then u=NB

[0089] (2)if t=NB or NM and f=NS or O then u=NB

[0090] (3)if t=NB or NM and f=PS then u=NM

[0091] (4)if t=NB or NM and f=PB or PM then u=O

[0092] Assuming the sampled feed frequency is f and the feed inlet temperature is t, calculate the clean water output u:

[0093]

[0094]

[0095] in, and “·” are both fuzzy matrix operators, R represents the fuzzy relationship, u1 is the control output obtained by the first fuzzy relationship, NB u is the fuzzy set NB f , N.M. f or fuzzy set NB t , N.M. t When the mapping fuzzy set is established at the same time, if the membership function values ​​of the feed frequency f and the feed inlet temperature t correspond to the quantified level and take 1, and the rest are taken as 0, formula (2) is simplified to:

[0096]

[0097] In the formula, They are fuzzy sets NB f , N.M. fThe membership degree of the ith element of (i.e., the measured feeding frequency is the ith level), They are fuzzy sets NB t , N.M. t The membership degree of the jth element of (i.e. the measured temperature is the jth level) is obtained in the same way. The control quantities u2, u3, ... u n , then the control quantity is a fuzzy set u expressed as:

[0098] u=u1+u2+u3+……+u n (4)

[0099] The maximum membership method is used to convert the control quantity from fuzzy quantity to precise quantity, that is, the element with the largest membership in the fuzzy subset is selected as the control quantity. If there are several elements with the largest membership, the control quantity takes their average value. The fuzzy control word table can be obtained as shown in Table 4:

[0100] Table 4 Output control word table

[0101]

[0102] In step 103, through the control word table generated in the previous step, when we perform real-time control again, we can directly query the control strategy that should be adopted under the current state according to the currently collected feeding frequency and feeding temperature data, obtain the corresponding control word u, and then multiply it by the proportional factor k to obtain the fuzzy water addition amount a for this time.

[0103] In step 104, a moisture content prediction model is first constructed, specifically:

[0104] The historical data of the process of adding water to the return powder of lead-zinc sintering under industrial environment were collected. After data preprocessing, the parameters such as conveyor vibration frequency, sediment flow, clean water flow, cooling cylinder feed inlet temperature and discharge port material moisture content were selected to construct the optimal historical data set. 40,000 sets of conveyor vibration frequency, sediment flow, clean water flow, cooling cylinder feed inlet temperature and discharge port material moisture content data in the historical data set were selected and divided into training set, validation set and test set according to the ratio of 6:2:2. The data of the optimal historical data set were fully utilized through appropriate offline training. The model training was carried out using the Levenbergmarquardt algorithm to obtain the basic time series neural network model. Specifically, for the water addition control process, the input of the offline time series neural network consists of the sediment flow weight vector x1(n), the feed frequency weight vector x2(n), the feed port temperature weight vector x3(n), the clean water flow weight a(n) and the discharge port moisture content y(n) and their delays, which can be expressed as:

[0105] Input(n)=[x(ndx ),...x(n),a(nd x ),...a(n),y(nd x ),...y(nd x )] (5)

[0106] The output Output(n) of the time series neural network model is defined as:

[0107]

[0108] The Levenberg Marquardt (LM) algorithm combines the advantages of the Newton method and the Gauss-Newton method, and can achieve effective supervised learning in the training process of the neural network model. Its weight update formula is:

[0109] W h+1 =W h -J h r h (H n +μI) -1 (7)

[0110] Where: W is the weight matrix; h is the number of iterations; J h is the Jacobian matrix; r h is the residual error vector; H is the Hessian matrix; μ is the penalty factor; I is the identity matrix.

[0111]

[0112] s lm is the iteration step size, when μ>0, the covariance matrix is always a positive number, which ensures that s lm Can represent the descending direction of f(x):

[0113]

[0114] The definition of the descending direction of the function comes from the expansion of the Taylor function. If the following formula s lm <0, it means that each small step will cause the function value f(x) to decrease:

[0115] f(x)≈f(x0)+f(x0)(x-x0)=f(x0)s lm (10)

[0116] In order to obtain the time series neural network model, we use the sediment flow, clean water flow, feeding frequency, and feeding port temperature as input variables, and the moisture content at the discharge port as the output variable. We set a single hidden layer, the number of hidden layer neurons is 10, the delay is 2 cycles, and the activation function is the sigmoid function. The data was divided into training set, validation set and test set in a ratio of 6:2:2. Some data from the training set were selected to obtain the output response of the time series neural network. The experimental results are shown in the figure. Figure 3 As shown in the figure, it can be clearly seen that the model output follows the target value well, and most of the moisture errors are far less than ±0.2.

[0117] Figure 4 The scatter plot shows the correlation between the output value of the time series neural network model and the target value in the training set, validation set, test set and overall data. Figure 4 We can see that the regression rate R of the training set is 0.9972, the regression rate R of the validation set is 0.99883, and the regression rate R of the test set is 0.9951. The overall regression rate R is 0.99713. Therefore, we believe that the model training effect is very good and all the information between the data has been fully learned.

[0118] In order to verify whether the model training has achieved the desired effect, we added 2000 sets of data to conduct additional tests on the trained model. The response curve is as follows: Figure 5 As shown, it can be seen that the model output value follows the target curve well, and the error is limited to ±0.1.

[0119] In step 105, according to manual experience, the historical data set M of the deviation between the predicted value and the set value of the moisture content at the outlet during the mixing process is mapped to the moisture content deviation classification set N in the form of function mapping through the binning method, and the deviation category function y=f(x) corresponding to the moisture content of the mixed moisture content deviation x(x∈M) is obtained. Since the measurement range of the moisture meter is between 0% and 10%, it is assumed that when the difference between the predicted value and the set value of the moisture content is less than -0.8%, -0.8% to -0.2%, -0.2% to 0.2%, 0.2% to 1%, and greater than 1%, the moisture content deviation categories are 1, 2, 3, 4, and 5 respectively. The classification results are shown in Table 5:

[0120] Table 5 Moisture content deviation classification set

[0121]

[0122] In step 106, the fuzzy water addition amount output by the fuzzy control word table is corrected online according to the deviation classification table in the previous step to obtain the target water addition amount for this time. The data of each raw material in the next lead-zinc ore sintering return powder mixing and water addition process are collected, and the current conveyor vibration frequency, bottom mud flow, clean water flow, and cooling cylinder feed port temperature are input into the trained moisture content prediction model to obtain the predicted value of the moisture content of the material at the discharge port after this water addition, and then the deviation classification of the moisture content prediction value and the preset moisture content value is obtained, and then the current fuzzy water addition amount a is adjusted according to the deviation category and manual operation experience:

[0123] If the deviation category y is 1, the target water addition amount P = a × (1 + 15%);

[0124] If the deviation category y is 2, the target water addition amount P = a × (1 + 7%);

[0125] If the deviation category y is 3, the target water addition amount P = a;

[0126] If the deviation category y is 4, the target water addition amount P = a × (1-7%);

[0127] If the deviation category y is 5, the target water addition amount P = a × (1-15%);

[0128] Among them, P is the target water addition amount this time, and a is the fuzzy water addition amount this time output by the fuzzy control word table.

[0129] The present invention also provides an automatic moisture control system for sintering return powder materials, comprising a processor and a memory connected to the processor, wherein the memory stores an automatic moisture control program for sintering return powder materials, and when the automatic moisture control program for sintering return powder materials is executed by the processor, the steps of the above-mentioned automatic moisture control method for sintering return powder materials are implemented.

[0130] like Figure 6 As shown, the automatic moisture control system for sintering return powder material also includes: a cooling cylinder, a clean water storage device, a bottom mud storage device, a vibrating feeder, a material conveyor belt, and an infrared thermometer for measuring the temperature of the cooling cylinder inlet and outlet;

[0131] The clean water storage device, the bottom mud storage device and the vibrating feeder are connected to the feeding end of the cooling cylinder;

[0132] The material conveyor belt is connected to the discharge end of the cooling cylinder;

[0133] A clean water electric valve and a clean water flow meter are provided in sequence on the pipeline between the clean water storage device and the cooling cylinder;

[0134] A manual valve and a sediment flow meter are provided in sequence on the pipeline between the sediment storage device and the cooling cylinder;

[0135] A moisture meter is also provided on the material conveyor belt.

[0136] The material conveyor belt is also provided with a reserve water device, and the reserve water device is equipped with a reserve water spraying device to avoid the occurrence of dust when the moisture content is low.

[0137] In view of the unstable nature of the sediment and the fact that it sometimes clogs the valve, the control system of the present invention retains the original manual adjustment method of the sediment, replaces the clean water electric regulating valve, installs a sediment flowmeter and a clean water flowmeter to measure the real-time sediment flow and clean water flow, and the insufficient sediment volume is supplemented by clean water.

[0138] The control system of the present invention uses a more advanced NIR moisture meter, which can directly display the moisture content in the form of a number, making it convenient for workers to determine the moisture range that meets the requirements. The design of the backup water addition function is mainly to take into account the safe operation of the system. The backup water is designed for the low or too low moisture content of the material at the discharge port. Low moisture will not only affect the efficiency of the next process of sintering and granulation, but may also cause dust on the conveyor belt, causing relatively large pollution to the environment. In addition, if the cooling drum is not added with water or the amount of water added is too little due to various reasons, resulting in too low moisture content of the material at the discharge port, it may even directly burn the conveyor belt and bring extremely adverse interference to the subsequent process. If an accident occurs, the loss to production will be immeasurable.

[0139] A water spraying device is installed on the material conveyor belt to avoid dust when the moisture content is low. At the same time, signal cables are laid for data transmission to transmit real-time moisture monitoring data to the PLC system and the sintering plant server, providing real-time data parameters for the control system and enabling production operators to understand the real-time changing moisture data. In addition, the control program is compiled according to the moisture control model, and the human-computer interaction interface---automatic water addition control screen is installed on the industrial control computer in the sintering duty room. The upper control interface uses WINCC configuration software, which has a user-friendly interface and powerful functions. Design and manufacture the return powder cooling and water addition control cabinet, and configure and install the SMARTLINE touch screen on the control cabinet panel to display the operating status of the cooling cylinder in real time, and control the amount of water added to the return powder cooling in real time, so as to improve the informatization and intelligence of the production process.

[0140] The working mode of the control system of the present invention is as follows: the expected moisture content value of the conveyor belt material is set on the touch screen or the WINCC operation interface, the discharge amount is roughly estimated according to the collected vibration feeder frequency, and according to the discharge amount and the feed inlet temperature, the operator adjusts the sediment flow to a more suitable position by adjusting the sediment manual valve (the specific sediment flow is related to the sediment properties on site, and it is necessary to ensure a certain amount of sediment usage, and not to block the flow meter and pipeline due to excessive sediment concentration. We generally believe that it is more appropriate to adjust to about 50% of the total water addition amount, and the remaining deficiency is supplemented by clean water, that is, the actual control output), and then the collected sediment flow, feed temperature, vibration feeder frequency and clean water flow are feedforward compensated by using the fuzzy control principle, and the current fuzzy water addition amount is estimated in combination with expert experience, and a moisture content prediction model is built by using a neural network, and then the fuzzy output is corrected online according to the deviation between the predicted moisture content and the set value to obtain the target water addition amount this time. The fuzzy controller transmits the control signal of the corrected target water addition amount to the clean water electric valve. The clean water electric valve immediately acts after receiving the adjustment signal to complete the first step of moisture adjustment. Then, according to the deviation between the real-time measurement result of the conveyor belt material moisture meter and the set moisture content set value, the PID control theory is applied to quantify this deviation to the amount of clean water added to complete the second step of moisture adjustment. At this point, the moisture content of the conveyor belt material is basically achieved. In order to avoid unexpected situations and changes in moisture content due to force majeure, a backup water device is installed on the conveyor belt for emergency use. When it is detected that the moisture content of the conveyor belt material is low and the temperature of the discharge port is high, the backup water electric valve receives the adjustment signal and immediately acts in a graded manner according to the basic principle of fuzzy control, so as to avoid dust caused by too low moisture content of the conveyor belt material and equipment damage caused by too high material temperature in extreme environments. At this point, the phased functions of the entire lead-zinc ore sintering return powder moisture content automatic control system are fully realized.

[0141] This system adopts the method of correcting the output value of the fuzzy controller by the output of the prediction model, which greatly improves the large lag, nonlinearity, slow time variation, and strong coupling problems that cannot be solved independently by a single fuzzy control. After the introduction of the moisture prediction value, the deviation from the set moisture value is obtained, and the amount of water added this time can be predicted in advance. The method of moisture deviation classification is adopted, so that the system can more accurately correct the amount of water added without frequent actions. The role of the prediction model is limited by the fuzzy control-dominated method, avoiding the system instability caused by the sudden change of the prediction value caused by the sudden change of the industrial environment. At the same time, it solves the problems of low moisture content of conveyor belt materials and dust, high temperature of materials at the discharge port and damage to equipment, and high moisture content of conveyor belt materials and slipping materials caused by inaccurate judgment, untimely adjustment, and unstable moisture caused by manual experience operation. It not only greatly reduces the labor intensity of workers and improves labor efficiency, but also improves the operating environment of operators, reduces the impact of dust on the surrounding environment, improves the level of factory environmental management, adapts to the requirements of my country's environmental protection laws and regulations, improves the efficiency and quality of sintering production, reduces labor costs, and achieves the coordination and unity of ecological and economic benefits.

[0142] This system can realize automatic dynamic control of the amount of water added to the sintering return powder under different working conditions so that the moisture content of the return powder meets the sintering production requirements and displays the real-time moisture content and water addition data of the production process, as well as the operation mode. Its advantage lies in the informatization and intelligence of the production process. This system is suitable for the water addition control of the raw material mixing process in various sintering production processes. It can be designed as an independent control system or implemented using a PLC control system based on automatic control of the production process. It is less restricted by on-site conditions and has low cost. It solves the environmental pollution problem caused by the production process, improves the production efficiency of the factory, reduces the labor intensity of the staff, and achieves the coordination and unity of ecological and economic benefits.

[0143] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for automatically controlling the moisture content of sintered powder, characterized in that: The following steps are involved: Obtaining production data of sintering return powder and establishing a production database; the production database includes the feeding frequency of the return powder, the temperature of the feed port, the bottom mud flow, the clean water flow, the spare water flow, the discharge port temperature and the real-time moisture content of the material on the conveyor belt; A multi-factor control method is used to take the feeding frequency and feeding port temperature, which have the greatest impact on the moisture content of the material in the production database, as input variables, and the clean water flow rate as an output variable, to establish a fuzzy control word table between the input and output quantities; According to the currently collected data of feeding frequency and feeding port temperature, the fuzzy control word table is queried to output the fuzzy water addition amount for this time; According to the production data of the sintering return powder obtained, an optimal historical data set is constructed, and a moisture content prediction model is established using the optimal historical data set; the currently collected data is input into the prediction model, and the moisture content prediction value is output; Divide the deviation categories of the moisture content prediction value and the preset moisture content value according to the binning method; The fuzzy water addition amount output by the fuzzy control word table is corrected online according to the deviation category to obtain the target water addition amount.

2. The method for automatically controlling the moisture content of sintered powder according to claim 1, characterized in that: The multi-factor control method is used to take the feeding frequency and feeding port temperature, which have the greatest impact on the moisture content of the material in the production database, as input variables, and the clean water flow rate as the output variable, and establish a fuzzy control word table between the input and output quantities, specifically: Assuming the sampled feed frequency is f and the feed inlet temperature is t, calculate the clean water output u: in, and "·" are both fuzzy matrix operators, R represents the fuzzy relationship, u1 is the control output obtained by the first fuzzy relationship, NB u is the fuzzy set NB f , N.M. f or fuzzy set NB t , N.M. t When the mapping fuzzy set is established at the same time, if the membership function values ​​of the feed frequency f and the feed inlet temperature t correspond to the quantified level and take 1, and the rest are taken as 0, formula (2) is simplified to: In the formula, They are fuzzy sets NB f , N.M. f The membership degree of the ith element of , that is, the measured feeding frequency is the ith level; They are fuzzy sets NB t , N.M. t The membership degree of the jth element of , that is, the measured temperature is at the jth level; similarly, the control quantities u2, u3, ... u n , then the control quantity is a fuzzy set u expressed as: u=u1+u2+u3+……+u n (4) The maximum membership method is used to convert the control quantity from fuzzy quantity to precise quantity, and the fuzzy control word table is obtained.

3. The method for automatically controlling the moisture content of sintered powder according to claim 1, characterized in that: According to the obtained production data of sintering return powder, an optimal historical data set is constructed, and a moisture content prediction model is established using the optimal historical data set; Input the currently collected data into the prediction model and output the moisture content prediction value. Specifically: The preferred historical data set includes: feeding frequency, bottom mud flow, clean water flow, feeding port temperature and material moisture content at the discharge port; The moisture content prediction model is a moisture content prediction model based on a time series neural network; the currently collected data include sediment flow, clean water flow, feeding frequency, and feed port temperature as input variables of the prediction model, and the moisture content of the material at the discharge port is the output variable of the prediction model.

4. The method for automatically controlling the moisture content of sintered powder according to claim 3, characterized in that: The moisture content prediction model based on time series neural network is constructed specifically as follows: the preferred historical data set is divided into a training set, a validation set and a test set in a ratio of 6:2:2, and the training set data of the preferred historical data set is used for offline training, and the model is obtained by training using the Levenberg Marquardt algorithm.

5. The method for automatically controlling the moisture content of sintered powder according to claim 1, characterized in that: The deviation categories between the predicted moisture content value and the preset moisture content value are divided according to the binning method, specifically: The historical data set M of the deviation between the predicted value of the moisture content of the material at the outlet and the preset moisture content value during the mixing process is mapped to the moisture content deviation classification set N in the form of function mapping by the binning method, and the deviation category function y=f(x) corresponding to the moisture content deviation x of the mixed material is obtained, where x∈M; Since the measurement range of the moisture meter is between 0% and 10%, assuming that when the difference between the predicted moisture content and the set value is less than -0.8%, -0.8% to -0.2%, -0.2% to 0.2%, 0.2% to 1%, and greater than 1%, the corresponding moisture content deviation categories are 1, 2, 3, 4, and 5 respectively.

6. The method for automatically controlling the moisture content of sintered powder according to claim 5, characterized in that: The fuzzy water addition amount output by the fuzzy control word table is corrected online according to the deviation category to obtain the target water addition amount, including: If the deviation category y is 1, the target water addition amount P = a × (1 + 15%); If the deviation category y is 2, the target water addition amount P = a × (1 + 7%); If the deviation category y is 3, the target water addition amount P = a; If the deviation category y is 4, the target water addition amount P = a × (1-7%); If the deviation category y is 5, the target water addition amount P = a × (1-15%); Among them, P is the target water addition amount this time, and a is the fuzzy water addition amount this time output by the fuzzy control word table.

7. An automatic moisture control system for sintering return powder, characterized in that: It includes a processor and a memory connected to the processor, the memory stores an automatic moisture control program for sintering returned powder, and when the automatic moisture control program for sintering returned powder is executed by the processor, the steps of the automatic moisture control method for sintering returned powder as described in any one of claims 1-6 are implemented.

8. The automatic moisture control system for sintering return powder according to claim 7, characterized in that: Also includes: Cooling cylinder, clean water storage device, bottom mud storage device, vibrating feeder, material conveyor belt and infrared thermometer for measuring the temperature of the cooling cylinder inlet and outlet; The clean water storage device, the bottom mud storage device and the vibrating feeder are connected to the feeding end of the cooling cylinder; The material conveyor belt is connected to the discharge end of the cooling cylinder; A clean water electric valve and a clean water flow meter are provided in sequence on the pipeline between the clean water storage device and the cooling cylinder; A manual valve and a sediment flow meter are provided in sequence on the pipeline between the sediment storage device and the cooling cylinder; A moisture meter is also provided on the material conveyor belt.

9. The automatic moisture control system for sintering return powder according to claim 7, characterized in that: The material conveyor belt is also provided with a reserve water device, and the reserve water device is equipped with a reserve water spraying device to avoid the occurrence of dust when the moisture content is low.