A method for optimally setting the foam velocity by integrating multi-source flotation condition information

Abstract

The present invention relates to a method for optimally setting the foam velocity by integrating multi-source flotation operating conditions information, and relates to the technical field of intelligent optimization production of flotation. Flotation operating condition data is collected through a flotation production measurement and control platform, and several flotation operating condition data source case sets are obtained through data preprocessing; based on the flotation operating condition matching distance, combined with the K-Means clustering method, K groups of subclass case sets are obtained, and key production indexes are obtained by combining a dressing grade prediction model and a recovery rate calculation model. The optimal case under each group of sub-case sets is obtained through a foam velocity setting model and an evaluation and learning model; the new case of the flotation operating condition is retrieved in the case library. If there is no matching case, the new case is added to the flotation operating condition source case set, the case is corrected, and the case retrieval is continued. If a similar case is matched, the optimal matching case is obtained; and it is exported from the case library for case reuse to obtain the optimal setting value of the foam velocity; the problem of fluctuations in the flotation operating condition process is solved, and the key production indexes of flotation are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent optimization of flotation production, and particularly to a method for optimally setting the foam velocity by integrating multi-source flotation condition information. Background Art

[0002] For a long time, flotation production mainly relies on manual observation of the visual characteristics of the foam surface for adjustment operations, which has subjectivity and randomness, easily causes fluctuations in flotation conditions, and affects the quality of concentrate products and mineral recovery rates. Among them, the setting of the foam velocity value is of great significance for the control of the flotation process. At present, some flotation foam expert systems put into use generally manually set the foam velocity target value. However, the manual setting experience varies, with subjectivity and randomness, easily causing fluctuations in flotation conditions. Relying solely on manual operation cannot keep the flotation production index in the optimal production state for a long time. Therefore, according to the real-time flotation condition characteristic information data, optimizing the setting of the foam velocity to achieve the optimized production of the flotation process, ensuring the stable operation of the ore dressing production while meeting the requirements of the concentrate grade production plan index, increasing the concentrate output, improving the metal recovery rate, reducing energy consumption and production costs, reducing the fluctuations in flotation conditions, and effectively guiding the transformation and upgrading of the ore dressing industry are of great significance.

[0003] Therefore, it is necessary to design a method for optimally setting the foam velocity to solve the above technical problems. Summary of the Invention

[0004] For this purpose, the present invention provides a method for optimally setting the foam velocity by integrating multi-source flotation condition information to overcome the problems of reducing the fluctuations in flotation conditions and effectively guiding the transformation and upgrading of the ore dressing industry in the prior art.

[0005] To achieve the above object, the present invention provides a method for optimally setting the foam velocity by integrating multi-source flotation condition information, including:

[0006] Collecting flotation condition data through a flotation production measurement and control platform, and performing data preprocessing to obtain a number of flotation condition data source case sets;

[0007] Based on the flotation condition matching distance, combining the K-Means clustering method to obtain K groups of subclass case sets, combining the ore dressing grade prediction model and the recovery rate calculation model to obtain key production indexes, obtaining the optimal cases under each group of sub-case sets through the foam velocity setting model and the evaluation and learning model, and storing the cases to complete the initialization of the case library;

[0008] Searching for cases of the new flotation condition case in the case library. If there is no matching case, adding the new case to the flotation condition data source case set, changing the number of clustering groups to K + 1, synchronously updating the case library to achieve case correction, and continuing case search. If a similar case is matched, obtaining the optimal matching case by combining the evaluation and learning model;

[0009] Export the optimal matching case from the case library for case reuse to obtain the optimal setting value of the foam velocity;

[0010] Based on the flotation production measurement and control platform, optimize the setting of the foam velocity and conduct experimental verification and analysis of the optimized setting of the foam velocity.

[0011] Furthermore, obtain multi-source data through several sensors of the flotation production measurement and control platform, where

[0012] The flotation working condition data includes the visual characteristics of the flotation foam surface, the characteristic of the operation variables in the flotation process, and the characteristic of the flotation operation index. Among them, the visual characteristics of the flotation foam surface are also called the foam state variable characteristics;

[0013] The visual characteristics of the flotation surface include the color characteristics of the foam surface, the foam size and the number of foams per unit area; the foam surface texture, the foam stability, and the foam flow velocity.

[0014] The characteristic of the operation variables in the flotation process includes the ore feeding amount, the aeration amount, the height of the pulp liquid level, and the dosage of the reagent;

[0015] The characteristic of the flotation operation index includes the ore feeding grade, the concentrate grade, the tailing grade, and the recovery rate.

[0016] Furthermore, the data preprocessing includes dividing the flotation working condition data into Gaussian variables and non-Gaussian variables through distribution analysis. The Gaussian variables judge the outliers through the 3σ principle, and the non-Gaussian variables judge the outliers through the box plot. Denote the outlier boundary as Limit the outliers according to this outlier boundary condition, and screen the characteristics of the flotation working conditions in combination with the Spearman correlation coefficient. The Spearman correlation coefficient is expressed as:

[0017]

[0018] Among them, the paired values of the two variables are ranked in order respectively. R i represents the rank of x aj , Q i represents the rank of x ij (i≠a), R i -Q i is the difference between the ranks of x aj , x [[ID=4,6]] ij (i≠a). The value range of the correlation coefficient r is -1≤r≤1.

[0019] Further, the obtaining of several flotation condition data source case sets includes selecting features with a correlation coefficient |r| ≤ 0.5 as the input features of the BPNN network, imputing missing values in the samples based on the BPNN, converting multi-source data into flotation condition features of a unified standard, and then performing data integration to obtain a flotation basic data source case set. Each source case is expressed as:

[0020] C j = <X j ; V Set,j >, j = 1, 2, …, n,

[0021] where V Set,j is the output of the j-th source case; X j is the input of the j-th source case,

[0022] X j is expressed as:

[0023] X j = (x 1j , x 2j , …, x mj ),

[0024] where X ij (i = 1, 2, …, m) is the value of the j-th input feature variable x i .

[0025] Further, the foam velocity setting model uses m features from the flotation foam state variable features and the flotation process operation variable features as the case input features x i (i = 1, 2, …, m) of the foam velocity setting model, sets a reference case B = (b1, …, b m ), representing the reference values of the x1 to x m working condition features. Then, all input features are normalized to obtain x' i (i = 1, 2, …, m) and B' = (b'1, …, b' m ). The influence factors of all input features on the key flotation indicators are preset as λ i (i = 1, 2, …, m), and the Euclidean distance / matching distance between each source case and the reference case is calculated:

[0026]

[0027] Then, in step two, the source case set is divided into K sub-case sets based on the clustering method, and the key production indicators concentrate grade η C (t), recovery rate γ(t), and concentrate output M C (t) obtained by combining the ore dressing grade prediction model and the recovery rate calculation model, as well as the measured concentrate output M C (t) measured by weighing are obtained.

[0028] Furthermore, let the new case of the flotation condition be \(T=(t_1,\ldots,t m )\), representing the current flotation condition data of \(x_1\sim x m Similar to the method of calculating the matching distance, the matching distance \(d j (j = 1,\ldots,K)\) between the new case and each classical case in the case library can be obtained. Then, the matching distance is converted into a similarity \(Sim j \in[0,1]\):

[0029]

[0030] Furthermore, through the similarity boundary condition \(Sim j \geq\lambda\), a case set \(M k (k = 0,1,2,\ldots,K)\) that matches the description features of the target case \(T\) is obtained from the case library, and case retrieval is completed;

[0031] If the number of matching cases \(k\geq2\), the matching case \(M j \) with the optimal performance index \(J\) is obtained from the cases matching the target case \(T\) by combining the DPSO algorithm and the set evaluation index \(J\);

[0032] If the number of matching cases \(k = 1\), the case \(M1\) is directly taken out for reuse;

[0033] If \(k = 0\), that is, no case similar to the description features of the target case \(T\) can be retrieved from the case library, the target case is added to the source case set, the clustering classification number \(K + 1\), and the source case set is clustered and stratified sampled into \(K + 1\) case subsets through the matching distance. Then, \(K + 1\) optimal typical cases are selected from the case subsets by combining the set evaluation index and the DPSO optimization algorithm to achieve case correction; and the \(K + 1\) corrected typical cases are stored for case storage, a case library with a quantity of \(K + 1\) is generated, and then new case retrieval is performed again to obtain the optimal matching case \(M j 。

[0034] Furthermore, the corresponding optimal matching case \(C j =X j ;V Set,j is taken out for reuse, and its corresponding case solution \(V Set,j is the set value \(V Set of the flotation foam speed, denoted as \(V Set =V Set,j to achieve the optimized setting of the foam speed.

[0035] Compared with the prior art, the beneficial effects of the present invention are as follows. Based on the multi-source flotation production condition data of the on-site copper flotation production measurement and control platform, the present invention realizes integrated modeling and setting by combining an index prediction model, an evaluation and learning model, and an intelligent optimization method, etc., to obtain the optimal setting value of the foam velocity. The comparison results with the manual setting experiment show that the setting value of the foam velocity and the fluctuation degree of the flotation grade of this method are smaller. Compared with the manual setting, the qualified rate of the comprehensive production index is increased by 12%, and the production increase rate of the concentrate output is increased by 13.9%, which has certain application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 FIG. is a flowchart of a foam velocity optimization setting method for integrating multi-source flotation condition information;

[0037] Figure 2 FIG. is a topological structure diagram of foam velocity setting based on a flotation production measurement and control platform;

[0038] Figure 3 FIG. is a structure diagram of a foam velocity optimization setting method;

[0039] Figure 4 FIG. is a structure diagram of a foam velocity optimization setting model and an evaluation and learning model;

[0040] Figure 5 FIG. is a comparison waveform diagram of concentrate grade between the optimization setting method and the manual setting method;

[0041] Figure 6 FIG. is a comparison waveform diagram of metal recovery rate between the optimization setting method and the manual setting method;

[0042] Figure 7 FIG. is a comparison waveform diagram of concentrate output between the optimization setting method and the manual setting method;

[0043] Figure 8 FIG. is a comparison waveform diagram of foam velocity between the optimization setting method and the manual setting method. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0044] In order to make the objectives and advantages of the present invention clearer and more understandable, the present invention will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0045] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present invention and do not limit the protection scope of the present invention.

[0046] It should be noted that in the description of the present invention, the terms indicating directions or positional relationships such as "upper", "lower", "left", "right", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings. This is only for convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation to the present invention.

[0047] In addition, it should also be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installation", "connection", and "coupling" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0048] Please refer to Figure 1 as shown, which is a flowchart of the foam velocity optimization setting method for integrating multi-source flotation working condition information in an embodiment of the present invention.

[0049] The foam velocity optimization setting method for integrating multi-source flotation working condition information provided by the embodiment of the present invention specifically includes the following implementation steps:

[0050] Step S1: Collect flotation working condition data through a flotation production measurement and control platform, and perform data preprocessing to obtain several flotation working condition data source case sets;

[0051] Step S2: Based on the flotation working condition matching distance, combine the K-Means clustering method to obtain K groups of subclass case sets, combine the ore dressing grade prediction model and the recovery rate calculation model to obtain key production indexes, and obtain the optimal cases under each group of sub-case sets through the foam velocity setting model and the evaluation and learning model, and perform case storage to complete the initialization of the case library;

[0052] Step S3: Retrieve the new flotation working condition case in the case library. If there is no matching case, add the new case to the flotation working condition data source case set, the number of clustering groups becomes K + 1, synchronously update the case library, implement case correction, and continue case retrieval. If a similar case is matched, combine the evaluation and learning model to obtain the optimal matching case;

[0053] Step S4: Export the optimal matching case from the case library, perform case reuse, and obtain the optimal setting value of the foam velocity;

[0054] Step S5: Perform foam velocity optimization setting and experimental verification and analysis of the foam velocity optimization setting based on the flotation production measurement and control platform.

[0055] AsFigure 2 As shown

[0056] Specifically, multi-source characteristic data of the flotation working condition is obtained through the flotation production measurement and control platform. The characteristic data of the foam state variable is obtained by the platform camera to acquire the image data of the foam on the flotation surface and is obtained through a vision algorithm. The characteristic data of the foam state variable in the embodiment includes the foam velocity V f (mm / s), foam stability Std(%), and foam quantity N f (no. / dm 2 ); the characteristic of the operation variable in the flotation process is obtained through relevant sensors. The characteristic of the operation variable in the flotation process in the embodiment includes the ore feed amount M P (t / h), the aeration rate Q A (m 3 / h), the height h of the pulp layer liquid level P (m), the concentrate amount M C (t / h), and the tailings amount M T (t / h). The ore source of the ore feed is basically stable with little fluctuation. The ore feed grade η P (%) can be obtained by an empirical calculation model. First, in order to provide prior information on the data distribution, the data is analyzed for distribution. The Gaussian variable uses the 3σ principle to judge outliers, and the non-Gaussian variable uses the box plot to judge outliers. Since outliers may contain abnormal information about the flotation working condition fluctuations, on the one hand, in order to retain the information volume of the data in the preprocessing process, and on the other hand, the data should not deviate too much from the flotation working condition law, so the outlier boundary conditions are used to process the outliers. Finally, in order to perform feature selection and avoid multicollinearity, decorrelation processing is performed to reduce the calculation cost. Therefore, the Spearman correlation coefficient is combined to screen the flotation working condition characteristics, and the characteristics with the correlation coefficient |r|≤0.5 are selected as the input characteristics x i (t)(i = 1, 2,..., m), and then the missing values of the samples are imputed based on BPNN. In order to remove redundant or duplicate flotation working condition cases, the multi-source data is converted into flotation working condition characteristics of a unified standard, and the flotation basic data source case set is constructed through data integration, and the data preprocessing is completed

[0057] In the embodiment of the present invention, the Spearman correlation coefficient is expressed as:

[0058]

[0059] where, the paired values of the two variables are ranked in order respectively. R i represents the rank of x aj , and Q i represents the rank of x[[ID=4,2]] ij (i≠a), and R i -Qi is x aj and x ij (i≠a), and the value range of the correlation coefficient r is -1 ≤ r ≤ 1.

[0060] In the embodiments of the present invention, ranking in sequence includes ranking from large to small or from small to large.

[0061] In the embodiments of the present invention, the flotation working condition data includes the visual characteristics of the flotation foam surface, the characteristics of the operation variables in the flotation process, and the characteristics of the flotation operation indexes. Among them, the visual characteristics of the flotation foam surface are also called the characteristics of the foam state variables. The visual characteristics of the flotation surface include: the color characteristics of the foam surface, such as the foam yellowness component, the average R / G / B value, the H / S / V value, and the gray scale; the foam size and the number of foams per unit area; the foam surface texture, such as the energy, entropy, and contrast of the foam; the foam stability; the foam flow velocity, simply referred to as the foam velocity; and the characteristics of the operation variables in the flotation process include the ore feed amount, the aeration amount, the height of the pulp level, and the chemical addition amount; the characteristics of the flotation operation indexes include the ore feed grade, the concentrate grade, the tailing grade, and the recovery rate.

[0062] such as Figure 3 shown, assuming that V Set is the set value of the foam velocity, and x1(t) to x6(t) are the flotation working condition characteristic data, which are respectively the foam stability S td and the number of foams per unit volume N f and the aeration amount Q A and the height h of the pulp layer P and the ore feed amount M P and the ore feed grade η P , and Ω(t) is an unmeasurable interference factor, such as the heat value fluctuation in the flotation cell, etc. and are the requirements indexes of the flotation production plan for this flotation cell, which are respectively the production plan target values of the concentrate output, the concentrate grade, and the recovery rate. If there are n groups of historical data, there are correspondingly n set values of the foam velocity V Set,j (j = 1, 2,..., n), and then they are composed into a basic data source case set. Each source case can be expressed as:

[0063] C j = <X j ; V Set,j >, j = 1, 2,..., n, (2)

[0064] Among them, V Set,j is the output of the jth source case (the foam velocity, i.e., the set value); X j is the input of the jth source case (i.e., the x1 to x6 historical data), and can be expressed as:

[0065] Xj =(x 1j ,x 2j ,x 3j ,x 4j ,x 5j ,x 6j ), (3)

[0066] Among them, X ij (i=1,2,…,6) is the jth input feature variable x i value.

[0067] Specifically, the froth speed setting model uses the flotation froth state variable characteristics and the six flotation operating condition characteristics in the flotation process operation variable characteristics as the case input characteristics x of the froth speed setting model i (i=1,2,…,6), set the benchmark case B=(b1,…,b6), which represents the benchmark value of the working condition characteristics of x1~x6. In order to eliminate the influence of the flotation working condition dimension, all input features are then normalized to obtain x′ i (i=1,2,…,6) and B′=(b′1,…,b′6), assuming that the influence factor of all input features on the flotation key indicators corresponds to λ i (i=1,2,…,6)=1, calculate the Euclidean distance / matching distance between each source case and the reference case:

[0068]

[0069] like Figure 3 As shown,

[0070] Specifically, in step S2, by selecting the operating variable characteristics and the operating index characteristics, the concentrate grade η can be predicted by combining the flotation condition data with the mineral processing grade prediction model. C and tailings grade η T : The ore dressing grade prediction model can be expressed as:

[0071]

[0072] Among them, η C (t) is the concentrate grade of the flotation cell at the current moment, η T (t) is the tailings grade of the flotation tank at the current moment, η C (t-1) is the concentrate grade in the flotation cell at the previous moment, η T (t-1) is the tailings grade in the flotation tank at the previous moment; other related variables are characteristic data at the current moment, V LP is the volume of slurry in the slurry layer, V BP is the volume of gas in the slurry layer, V LF is the volume of slurry in the foam layer, V BFis the gas volume in the foam layer, Q A is the air flow rate, Q C is the flow rate of mineral particles to the concentrate pool, Q T is the flow rate of mineral particles to the tailings pool; η C is the concentrate grade at the current moment in the flotation cell, η T is the tailings grade at the current moment in the flotation cell; M P is the incoming ore mass, M C is the concentrate mass, M T is the tailings mass, the operating condition parameters [ω1, ω2] are constants, taking [0.88512, -0.17754]; the system parameters related to the model are expressed as:

[0073] S = (d T / 2) 2 *π, V T = Sh T , (6)

[0074] V BP = Sh P -V LP , (7)

[0075]

[0076]

[0077] V LF = V T -Sh P -V BF , (10)

[0078] wherein, the flotation cell is cylindrical, S = 2.25π is the bottom area of the flotation cell (m 2 ), d T = 3 is the diameter of the flotation cell (m), V T = 19.125π is the total volume of the flotation cell (m 3 ), h T = 8.5 is the height of the flotation cell (m), h P is the liquid level height of the pulp layer;

[0079] Combined with the incoming ore grade η P , the recovery rate γ is obtained:

[0080]

[0081] Furthermore, in the second step, the source case set is divided into K sub-case sets based on the clustering method, and the key production index concentrate grade η C(t), recovery rate γ(t), and concentrate output M obtained by measurement and weighing C (t), in order to obtain typical cases for foam velocity setting, the matching cases that optimize the ore dressing performance index J in each subset of sub-cases are obtained through the foam velocity setting value evaluation and learning model. Specifically: initialize the basic parameters of the DPSO algorithm, the position and velocity of the particle swarm, calculate the dynamic inertia weight value of the particle swarm, update the velocity value and position of the particle swarm through iteration, use the set value performance evaluation index J as the fitness function of DPSO, update the individual optimal position and the global optimal position, that is, update the index value under the case that optimizes the set value performance evaluation index J. Finally, the case can be retrieved from the subset of sub-cases through the index value of the optimal case, and the case is stored to construct a typical case set. Among them, in order to measure the performance of the foam velocity setting value, combining the key production indexes of flotation and the flotation production plan indexes, the performance index J can be expressed as:

[0082]

[0083] Among them, ω is the weight factor, taking 0.5.

[0084] As Figure 4 shown, <![CDATA[

[0085] ]]>Specifically, in the step S3, let the new case T of the flotation working condition be T=(t1,…,t6), representing the current working condition data of x1~x6 flotation. Similar to the method of calculating the matching distance in step two, the matching distance d between the new case and each classical case in the case library can be obtained j (j = 1,…,K). In order to efficiently retrieve the matching cases with similar working conditions to the current new flotation working condition case from the case library, the similarity method is used for retrieval. Therefore, the matching distance is converted into similarity Sim j ∈[0,1]:

[0086]

[0087] Specifically, in the step S3, in order to obtain matching cases within the allowable range of similar working conditions, so through the similarity boundary condition Sim j ≥λ(λ = 0.8), the case set M that matches the description features of the target case T is obtained from the case library k (k = 0,1,……,K), and the case retrieval is completed. If the number of matching cases k≥2, in order to obtain the excellent cases with the best performance from all matching cases, then combine the DPSO algorithm and the set value evaluation index J to obtain the matching case M with the best performance index J from the cases that match the target case T j .

[0088] Specifically, in the step S3, if the number of matching cases k = 1, the case M1 is directly taken out for reuse. If k = 0, that is, no case similar to the target case description feature T is retrieved from the case library, in order to increase the diversity and adaptability of typical cases in the flotation case library, the target case is added to the source case set, the clustering classification number K is incremented by 1, and the source case set is clustered and stratified sampled into K + 1 case subsets according to the matching distance. Then, K + 1 optimal typical cases are selected from the case subsets by combining the set evaluation index and the DPSO optimization algorithm to achieve case correction. Next, the K + 1 corrected typical cases are stored to generate a case library with a quantity of K + 1, and then the retrieval of new cases is performed again to obtain the optimal matching case M j .

[0089] Specifically, in the step S4, the corresponding optimal matching case C j =X j ; V Set,j is taken out for reuse, and its corresponding case solution V Set,j is the set value V Set of the flotation foam speed, denoted as V Set =V Set,j , realizing the optimal setting of the foam speed.

[0090] Specifically, in the step S5, by setting the optimized value V Set of the foam speed on the flotation production measurement and control platform, combining the fuzzy processing, expert control strategy, and defuzzification processing of control variables, the target values of the aeration rate and the pulp layer liquid level are obtained, and PID control is performed to achieve the automatic control of flotation.

[0091] The present invention conducts experiments on the multi-source flotation production condition data of the on-site copper flotation production measurement and control platform, such as Figure 5 , Figure 6 , Figure 7 , Figure 8 , as shown in Table 1, Table 2, Table 3, and Table 4. The set value of the foam speed and the fluctuation degree of the flotation grade of this method are smaller. Compared with manual setting, the qualified rate of the comprehensive production index is increased by 12%, and the production increase rate of the concentrate output is increased by 13.9%, having certain application value.

[0092] Table 1 Comparison of concentrate grade effect data of setting methods

[0093]

[0094] Table 2 Comparison of metal recovery rate effect data of setting methods

[0095]

[0096] Table 3 Comparison of concentrate output effect data of setting methods

[0097]

[0098] Table 4 Comparison of Total Effect Data of Setting Methods

[0099]

[0100] So far, the technical solution of the present invention has been described in conjunction with the preferred embodiments shown in the drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0101] The above are only the preferred embodiments of the present invention and are not used to limit the present invention; for those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for optimizing the setting of froth velocity by integrating multi-source flotation operating condition information, characterized in that: include: The flotation working condition data is collected through the flotation production measurement and control platform, and the data is preprocessed to obtain several flotation working condition data source case sets; Based on the flotation condition matching distance, the K-Means clustering method is combined to obtain K groups of sub-class case sets. The mineral processing grade prediction model and recovery rate calculation model are combined to obtain key production indicators. The foam velocity setting model and the evaluation and learning model are used to obtain the optimal case in each sub-case set, store the case, and complete the case library initialization. The new flotation case is searched in the case library. If there is no matching case, the new case is added to the flotation source case set, the number of clustering groups becomes K+1, the case library is updated synchronously, the case is corrected, and the case search is continued. If a similar case is matched, the evaluation and learning model are combined to obtain the best matching case. The optimal matching case is exported from the case library, and the case is reused to obtain the optimal setting value of the foam speed; Conducting optimization of foam velocity settings based on the flotation production measurement and control platform, as well as experimental verification and analysis of the optimization of foam velocity settings; Let the new flotation case T=(t1,…,t m ), indicating x1~x m Float the current working condition data and calculate the matching distance in the same way as the matching distance method to obtain the matching distance d between the new case and each classic case in the case library j (j=1,…,K), convert the matching distance into similarity Sim j ∈[0,1]: Through the similarity boundary condition Sim j ≥λ Obtain a case set M that matches the description features of the target case T from the case library k (k=0,1,2,…,K), complete case retrieval; Initialize the basic parameters of the DPSO algorithm, the position and velocity of the particle swarm, calculate the dynamic inertia weight value of the particle swarm, iteratively update the velocity value and position of the particle swarm, use the set value performance evaluation index J as the fitness function of DPSO, update the individual optimal position and the global optimal position, that is, update the index value of the optimal case under the set value performance evaluation index J, and finally use the index value of the optimal case to extract the case from the sub-case set, store the case, and build a typical case set; If the number of matching cases k≥2, the DPSO algorithm and the set value evaluation index J are combined to obtain the matching case M with the best performance index J from the cases matching the target case T. j ; If the number of matching cases k = 1, then the M1 case is directly taken out and reused; If k = 0, that is, no case with similar description feature T as the target case can be retrieved in the case library, then the target case is added to the source case set, the cluster classification number is K+1, and the source case set is clustered and stratified by matching distance and divided into K+1 case subsets. The set value evaluation index and DPSO optimization algorithm are combined to select K+1 optimal typical cases from the case subset to achieve case correction; and the K+1 corrected typical cases are stored as cases to generate a case library with a number of K+1, and then the new case is searched again to obtain the optimal matching case M j .

2. The method for optimizing the setting of the foam velocity by integrating multi-source flotation operating condition information according to claim 1, characterized in that: Multi-source data is obtained through several sensors of the flotation production measurement and control platform, including: Flotation condition data include flotation foam surface visual features, flotation process operation variable features and flotation operation index features, among which flotation foam surface visual features are also called foam state variable features; The visual characteristics of the flotation surface include the color characteristics of the foam surface, foam size and the amount of foam per unit area; foam surface texture, foam stability, foam flow rate, The operational variable characteristics of the flotation process include the amount of ore fed, the amount of aeration, the height of the slurry level, and the amount of reagent added; The flotation operation index characteristics include feed grade, concentrate grade, tailings grade and recovery rate.

3. The method for optimizing the setting of the foam velocity by integrating multi-source flotation operating condition information according to claim 1, characterized in that: The data preprocessing includes dividing the flotation condition data into Gaussian variables and non-Gaussian variables through distribution analysis, judging the outliers of Gaussian variables by the 3σ principle, and judging the outliers of non-Gaussian variables by box plots, and recording the outlier boundaries as The outlier boundary condition is used to limit the outlier, and the Spearman correlation coefficient is used to screen the flotation working condition characteristics. The Spearman correlation coefficient is expressed as: Among them, for two variables x aj and x ij The paired values are ranked in order, R i Represents x aj The rank, Q i Represents x ij The rank of (i≠a), R i -Q i is x aj 、x ij The rank difference of (i≠a), the correlation coefficient r range is -1≤r≤1.

4. The method for optimizing the setting of froth velocity by integrating multi-source flotation operating condition information according to claim 1, characterized in that: The method of obtaining a plurality of flotation condition data source case sets includes selecting features with a correlation coefficient |r| ≤ 0.5 as input features of the BPNN network, interpolating missing sample values based on the BPNN, converting multi-source data into flotation condition features of a unified standard, and then performing data integration to obtain a flotation basic data source case set, where each source case is represented as: C j =<X j ;V Set,j >,j=1,2,…,n, Among them, V Set,j Output for the jth source case; X j For the j-th source case input, X j Expressed as: X j =(x 1j ,x 2j ,…,x mj ), Among them, X ij (i=1,2,…,m) is the jth input feature variable x i value.

5. The method for optimizing the setting of foam velocity by integrating multi-source flotation operating condition information according to claim 1, characterized in that: The foam speed setting model uses m features of the flotation foam state variable features and the flotation process operation variable features as case input features x of the foam speed setting model i (i=1,2,…,m), set the benchmark case B=(b1,…,b m ), indicating x1~x m The working condition characteristic baseline value, then all input features are normalized to obtain x′ i (i=1,2,…,m) and B′=(b′1,…,b′ m ), the influence factor of all input features on the key flotation indicators is preset to be λ i (i=1,2,…,m), calculate the Euclidean distance / matching distance between each source case and the reference case: Then, the source case set is divided into K sub-case sets based on the clustering method, and the key production indicator concentrate grade η is obtained by combining the mineral processing grade prediction model and the recovery rate calculation model. C (t), recovery rate γ(t), and the concentrate yield M obtained by measurement and weighing C (t).

6. The method for optimizing the setting of froth velocity by integrating multi-source flotation operating condition information according to claim 1, characterized in that: The corresponding optimal matching case C j =X j ; V Set,j Take out and reuse, the corresponding case solution V Set,j Set the value V for the flotation foam velocity Set , denoted as V Set =V Set,j , to achieve optimal setting of foam speed.

Images