Mineral flotation process industrial knowledge graph construction method based on multi-level cross-process association

Abstract

The invention discloses a mineral flotation industrial knowledge graph construction method based on multi-level cross-process association. The flotation process is systematically expressed by combining ontology modeling and a tree structure. According to the method, the flotation process is regarded as a dynamic tree, root nodes are used for representing feeding parameters, tree trunk and branch nodes correspond to main process steps and sub-processes of the main process steps respectively, and leaf nodes represent process parameters and performance index feedback. According to the method, the ontology is introduced into the knowledge graph, semantic dynamic modeling is performed on process data, process association and input-output relationships, and unified description and integrated expression of multi-source heterogeneous data are realized by using ontology semantics. Through the method, the problem of visual modeling of multi-parameter complex association in the flotation process is effectively solved, a clear process expression and knowledge organization form is provided, and a foundation is laid for intelligent analysis and optimization of the mineral flotation process.

Description

Technical Field

[0001] The present invention relates to the field of mineral flotation, and in particular to a method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process associations. Background Art

[0002] Mineral flotation is a mineral processing technology widely used in mineral separation and enrichment, involving multiple processes, parameters and their dynamic relationships. Traditional process optimization methods are mostly based on empirical rules or single process analysis, which makes it difficult to effectively model complex multi-level parameter relationships and feedback mechanisms between processes. In addition, existing industrial knowledge graphs usually adopt static network structures, which cannot adapt to the dynamic changes in input-output relationships and the expression requirements of upstream and downstream relationships in flotation processes, and it is difficult to quantify complex relationships. Therefore, improvements are needed. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a method for constructing an industrial knowledge graph of mineral flotation process based on multi-level cross-process associations.

[0004] The purpose of the present invention is achieved through the following technical solutions:

[0005] A method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process associations comprises the following steps:

[0006] Step 1: Collect process data in flotation process;

[0007] Step 2: construct a flotation tree model and define the input-output relationship of the flotation tree model according to process data;

[0008] Step 3: Use ontology modeling to semantically describe the concepts and relationships in the flotation tree model to form an ontology modeled flotation tree model

[0009] Step 4: Use the graph database Neo4j to construct a knowledge graph based on the flotation tree model of the ontology modeling, and use visualization tools to display the graph.

[0010] As a further improvement, in step one, the process data includes raw material parameters, particle size distribution, moisture content, equipment status, environmental variables and reagent dosage.

[0011] In a further improvement, the raw material parameters include the type of ore minerals; the equipment status includes the stirring tank rotation speed and the bubble generator gas flow rate; the environmental variables include the slurry temperature and pH value; and the reagent dosage includes the collector dosage and the foaming agent dosage.

[0012] As a further improvement, the process data is normalized and preprocessed, and the specific method is as follows: outliers and duplicate values ​​in the ore flotation process are removed and then the data in different formats in the ore flotation process are converted into a unified structured format.

[0013] For further improvement, the specific steps of step 2 are as follows:

[0014] Step S21: The flotation tree model hierarchy is defined as follows: root node: mineral feed parameters, which include mineral type, particle size distribution and moisture content; trunk node: flotation process steps, which include fast roughing, roughing, fine separation and scavenging; branch node: sub-process parameters, which include bubble distribution, pulp stability and pulp flow rate; leaf node: final output production index, which includes concentrate grade and concentrate recovery rate;

[0015] Step S22: define the input-output relationship of the flotation tree model:

[0016] Set up the flotation tree model nutrient delivery: Set the final concentrate concentration of the target mineral to C 精矿 , the transfer process considers the efficiency factor of each step and the possible loss factor, and the formula is expressed as:

[0017]

[0018] Where: C 精矿 is the final concentrate concentration, C 原矿 is the original ore concentration, E 步骤i is the flotation efficiency factor of the i-th step, L 步骤i is the loss factor of the i-th step, k j is the influence coefficient of slurry pH value on the efficiency of each process step, pH j is the pH value of step j, pH opt,j is the optimal pH value of the jth step; n is the total number of flotation steps, and m is the number of steps actually passed during the flotation process.

[0019] Further improvement, efficiency factor E 步骤i and loss factor L 步骤i , modeled by the following dynamic formula:

[0020]

[0021] L 步骤i =α i ×C 步骤i,输入

[0022] Where: C 步骤i,输入 and C 步骤i,输出 are the input and output mineral concentrations of the i-th step, α i is the loss coefficient of the i-th step, reflecting the mineral loss caused by operating conditions during the flotation process; the operating conditions include the amount of reagent used and the pulp concentration.

[0023] Further improvement is to make the scanning step dynamic:

[0024]

[0025] Where: C 原矿 is the concentration of the target mineral in the original ore, C 尾矿 is the concentration of target mineral in tailings, β pH and β 温度 are the coefficients of the influence of pulp pH value and temperature on recovery rate, pH opt and T opt are the optimal pH value and temperature, respectively, and pH and T are the current pH value and temperature of the slurry, respectively.

[0026] For further improvement, the specific steps of step three are as follows:

[0027] Step S31: define the basic structure of the ontology model: concept classes and relationship classes; the concept classes include lead minerals, zinc minerals and flotation cells, and the relationship classes include the collector affecting the lead recovery rate and the pH value regulating the slurry stability;

[0028] The amount of collector added is X 捕收剂 and recovery rate R 回收 The relationship between is modeled by the following formula:

[0029]

[0030] Among them, R 基准 is the baseline recovery, i.e., the recovery without collector, a and b are experimental fitting parameters, representing the nonlinear effect of collector on the recovery;

[0031] The relationship between pH value and slurry stability is expressed by the following formula:

[0032] L 稳定性 =k pH ·(pH-pH opt ) 2

[0033] Among them, k pH is the stability response factor, pH opt The optimal pH value.

[0034] As a further improvement, the step 4 includes the following steps:

[0035] Step S41: knowledge graph construction, importing the flotation tree model of ontology modeling into Neo4j in RDF / XML format, establishing a bidirectional dynamic relationship between nodes and edges, where nodes include mineral types, reagents and flotation cells; edges include process parameter associations and efficiency transfer.

[0036] Step S42: Knowledge graph visualization: The graph dynamically displays the transfer process of the mineral flotation process from raw materials to final products; visualizes the relationship between collector dosage and lead concentrate grade; and dynamically affects the pH value of the slurry in each process tank on the recovery rate of the target mineral.

[0037] The method for constructing an industrial knowledge graph for a mineral flotation process based on multi-level cross-process associations as described in claim 1, wherein the mineral flotation includes lead-zinc ore flotation.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] The present invention uses a knowledge graph construction method with multi-level cross-process associations to abstract the mineral flotation process flow into a hierarchical tree structure, and combines ontology modeling technology to express the processes, parameters and their relationships in a semantic way, thus overcoming the deficiency that traditional static networks are difficult to adapt to the dynamic changes of flotation processes. By introducing a unified integration and dynamic modeling mechanism for multi-source data, the problem of multi-variable and multi-level parameter associations being difficult to quantify in complex processes is effectively solved, while achieving accurate expression of logical associations and dynamic feedback of upstream and downstream processes. The present invention can not only comprehensively and systematically characterize the input-output relationship of the flotation process, but also support the discovery of hidden influencing factors through semantic reasoning and association mining, providing a scientific basis for the optimal design and intelligent control of the flotation process, and improving the efficiency and accuracy of flotation process modeling and analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The present invention is further described with reference to the accompanying drawings, but the contents in the accompanying drawings do not constitute any limitation to the present invention.

[0041] Figure 1 This is a system structure diagram of the present invention, showing a schematic diagram of the overall process from data collection to knowledge graph generation and optimization.

[0042] Figure 2 It is a schematic diagram of the tree model of the present invention, which represents a schematic diagram of the hierarchical structure of the flotation process.

[0043] Figure 3 This is a diagram of the bidirectional transmission mechanism of the present invention, simulating a dynamic flow diagram of "transmitting feedback downward and transmitting demand upward". DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples.

[0045] Example 1

[0046] See Figure 1The present invention provides a method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association, which at least includes the following steps:

[0047] Step S1: Collect and integrate multi-source data in the flotation process to provide data support for subsequent modeling.

[0048] Step S2: Based on the bidirectional dynamic process characteristics of the flotation process, a tree model with a hierarchical association relationship is constructed.

[0049] Step S3: The concepts and relationships in the flotation process are semantically described through ontology modeling, so as to achieve accurate construction of the knowledge graph.

[0050] Step S4: Build a knowledge graph through the graph database Neo4j and use visualization tools to display the graph.

[0051] Specifically in a lead flotation embodiment of the present invention, a method for constructing a multi-level, cross-process-related industrial knowledge graph of a mineral flotation process of the present invention comprises the following steps:

[0052] Step S1: The data of lead-zinc ore flotation process is used as the main source, including production data and relevant technical information collected at the mine site. Step S1 specifically includes the following steps:

[0053] Step S11: Collect key data in the flotation process of lead-zinc ore. Raw material parameters: lead-zinc ore mineral type (sphalerite, galena, pyrite, etc.), particle size distribution (0.01-0.1mm, 0.1-1mm, etc.), moisture content (8%, 10%, etc.); equipment status: stirring tank speed (200rpm, 300rpm, etc.), bubble generator gas flow (0.2m 3 / min, etc.); environmental variables: slurry temperature (25°C, etc.), pH value (7.5-9.0, etc.); reagent dosage: collector (ethylthiocarb, 50g / t, 100g / t, etc.), foaming agent (pine oil, 20g / t, 30g / t, etc.).

[0054] Step S12: pre-processing the collected data. Data cleaning: removing abnormal values ​​and duplicate values ​​in the lead-zinc ore flotation process; data formatting: converting data of different formats in the lead-zinc ore flotation process into a unified structured format.

[0055] Step S2: Combined with the lead-zinc ore flotation process, based on the bidirectional dynamic process construction concept of the flotation tree model, a hierarchical association model is designed. Step S2 specifically includes the following steps:

[0056] Step S21: Step S21: Model level definition. Root node: mineral feed parameters, such as mineral type, particle size distribution, moisture content, etc.; Trunk node: main flotation process steps, including fast roughing, roughing, fine flotation, and scavenging; Branch node: sub-process parameters, such as bubble distribution, slurry stability, slurry flow rate, etc.; Leaf node: final output production indicators, such as lead concentrate grade and zinc concentrate recovery rate.

[0057] Step S22: define the input-output relationship of the flotation tree model.

[0058] Flotation tree model nutrient delivery, setting the final concentration of target mineral to C 精矿 , the transfer process takes into account the efficiency factor of each step and the possible loss factor, and this transfer process is regulated by multiple dynamic parameters, taking into account the interaction between different steps in the flotation process and the influence of environmental factors (such as pH value, temperature, etc.), and the formula can be expressed as:

[0059]

[0060] Where: C 精矿 is the final concentrate concentration, C 原矿 is the original ore concentration, E 步骤i is the flotation efficiency factor of the i-th step, L 步骤i is the loss factor of the i-th step, k j is the influence coefficient of slurry pH value on the efficiency of each process step, pH j is the pH value of step j, pH opt,j is the optimal pH value for step j.

[0061] Furthermore, the efficiency factor E 步骤i and loss factor L 步骤i , can be modeled by the following dynamic formula:

[0062]

[0063] L 步骤i =α i ×C 步骤i,输入

[0064] Where: C 步骤i,输入 and C 步骤i,输出 are the input and output mineral concentrations of the i-th step, α i is the loss coefficient of the i-th step, reflecting the mineral loss caused by operating conditions (such as reagent usage, pulp concentration, etc.) during the flotation process.

[0065] The absorption mechanism of the flotation tree model is the scavenging recovery rate. In the scavenging step, the residual target mineral is recovered. The scavenging recovery rate R 扫选It reflects the recovery degree of residual minerals in tailings. The formula is:

[0066]

[0067] Where: C 原矿 is the concentration of the target mineral in the original ore, C 尾矿 is the concentration of target mineral in tailings.

[0068] In addition, the scavenging recovery rate is also affected by environmental variables such as pulp concentration and pH value, which can be further dynamically processed through the following model:

[0069]

[0070] Where: β pH and β 温度 is the coefficient of the effect of pulp pH and temperature on recovery rate, pH opt and T opt are the optimal pH value and temperature, pH and T are the current pH value and temperature of the slurry, respectively.

[0071] Step S3: Based on the semantic description of ontology modeling, combined with the flotation tree model and the dynamic mechanism of the tree, ontology modeling is used to realize the semantic description of the flotation process. Step S3 specifically includes the following steps:

[0072] Step S31: define the basic structure of the flotation tree model ontology. Concept class: such as "lead mineral", "zinc mineral", "flotation cell", etc.; relationship class: such as "collector affects lead recovery rate", "pH value regulates pulp stability";

[0073] The relationship between collector and recovery rate, the amount of collector added X 捕收剂 and recovery rate R 回收 The relationship between is usually nonlinear and can be modeled by the following formula:

[0074]

[0075] Among them, R 基准 is the baseline recovery rate (the recovery rate without collector), a and b are the experimental fitting parameters, which represent the nonlinear effect of collector on the recovery rate.

[0076] The relationship between pH value and slurry stability. The effect of pH value on slurry stability can be expressed by the following formula:

[0077] L 稳定性 =k pH ·(pH-pH opt ) 2

[0078] Among them, kpH is the stability response factor, pH opt The optimal pH value.

[0079] Step S32: Instantiate the concepts in the flotation tree model ontology and correspond them to the tree model nodes. Based on the ontology model defined in step S31, map the specific flotation process data to the ontology concepts and generate instances. For example, the concept of "lead mineral" is instantiated as "galena" and "sphalerite", and the "flotation cell" is instantiated as "fast roughing cell", "fine cell", etc. Use the Protege tool to define instance attributes, for example: the attributes of the galena instance: particle size range (0.01-0.1mm), content (4.5%). The attributes of the flotation cell instance: processing capacity (500t / d), bubble generator flow (0.2m 3 / min). Then, each node in the tree model is mapped one by one to the concept instance in the ontology model, and the description logic (OWL DL) is used to define association rules for each mapping node, such as: if the flow rate of the bubble generator of a flotation tank exceeds 0.3m 3 / min, a “low stability” mark is added to the slurry stability node; if the collector dosage increases by 10g / t, the estimated value of the lead concentrate recovery rate in the leaf node is dynamically updated.

[0080] Step S4: Knowledge graph construction and visualization. Step S4 specifically includes the following steps:

[0081] Step S41: knowledge graph construction, importing the ontology model into Neo4j in RDF / XML format, and establishing a bidirectional dynamic relationship between nodes and edges. Nodes: mineral types, reagents, flotation cells, etc.; edges: process parameter association, efficiency transfer, etc.

[0082] Step S42: Visualization of knowledge graph. The graph dynamically displays the transfer process of lead-zinc ore flotation process from raw materials to final products; visualizes the relationship between collector dosage and lead concentrate grade; and dynamically affects the pH value of the slurry in each process tank on the recovery rate of the target mineral.

[0083] Through the above steps, the present invention not only realizes the full-process modeling of the lead-zinc ore flotation process, but also introduces a two-way dynamic process transmission mechanism of the flotation tree model, making the knowledge graph more dynamic and relevant, and providing a theoretical basis and technical support for complex process optimization.

[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the essence and scope of the technical solution of the present invention.

Claims

1. A method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association, characterized in that: The steps include: Step 1: Collect process data in flotation process; Step 2: construct a flotation tree model and define the input-output relationship of the flotation tree model according to process data; Step 3: Use ontology modeling to semantically describe the concepts and relationships in the flotation tree model to form an ontology modeled flotation tree model Step 4: Use the graph database Neo4j to construct a knowledge graph based on the flotation tree model of the ontology modeling, and use visualization tools to display the graph.

2. The method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association according to claim 1, characterized in that: In step 1, the process data includes raw material parameters, particle size distribution, moisture content, equipment status, environmental variables and reagent dosage.

3. The method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association according to claim 2, characterized in that: The raw material parameters include the type of ore minerals; the equipment status includes the stirring tank rotation speed and the bubble generator gas flow rate; the environmental variables include the slurry temperature and pH value; and the reagent dosage includes the collector dosage and the foaming agent dosage.

4. The method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association according to claim 1, characterized in that: The process data is normalized and preprocessed, and the specific method is as follows: outliers and duplicate values ​​in the ore flotation process are removed and then the data in different formats in the ore flotation process are converted into a unified structured format.

5. The method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association according to claim 1, characterized in that: The specific steps of step 2 are as follows: Step S21: The flotation tree model hierarchy is defined as follows: root node: mineral feed parameters, the mineral feed parameters include mineral type, particle size distribution and moisture content; trunk node: flotation process steps, the flotation process steps include fast roughing, roughing, fine separation and scavenging; branch node: sub-process parameters, the sub-process parameters include bubble distribution, pulp stability and pulp flow rate; leaf node: final output production indicators, the final output production indicators include concentrate grade and concentrate recovery rate; Step S22: define the input-output relationship of the flotation tree model: Set up the flotation tree model nutrient delivery: Set the final concentrate concentration of the target mineral to C 精矿 , the transfer process considers the efficiency factor of each step and the possible loss factor, and the formula is expressed as: Where: C 精矿 is the final concentrate concentration, C 原矿 is the original ore concentration, E 步骤i is the flotation efficiency factor of the i-th step, L 步骤i is the loss factor of the i-th step, k j is the influence coefficient of slurry pH value on the efficiency of each process step, pH j is the pH value of step j, pH opt,j is the optimal pH value of the jth step; n is the total number of flotation steps, and m is the number of steps actually passed during the flotation process.

6. The method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association according to claim 5, characterized in that: Efficiency Factor E 步骤i and loss factor L 步骤i , modeled by the following dynamic formula: L 步骤i =a i ×C 步骤i,输入 Where: C 步骤i,输入 and C 步骤i,输出 are the input and output mineral concentrations of the i-th step, α i is the loss coefficient of the i-th step, reflecting the mineral loss caused by operating conditions during the flotation process; the operating conditions include the amount of reagent used and the pulp concentration.

7. The method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association according to claim 5, characterized in that: Dynamic processing of the scanning step: Where: C 原矿 is the concentration of the target mineral in the original ore, C 尾矿 is the concentration of target mineral in tailings, β pH and β 温度 are the coefficients of the influence of pulp pH value and temperature on recovery rate, pH opt and T opt are the optimal pH value and temperature, respectively, and pH and T are the current pH value and temperature of the slurry, respectively.

8. The method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association according to claim 5, characterized in that: The specific steps of step three are as follows: Step S31: define the basic structure of the ontology model: concept classes and relationship classes; the concept classes include lead minerals, zinc minerals and flotation cells, and the relationship classes include the collector affecting the lead recovery rate and the pH value regulating the slurry stability; The amount of collector added is X 捕收剂 and recovery rate R 回收 The relationship between is modeled by the following formula: Among them, R 基准 is the baseline recovery, i.e., the recovery without collector, a and b are experimental fitting parameters, representing the nonlinear effect of collector on the recovery; The relationship between pH value and slurry stability is expressed by the following formula: L 稳定性 =k pH ·(pH-pH opt ) 2 Among them, k pH is the stability response factor, pH opt The optimal pH value.

9. The method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association according to claim 1, characterized in that: The step 4 comprises the following steps: Step S41: knowledge graph construction, importing the flotation tree model of ontology modeling into Neo4j in RDF / XML format, establishing a bidirectional dynamic relationship between nodes and edges, where nodes include mineral types, reagents and flotation cells; edges include process parameter associations and efficiency transfer. Step S42: Visualization of knowledge graph: The graph dynamically displays the transfer process of the mineral flotation process from raw materials to final products; visualizes the relationship between collector dosage and lead concentrate grade; and dynamically affects the pH value of the slurry in each process tank on the recovery rate of the target mineral.

10. The method for constructing an industrial knowledge graph of a mineral flotation process based on multi-level cross-process association according to claim 1, characterized in that: The mineral flotation includes lead-zinc ore flotation.

Images