Method and device for adjusting operation parameters in flotation process based on Bayesian reasoning
Through Bayesian inference method, filter historical cases, calculate similarity and build incremental databases, train Bayesian network model, solve the subjective arbitrary problem of operational parameter adjustment during flotation, and realize the quality stability of concentrate grade and accuracy.
Patent Information
- Application Number
- CN202411373944.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-29
- Publication Date
- 2024-11-26
AI Technical Summary
During the flotation process, the adjustment of operating parameters depends on manual visual judgment, and there are subjective arbitrary and difficult to adjust in a timely and accurate manner, which affects the quality of concentrate grade.
Using Bayesian reasoning based method, we use filtering historical cases that meet the concentrate quality standards, calculating similarity scores, building an incremental database, and training the Bayesian network model to automatically adjust the operating parameters.
The accurate and timely adjustment of operating parameters during flotation process is achieved, and the quality stability of concentrate grade is improved.
Smart Images

Figure CN119016209A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of flotation control, and in particular to a method and device for adjusting operating parameters in a flotation process based on Bayesian reasoning. Background Art
[0002] Froth flotation is a widely used beneficiation technology for nonferrous metals. It uses the difference in hydrophobicity of mineral particles to separate useful mineral particles from gangue. Due to the long production process and frequent changes in the properties of the raw ore, it is challenging to ensure the quality of the concentrate product within an acceptable range. At the flotation site, when the operating conditions change or the concentrate grade is unqualified, it is usually necessary to adjust the operating parameters to improve the concentrate grade.
[0003] Due to the complexity of the flotation process, current operational adjustments are still done manually. Operators adjust operating parameters by observing the visual characteristics of flotation foam, which is highly subjective and makes it difficult to adjust operating parameters in a timely and accurate manner to ensure qualified concentrate grade. Summary of the invention
[0004] In order to solve the above technical problems, the embodiments of the present application provide a method and device for adjusting operating parameters in a flotation process based on Bayesian reasoning.
[0005] In a first aspect, an embodiment of the present application provides a method for adjusting operating parameters in a flotation process based on Bayesian reasoning, the method comprising: Screening historical cases that meet the concentrate quality standard, and obtaining problem attributes and solution attributes of each of the historical cases; Calculate the similarity scores between the problem attributes of each of the historical cases and the problem attributes of the current query case, and obtain the most similar historical case and a preset number of neighboring similar historical cases according to the sorting results of the similarity scores; Calculating the difference between the problem attribute of the current query case and the solution attribute of the most similar historical case to obtain a rough operation adjustment solution; Building an incremental database based on the problem attributes and solution attributes of the most similar historical case and the problem attributes and solution attributes of the neighboring similar historical cases; Training a Bayesian network model based on the incremental database, and inputting the incremental data of the problem attributes of the current query case into the trained Bayesian network model to obtain a detailed operation adjustment solution; A target operation adjustment solution is obtained according to the coarse operation adjustment solution and the fine operation adjustment solution.
[0006] In one implementation, the obtaining of the problem attributes and solution attributes of the historical case includes: Calculating latent variables of the historical case according to the working condition data of the initial problem attributes in the historical case; Obtaining the operating parameters of the historical case before adjustment, and setting the latent variables of the historical case and the operating parameters before adjustment as the problem attributes of the historical case; The adjusted operating parameters of the historical case are obtained, and the adjusted operating parameters are set as solution attributes.
[0007] In one embodiment, before calculating the similarity score between the problem attribute of each of the historical cases and the problem attribute of the current query case, the method further includes: Performing standardization processing on the initial problem attributes of the historical case, wherein the standardization processing includes standardization processing of working condition data; The standardized working condition data is extracted through the sparse autoencoder, and the weight matrix and bias vector are obtained; Calculating the latent variables of the current query case according to the weight matrix, the bias vector and the operating condition data corresponding to the current query case; The problem attribute of the current query case is obtained according to the latent variables and the operating parameters before adjustment.
[0008] In one implementation, the calculating of the similarity score between the problem attribute of each historical case and the problem attribute of the current query case includes: Calculate the similarity score according to formula (1), the question attributes of the historical case and the question attributes of the current query case; Formula (1): ; in, is the similarity score between the historical case and the current query case, is the problem attribute of the current query case, is the problem attribute of the historical case.
[0009] In one embodiment, the step of constructing an incremental database based on the problem attributes and solution attributes of the most similar historical case and the problem attributes and solution attributes of the neighboring similar historical cases includes: Calculate the difference between the problem attribute of each neighboring similar historical case and the problem attribute of the most similar historical case to obtain incremental data of the problem attribute; Calculate the difference between the solution attribute of each neighboring similar historical case and the solution attribute of the most similar historical case to obtain solution attribute increment data; An incremental database is constructed based on the problem attribute incremental data and the corresponding solution attribute incremental data of each neighboring similar historical case.
[0010] In one embodiment, the training of the Bayesian network model according to the incremental database includes: Determine the Bayesian network model structure based on flotation experience knowledge; Inputting the problem attribute incremental data and the corresponding solution attribute incremental data of each of the neighboring similar historical cases in the incremental database into the Bayesian network model structure; The training is performed through the maximum likelihood estimation algorithm to obtain a trained Bayesian network model.
[0011] In one embodiment, the determining of the Bayesian network model structure based on flotation experience knowledge includes: Obtain operational variables related to concentrate quality; Based on the operational variables, determining nodes and edges in the Bayesian network model structure; Assigning a node type to each node in the Bayesian network model structure; The Bayesian network model structure is constructed according to the nodes, the node types and the edges using a graphical tool or a programming method.
[0012] In a second aspect, an embodiment of the present application provides a device for adjusting operating parameters in a flotation process based on Bayesian reasoning, and the device for adjusting operating parameters in a flotation process based on Bayesian reasoning includes: The first acquisition module is used to screen historical cases that meet the concentrate quality standard and obtain the problem attributes and solution attributes of each historical case; A first calculation module is used to calculate the similarity score between the problem attribute of each of the historical cases and the problem attribute of the current query case, and obtain the most similar historical case and a preset number of neighboring similar historical cases according to the sorting result of the similarity score; A second calculation module is used to calculate the difference between the problem attribute of the current query case and the solution attribute of the most similar historical case to obtain a rough operation adjustment solution; A construction module, configured to construct an incremental database according to the problem attributes and solution attributes of the most similar historical case and the problem attributes and solution attributes of the neighboring similar historical cases; A training module, used for training a Bayesian network model according to the incremental database, and inputting the incremental data of the problem attributes of the current query case into the trained Bayesian network model to obtain a detailed operation adjustment solution; The second acquisition module is used to acquire a target operation adjustment solution according to the coarse operation adjustment solution and the fine operation adjustment solution.
[0013] In a third aspect, an embodiment of the present application provides an electronic device including a memory and a processor, wherein the memory is used to store a computer program, and when the computer program is run by the processor, the method for adjusting operating parameters in a flotation process based on Bayesian reasoning provided in the first aspect is executed.
[0014] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium storing a computer program, which, when executed on a processor, executes the method for adjusting operating parameters in a flotation process based on Bayesian reasoning provided in the first aspect.
[0015] The above-mentioned method for adjusting operating parameters in the flotation process based on Bayesian reasoning provided by the present application screens each historical case that meets the concentrate quality standard, and obtains the problem attributes and solution attributes of each of the historical cases; calculates the similarity score between the problem attributes of each of the historical cases and the problem attributes of the current query case, and obtains the most similar historical case and a preset number of neighboring similar historical cases according to the sorting result of the similarity score; calculates the difference between the problem attribute of the current query case and the solution attribute of the most similar historical case to obtain a rough operation adjustment solution; constructs an incremental database based on the problem attributes and solution attributes of the most similar historical case and the problem attributes and solution attributes of the neighboring similar historical cases; trains a Bayesian network model based on the incremental database, and inputs the incremental data of the problem attributes of the current query case into the trained Bayesian network model to obtain a fine operation adjustment solution; obtains a target operation adjustment solution based on the rough operation adjustment solution and the fine operation adjustment solution. The present application obtains a rough operation adjustment solution through the most similar case, obtains a fine operation adjustment solution through the incremental database training Bayesian network model, and finally obtains a target operation adjustment solution, which ensures accurate and timely adjustment of operating parameters during flotation. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present application and should not be regarded as limiting the scope of protection of the present application. In each of the drawings, similar components are numbered similarly.
[0017] Figure 1 A schematic diagram of a process of adjusting operating parameters in a flotation process based on Bayesian reasoning provided in an embodiment of the present application is shown; Figure 2 Another schematic diagram of the process of adjusting operating parameters in a flotation process based on Bayesian reasoning provided in an embodiment of the present application is shown; Figure 3 Another schematic diagram of a process of adjusting operating parameters in a flotation process based on Bayesian reasoning provided in an embodiment of the present application is shown; Figure 4 Another schematic diagram of a process of adjusting operating parameters in a flotation process based on Bayesian reasoning provided in an embodiment of the present application is shown; Figure 5 Another schematic diagram of a process of adjusting operating parameters in a flotation process based on Bayesian reasoning provided in an embodiment of the present application is shown; Figure 6 A structural schematic diagram of an operating parameter adjustment device in a flotation process based on Bayesian reasoning provided in an embodiment of the present application is shown.
[0018] Icons: 600 - device for adjusting operating parameters in flotation process based on Bayesian reasoning, 601 - first acquisition module, 602 - first calculation module, 603 - second calculation module, 604 - construction module, 605 - training module, 606 - second acquisition module. DETAILED DESCRIPTION
[0019] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments.
[0020] The components of the embodiments of the present application generally described and shown in the drawings herein may be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.
[0021] Hereinafter, the terms "including", "having" and their cognates, which may be used in various embodiments of the present application, are intended only to indicate specific features, numbers, steps, operations, elements, components, or combinations of the foregoing items, and should not be understood as first excluding the existence of one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing items or adding the possibility of one or more features, numbers, steps, operations, elements, components, or combinations of the foregoing items.
[0022] Furthermore, the terms “first”, “second”, “third”, etc. are merely used for distinguishing descriptions and are not to be understood as indicating or implying relative importance.
[0023] Unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meanings as those generally understood by those skilled in the art to which the various embodiments of the present application belong. The terms (such as those defined in generally used dictionaries) will be interpreted as having the same meanings as the contextual meanings in the relevant technical field and will not be interpreted as having idealized meanings or overly formal meanings unless clearly defined in the various embodiments of the present application.
[0024] Example 1 The embodiment of the present application provides a method for adjusting operating parameters in a flotation process based on Bayesian reasoning.
[0025] See also Figure 1 The method for adjusting operating parameters in the flotation process based on Bayesian reasoning includes steps S101-S106: S101: Screening historical cases that meet the concentrate quality standard, and obtaining problem attributes and solution attributes of each of the historical cases.
[0026] In this embodiment, historical cases that meet the concentrate grade are stored in the historical case database. The initial historical case database consists of multiple historical cases, which can be express, is a two-tuple, where Represents the initial problem attributes, which are composed of the operating conditions and the parameters of the operating variables before adjustment. It represents the adjusted operating variable parameters, i.e. solution attributes, operating conditions such as ore grade, foam image characteristics, and operating variable parameters such as slurry level and aeration volume.
[0027] It should be noted that by extracting the characteristics of the working conditions to obtain latent variables, the historical case database is expressed as , the initial problem attributes of the historical case are transformed into problem attributes, expressed as , , represents the latent variable of the extracted operating condition variable, is the number of historical cases, is the number of all latent variables extracted.
[0028] See also Figure 2 In one embodiment, the step of obtaining the problem attributes and solution attributes of each of the historical cases includes steps 1011-S1013: S1011: Calculate latent variables of the historical case according to the operating condition data of the initial problem attributes in the historical case.
[0029] In this embodiment, the initial question attributes of the historical case are , is a given operating condition, represents the manipulated variable parameter before adjustment, is the number of historical cases, is the number of all working conditions, including ore grade, foam image characteristics (X-axis velocity, Y-axis velocity, large bubble pixels, medium bubble pixels, small bubble pixels, green, red-blue ratio, yellow), according to the formula Calculate the latent variables of the problem attributes of each historical case, where is the weight matrix obtained through training of each historical case, is the bias vector, is the scaling factor, are the normalized operating condition data in the initial problem attributes of each historical case.
[0030] S1012: Obtain the operating parameters of the historical case before adjustment, and set the latent variables of the historical case and the operating parameters before adjustment as the problem attributes of the historical case.
[0031] In this embodiment, the operating parameters before adjustment of the historical cases are obtained by querying the historical data, and the operating parameters before adjustment are converted into Attributes of the questions that make up the historical case .
[0032] S1013: Acquire the adjusted operating parameters of the historical case, and set the adjusted operating parameters as solution attributes.
[0033] In this embodiment, by querying historical data, the operating parameters of historical cases before adjustment are , that is, the solution attribute, the case library The problem attributes and solution attributes of the case are expressed as .
[0034] See also Figure 3 In one embodiment, before S102, the method further includes steps S301-S304: S301: Standardizing the initial problem attributes of the historical case, wherein the standardization includes standardizing the working condition data.
[0035] In this example, the historical cases that meet the concentrate grade The initial problem attributes in Settlement Properties To standardize, and , is the number of historical cases, is the operating variable of all solution attributes, according to the formula , Initial problem attributes for each historical case Settlement Properties Standardize to get and , Represents the initial problem attributes The mean vector of Represents the initial problem attributes The standard deviation vector of Represents solution attributes The mean vector of Represents solution attributes The standard deviation vector of the initial problem attribute is and the standard deviation vector to save.
[0036] S302: Extract features of the normalized operating condition data through a sparse autoencoder, and obtain a weight matrix and a bias vector.
[0037] In this embodiment, the ore grade data and foam image feature data after standardization are input into the sparse autoencoder to extract features. The standardization eliminates the dimension differences and numerical range differences between the data under different working conditions. By constructing a sparse autoencoder network, the sparse autoencoder is trained using the back propagation algorithm and gradient descent. After the training, the weight matrix , the weight matrix is used to connect the weights of the input layer and the hidden layer of the sparse autoencoder network, as well as the weights connecting the hidden layer and the output layer. The bias vector , that is, the bias of the input layer, hidden layer, and output layer.
[0038] S303: Calculate the latent variables of the current query case according to the weight matrix, the bias vector and the operating condition data corresponding to the current query case.
[0039] In this embodiment, the initial question attribute of the current query case is expressed as ,in is the initial question attribute of the current query case, The operation parameters before adjustment for the current query case. Based on the mean vector of the initial problem attributes of each saved historical case and the standard deviation vector , standardize the initial question attributes of the current query case, that is, , according to the formula Calculate the latent variables for the current query case, where is the weight matrix obtained through training of each historical case, is the bias vector, is the scaling factor, is the working condition data in the standardized initial problem attributes, is the latent variable of the problem attribute of the current query case.
[0040] S304: Obtaining the problem attribute of the current query case according to the latent variables and the operating parameters before adjustment.
[0041] In this embodiment, is the latent variable of the problem attribute of the current query case, To adjust the operating parameters before, the problem attributes of the current query case can be expressed as .
[0042] S102: Calculate the similarity scores between the problem attributes of each of the historical cases and the problem attributes of the current query case, and obtain the most similar historical case and a preset number of neighboring similar historical cases according to the sorting results of the similarity scores.
[0043] In this embodiment, the historical case database is traversed to obtain the problem attributes of each historical case and calculate the problem attributes of the current query case. Problem attributes related to each historical case The similarity scores between them are then sorted in descending order to obtain the top L nearest historically similar historical cases.
[0044] In one embodiment, the step S102 includes: calculating the similarity score according to formula (1), the question attributes of the historical case and the question attributes of the current query case; formula (1): ;in, is the similarity score between the historical case and the current query case, is the problem attribute of the current query case, is the problem attribute of the historical case.
[0045] In this implementation, for each historical case that meets the concentrate grade, after extracting the characteristics of the working conditions of each historical case to obtain latent variables, the historical case database consists of Convert to , which is the initial problem attribute consisting of the operating conditions and the operating parameters before adjustment Transformed into a problem attribute consisting of latent variables and operational parameters before adjustment . Loop through each historical case in the historical case database to get the problem attributes , the problem attribute of the current query case Calculate according to formula (1), that is, obtain the Euclidean distance between two vectors , in order to obtain the similarity score between the historical case and the current query case.
[0046] S103: Calculate the difference between the problem attribute of the current query case and the solution attribute of the most similar historical case to obtain a rough operation adjustment solution.
[0047] In this embodiment, the similarity scores are sorted to obtain the most similar historical cases. , according to the problem attributes of the current query case The operating parameters before adjustment in Adjusted operating parameters to the most similar historical case Get the rough operation adjustment solution .
[0048] S104: constructing an incremental database according to the problem attributes and solution attributes of the most similar historical case and the problem attributes and solution attributes of the neighboring similar historical cases.
[0049] In this embodiment, the similarity scores of the problem attributes of the current query case and the problem attributes of each historical case are sorted to obtain the top L nearest similar historical cases and the most similar historical case, and the incremental data of the problem attributes are obtained based on the difference between the problem attributes of the nearest historical cases and the problem attributes of the most similar historical cases, and the incremental data of the solution attributes are obtained based on the difference between the solution attributes of the nearest historical cases and the solution attributes of the most similar historical cases, thereby obtaining an incremental database.
[0050] See also Figure 4 In one embodiment, the S104 includes steps S1041-S1043: S1041: Calculate the difference between the problem attribute of each neighboring similar historical case and the problem attribute of the most similar historical case to obtain problem attribute increment data.
[0051] In this embodiment, the historical case most similar to the current query case is represented as , each neighbor similar historical case is expressed as , where the difference between the problem attribute of each neighbor historical case and the problem attribute of the most similar historical case, and the incremental data of the problem attribute is expressed as . S1042: Calculate the difference between the solution attribute of each neighboring similar historical case and the solution attribute of the most similar historical case to obtain solution attribute increment data.
[0052] In this embodiment, the difference between the solution attribute of each neighboring historical case and the solution attribute of the most similar historical case is used to obtain the solution attribute increment data, which is expressed as .
[0053] S1043: Construct an incremental database based on the incremental data of problem attributes and the corresponding incremental data of solution attributes of each neighboring similar historical case.
[0054] In this embodiment, based on the similarity score, L nearest historical cases similar to the current query case are taken to construct a new incremental database. , where in the incremental database, both the incremental data of problem attributes and the incremental data of solution attributes include L values, Incremental data representing the problem attributes, Incremental data representing solution properties.
[0055] S105: Train a Bayesian network model according to the incremental database, and input the incremental data of the problem attributes of the current query case into the trained Bayesian network model to obtain a detailed operation adjustment solution.
[0056] In this embodiment, the structure is modeled by process knowledge, the adjusted operating variables are regarded as the result, the given working conditions and the operating variables before adjustment are regarded as the cause, and the model parameters are trained online by the maximum likelihood estimation algorithm using the incremental database constructed in step S104 to obtain the trained Bayesian network model. As evidence, the trained Bayesian network model is input, and the fine operation adjustment solution is obtained through model inference calculation. The mean is expressed as ; See also Figure 5 In one embodiment, the S105 includes steps S1051-S1053: S1051: Determine the Bayesian network model structure based on flotation experience knowledge.
[0057] In this embodiment, based on the expert knowledge and experience in the field of flotation, the structure of the Bayesian network is defined to ensure that the network structure can reflect the causal relationship between the operating conditions (such as the grade of the raw ore, the characteristics of the foam image: X-axis velocity, Y-axis velocity, large bubble pixels, medium bubble pixels, small bubble pixels, green, red-to-blue ratio, yellow, etc.) and the operating parameter variables before adjustment and the operating parameter variables after adjustment (such as slurry level, aeration volume). Usually, the adjusted operating variables are regarded as the result, while the raw ore grade and the operating conditions are regarded as the cause.
[0058] In one embodiment, the step S1051 includes: obtaining operating variables related to concentrate quality; determining nodes and edges in the Bayesian network model structure based on the operating variables; assigning a node type to each node in the Bayesian network model structure; and constructing the Bayesian network model structure using graphical tools or programming methods based on the nodes, the node types and the edges.
[0059] In this embodiment, the adjustment of operating variables such as the grade of the raw ore, foam image characteristics (X-axis flow velocity, Y-axis flow velocity, large bubble pixels, medium bubble pixels, small bubble pixels, green, red-to-blue ratio, yellow, etc.), and the operating variables before adjustment, such as the initial slurry level and the initial aeration volume, will be related to the quality of the concentrate obtained by flotation. Exemplarily, each operating variable corresponds to a node in the Bayesian network. For example, the ore grade, X-axis flow velocity, slurry level, etc. will all become independent nodes in the Bayesian network structure. Furthermore, if variable A is the cause of variable B, it is represented in the network by an edge pointing from A to B.
[0060] For example, in the Bayesian network structure, nodes can be discrete variables or continuous variables. According to the characteristics of the flotation process, each node is reasonably assigned a type, such as certain categorical data (such as foam size classification, including foam size oversize class, foam size normal class and foam size undersize class) or continuous variables, such as ore grade, slurry level, aeration volume, etc., which are usually expressed as continuous values within the real number range.
[0061] For example, a graphical tool (such as Netica) or a programming method (such as Python's pgmpy library) is used to construct a Bayesian network model structure. According to the defined nodes, node types, and edges, a network diagram is drawn in a graphical tool to ensure that the network structure can clearly reflect the causal relationship between the variables and facilitate subsequent model training and reasoning calculations.
[0062] S1052: Input the problem attribute incremental data and the corresponding solution attribute incremental data of each of the neighboring similar historical cases in the incremental database into the Bayesian network model structure.
[0063] In this embodiment, the incremental database is used to input the constructed Bayesian network model, whose input is the incremental data of the problem attributes, and whose output is the incremental data of the corresponding solution attributes.
[0064] S1053: Perform training through a maximum likelihood estimation algorithm to obtain a trained Bayesian network model.
[0065] In this embodiment, the maximum likelihood estimation algorithm is used to perform inference training on these data from incremental problem attribute data to incremental solution attribute data, so as to adjust the parameters in the Bayesian network model and finally obtain a trained Bayesian network model.
[0066] S106: Obtain a target operation adjustment solution according to the coarse operation adjustment solution and the fine operation adjustment solution.
[0067] In this embodiment, based on the rough operation adjustment solution obtained in S103 Compared with the S105 , add them together , and obtain the final target operation adjustment solution .
[0068] The method for adjusting operating parameters in the flotation process based on Bayesian reasoning provided in this embodiment screens historical cases that meet the concentrate quality standards, and obtains the problem attributes and solution attributes of each of the historical cases; calculates the similarity score between the problem attributes of each of the historical cases and the problem attributes of the current query case, and obtains the most similar historical case and a preset number of neighboring similar historical cases according to the sorting result of the similarity score; calculates the difference between the problem attribute of the current query case and the solution attribute of the most similar historical case to obtain a rough operation adjustment solution; constructs an incremental database based on the problem attributes and solution attributes of the most similar historical case and the problem attributes and solution attributes of the neighboring similar historical cases; trains a Bayesian network model based on the incremental database, and inputs the incremental data of the problem attributes of the current query case into the trained Bayesian network model to obtain a fine operation adjustment solution; obtains a target operation adjustment solution based on the rough operation adjustment solution and the fine operation adjustment solution. The present application obtains a rough operation adjustment solution through the most similar case, obtains a fine operation adjustment solution through the incremental database training Bayesian network model, and finally obtains a target operation adjustment solution, which ensures accurate and timely adjustment of operating parameters during flotation.
[0069] Example 2 In addition, an embodiment of the present application provides a device for adjusting operating parameters in a flotation process based on Bayesian reasoning, which is applied to electronic equipment.
[0070] like Figure 6 As shown, the operating parameter adjustment device 600 in the flotation process based on Bayesian reasoning includes: The first acquisition module 601 is used to screen historical cases that meet the concentrate quality standard and obtain the problem attribute and solution attribute of each historical case; A first calculation module 602 is used to calculate the similarity score between the problem attribute of each of the historical cases and the problem attribute of the current query case, and obtain the most similar historical case and a preset number of neighboring similar historical cases according to the sorting result of the similarity score; The second calculation module 603 is used to calculate the difference between the problem attribute of the current query case and the solution attribute of the most similar historical case to obtain a rough operation adjustment solution; A construction module 604 is used to construct an incremental database according to the problem attributes and solution attributes of the most similar historical case and the problem attributes and solution attributes of the neighboring similar historical cases; A training module 605 is used to train a Bayesian network model according to the incremental database, and input the incremental data of the problem attribute of the current query case into the trained Bayesian network model to obtain a detailed operation adjustment solution; The second acquisition module 606 is configured to acquire a target operation adjustment solution according to the coarse operation adjustment solution and the fine operation adjustment solution.
[0071] Optionally, the first acquisition module 601 shown is also used to calculate the latent variables of the historical case based on the operating condition data of the initial problem attributes in the historical case; obtain the operating parameters of the historical case before adjustment, and set the latent variables of the historical case and the operating parameters before adjustment as the problem attributes of the historical case; obtain the operating parameters of the historical case after adjustment, and set the adjusted operating parameters as solution attributes.
[0072] Optionally, the first calculation module 602 shown is also used to standardize the initial problem attributes of the historical case, and the standardization includes standardization of operating condition data; extracting the standardized operating condition data through a sparse autoencoder, and obtaining a weight matrix and a bias vector; calculating the latent variables of the current query case based on the weight matrix, the bias vector and the operating condition data corresponding to the current query case; and obtaining the problem attributes of the current query case based on the latent variables and the operating parameters before adjustment.
[0073] Optionally, the first calculation module 602 is further configured to calculate the similarity score according to formula (1), the question attributes of the historical case and the question attributes of the current query case; formula (1): ;in, is the similarity score between the historical case and the current query case, is the problem attribute of the current query case, is the problem attribute of the historical case.
[0074] Optionally, the construction module 604 shown is also used to calculate the difference between the problem attributes of each neighboring similar historical case and the problem attributes of the most similar historical case to obtain incremental data of problem attributes; calculate the difference between the solution attributes of each neighboring similar historical case and the solution attributes of the most similar historical case to obtain incremental data of solution attributes; and construct an incremental database based on the incremental data of problem attributes of each neighboring similar historical case and the corresponding incremental data of solution attributes.
[0075] Optionally, the training module 605 is also used to determine the Bayesian network model structure based on flotation experience knowledge; input the problem attribute incremental data and the corresponding solution attribute incremental data of each of the neighboring similar historical cases in the incremental database into the Bayesian network model structure; and perform training through the maximum likelihood estimation algorithm to obtain a trained Bayesian network model.
[0076] Optionally, the training module 605 is also used to obtain operating variables related to concentrate quality; based on the operating variables, determine the nodes and edges in the Bayesian network model structure; assign a node type to each node in the Bayesian network model structure; and construct the Bayesian network model structure using graphical tools or programmatic methods based on the nodes, the node types and the edges.
[0077] The device 600 for adjusting operating parameters in a flotation process based on Bayesian reasoning provided in this embodiment can implement the method for adjusting operating parameters in a flotation process based on Bayesian reasoning provided in Embodiment 1, which will not be described again to avoid repetition.
[0078] The operating parameter adjustment device in the flotation process based on Bayesian reasoning provided in this embodiment screens each historical case that meets the concentrate quality standard, and obtains the problem attribute and solution attribute of each historical case; calculates the similarity score between the problem attribute of each historical case and the problem attribute of the current query case, and obtains the most similar historical case and a preset number of neighboring similar historical cases according to the sorting result of the similarity score; calculates the difference between the problem attribute of the current query case and the solution attribute of the most similar historical case to obtain a rough operation adjustment solution; constructs an incremental database according to the problem attribute and solution attribute of the most similar historical case and the problem attribute and solution attribute of the neighboring similar historical case; trains a Bayesian network model according to the incremental database, and inputs the incremental data of the problem attribute of the current query case into the trained Bayesian network model to obtain a fine operation adjustment solution; obtains a target operation adjustment solution according to the rough operation adjustment solution and the fine operation adjustment solution. The present application obtains a rough operation adjustment solution through the most similar case, obtains a fine operation adjustment solution through the incremental database training Bayesian network model, and finally obtains a target operation adjustment solution, which ensures accurate and timely adjustment of operating parameters during flotation.
[0079] Example 3 In addition, an embodiment of the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program runs on the processor, the method for adjusting operating parameters in a flotation process based on Bayesian reasoning provided in Example 1 is executed.
[0080] The electronic device provided in the embodiment of the present invention can execute the steps of the method for adjusting operating parameters in the flotation process based on Bayesian reasoning provided in the above method embodiment 1, which will not be described again to avoid repetition.
[0081] Example 4 The present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for adjusting operating parameters in a flotation process based on Bayesian reasoning provided in Example 1 is implemented.
[0082] In this embodiment, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
[0083] The computer-readable storage medium provided in this embodiment can implement the method for adjusting operating parameters in the flotation process based on Bayesian reasoning provided in Example 1, and will not be described again here to avoid repetition.
[0084] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or terminal including the element.
[0085] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.
[0086] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.
Claims
1. A method for adjusting operating parameters in a flotation process based on Bayesian reasoning, characterized in that: The method comprises: Screening historical cases that meet the concentrate quality standard, and obtaining problem attributes and solution attributes of each of the historical cases; Calculate the similarity scores between the problem attributes of each of the historical cases and the problem attributes of the current query case, and obtain the most similar historical case and a preset number of neighboring similar historical cases according to the sorting results of the similarity scores; Calculating the difference between the problem attribute of the current query case and the solution attribute of the most similar historical case to obtain a rough operation adjustment solution; Building an incremental database based on the problem attributes and solution attributes of the most similar historical case and the problem attributes and solution attributes of the neighboring similar historical cases; Training a Bayesian network model based on the incremental database, and inputting the incremental data of the problem attributes of the current query case into the trained Bayesian network model to obtain a detailed operation adjustment solution; A target operation adjustment solution is obtained according to the coarse operation adjustment solution and the fine operation adjustment solution.
2. The method according to claim 1, characterized in that: The step of obtaining the problem attributes and solution attributes of the historical case includes: Calculating latent variables of the historical case according to the working condition data of the initial problem attributes in the historical case; Obtaining the operating parameters of the historical case before adjustment, and setting the latent variables of the historical case and the operating parameters before adjustment as the problem attributes of the historical case; The adjusted operating parameters of the historical case are obtained, and the adjusted operating parameters are set as solution attributes.
3. The method according to claim 2, characterized in that Before calculating the similarity scores between the problem attributes of each of the historical cases and the problem attributes of the current query case, the method further includes: Performing standardization processing on the initial problem attributes of the historical case, wherein the standardization processing includes standardization processing of working condition data; The standardized working condition data is extracted through the sparse autoencoder, and the weight matrix and bias vector are obtained; Calculating the latent variables of the current query case according to the weight matrix, the bias vector and the operating condition data corresponding to the current query case; The problem attribute of the current query case is obtained according to the latent variables and the operating parameters before adjustment.
4. The method according to claim 1, characterized in that: The calculating of the similarity scores between the problem attributes of each historical case and the problem attributes of the current query case includes: Calculate the similarity score according to formula (1), the question attributes of the historical case and the question attributes of the current query case; Formula (1): ; in, is the similarity score between the historical case and the current query case, is the problem attribute of the current query case, is the problem attribute of the historical case.
5. The method according to claim 1, characterized in that The step of constructing an incremental database according to the problem attributes and solution attributes of the most similar historical case and the problem attributes and solution attributes of the neighboring similar historical cases includes: Calculate the difference between the problem attribute of each neighboring similar historical case and the problem attribute of the most similar historical case to obtain incremental data of the problem attribute; Calculate the difference between the solution attribute of each neighboring similar historical case and the solution attribute of the most similar historical case to obtain solution attribute increment data; An incremental database is constructed based on the problem attribute incremental data and the corresponding solution attribute incremental data of each neighboring similar historical case.
6. The method according to claim 1, characterized in that The step of training the Bayesian network model according to the incremental database comprises: Determine the Bayesian network model structure based on flotation experience knowledge; Inputting the problem attribute incremental data and the corresponding solution attribute incremental data of each of the neighboring similar historical cases in the incremental database into the Bayesian network model structure; The training is performed through the maximum likelihood estimation algorithm to obtain a trained Bayesian network model.
7. The method according to claim 6, characterized in that Determining the Bayesian network model structure based on flotation experience knowledge includes: Obtain operational variables related to concentrate quality; Based on the operational variables, determining nodes and edges in the Bayesian network model structure; Assigning a node type to each node in the Bayesian network model structure; The Bayesian network model structure is constructed according to the nodes, the node types and the edges using a graphical tool or a programming method.
8. A device for adjusting operating parameters in a flotation process based on Bayesian reasoning, characterized in that: The device comprises: The first acquisition module is used to screen historical cases that meet the concentrate quality standard and obtain the problem attributes and solution attributes of each historical case; A first calculation module is used to calculate the similarity score between the problem attribute of each of the historical cases and the problem attribute of the current query case, and obtain the most similar historical case and a preset number of neighboring similar historical cases according to the sorting result of the similarity score; A second calculation module is used to calculate the difference between the problem attribute of the current query case and the solution attribute of the most similar historical case to obtain a rough operation adjustment solution; A construction module, configured to construct an incremental database according to the problem attributes and solution attributes of the most similar historical case and the problem attributes and solution attributes of the neighboring similar historical cases; A training module, used for training a Bayesian network model according to the incremental database, and inputting the incremental data of the problem attributes of the current query case into the trained Bayesian network model to obtain a detailed operation adjustment solution; The second acquisition module is used to acquire a target operation adjustment solution according to the coarse operation adjustment solution and the fine operation adjustment solution.
9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is run on the processor, the method for adjusting operating parameters in a flotation process based on Bayesian reasoning according to any one of claims 1 to 7 is executed.
10. A computer-readable storage medium, characterized in that: The computer program stores a computer program, which, when running on a processor, executes the method for adjusting operating parameters in a flotation process based on Bayesian reasoning as claimed in any one of claims 1 to 7.
