CN116720139A - Mine paste slurry yield stress prediction method and system - Google Patents
Mine paste slurry yield stress prediction method and system Download PDFInfo
- Publication number
- CN116720139A CN116720139A CN202310399722.8A CN202310399722A CN116720139A CN 116720139 A CN116720139 A CN 116720139A CN 202310399722 A CN202310399722 A CN 202310399722A CN 116720139 A CN116720139 A CN 116720139A
- Authority
- CN
- China
- Prior art keywords
- characteristic information
- information entropy
- root node
- data
- paste slurry
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000002002 slurry Substances 0.000 title claims abstract description 109
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000003066 decision tree Methods 0.000 claims abstract description 73
- 238000007637 random forest analysis Methods 0.000 claims abstract description 50
- 239000004576 sand Substances 0.000 claims abstract description 47
- 238000012549 training Methods 0.000 claims abstract description 30
- 238000012795 verification Methods 0.000 claims abstract description 18
- 238000013524 data verification Methods 0.000 claims abstract description 11
- 238000010276 construction Methods 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 11
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 238000013138 pruning Methods 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 230000009194 climbing Effects 0.000 claims description 4
- 238000013502 data validation Methods 0.000 claims 1
- 239000011343 solid material Substances 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 238000011835 investigation Methods 0.000 description 2
- 239000004571 lime Substances 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 239000004568 cement Substances 0.000 description 1
- 230000018044 dehydration Effects 0.000 description 1
- 238000006297 dehydration reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000010881 fly ash Substances 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005086 pumping Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application provides a method and a system for predicting yield stress of paste slurry of a mine, which relate to the technical field of mine filling, and the method comprises the following steps: the slump characteristic information entropy, concentration characteristic information entropy and gray-sand ratio characteristic information entropy of the target paste slurry are input into a sequence adjustment model for training, after first node characteristic information is obtained, a decision tree set is built, the decision tree set is combined, a random forest model is built, data verification is carried out on the target paste slurry data input into the random forest model, the random forest model is optimized after a verification result is obtained, the target paste slurry data input into the optimized random forest model is used for predicting the yield stress of the target paste slurry, the technical problem that the determination of the paste slurry data is inaccurate enough in the prior art, the prediction accuracy of the paste slurry yield stress is low is solved, the precise determination of the paste slurry data is realized, and the prediction accuracy of the paste slurry yield stress is further improved.
Description
Technical Field
The application relates to the technical field of mine filling, in particular to a method and a system for predicting yield stress of mine paste slurry.
Background
Paste filling refers to a method for making materials into paste slurry without critical flow rate and little/no dehydration, pumping the paste slurry to an underground working surface through a pipeline under the action of a high-density solid filling pump and dead weight, and timely filling goaf. The yield stress is taken as an important parameter of paste rheological property and is a mode for judging paste pipe conveying quality, and most of the current measuring methods adopt paddle rheometers for measurement, but the method has the problems that the measuring principles of different rheometers are the same but the measuring methods are different, so that the measuring results are different.
In the prior art, the data of the paste slurry are not accurately measured, so that the yield stress prediction accuracy of the paste slurry is low.
Disclosure of Invention
The application provides a method and a system for predicting yield stress of paste slurry in a mine, which are used for solving the technical problems that in the prior art, the determination of the paste slurry data is not accurate enough, so that the accuracy of predicting the yield stress of the paste slurry is low.
In view of the above problems, the application provides a method and a system for predicting yield stress of mine paste slurry.
In a first aspect, the application provides a method for predicting yield stress of paste slurry of a mine, the method comprising: acquiring slump characteristic information entropy and concentration characteristic information entropy of target paste slurry, and gray-sand ratio characteristic information entropy; inputting the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy into a sequence adjustment model for training, and obtaining first root node characteristic information; constructing a decision tree set based on the first root node characteristic information; combining the decision tree sets to establish a random forest model; inputting target paste slurry data into the random forest model for data verification to obtain a verification result; and optimizing the random forest model based on the verification result, inputting the data of the target paste slurry into the optimized random forest model, and predicting the yield stress of the target paste slurry.
In a second aspect, the application provides a mine paste slurry yield stress prediction system, the system comprising: the information entropy acquisition module is used for acquiring slump characteristic information entropy, concentration characteristic information entropy and gray-sand ratio characteristic information entropy of the target paste slurry; the training module is used for inputting the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy into a sequence adjustment model for training, and acquiring first root node characteristic information; the decision tree set construction module is used for constructing a decision tree set based on the first root node characteristic information; the model building module is used for combining the decision tree sets and building a random forest model; the data verification module is used for inputting the target paste slurry data into the random forest model for data verification, and obtaining a verification result; and the prediction module is used for optimizing the random forest model based on the verification result, inputting the target paste slurry data into the optimized random forest model, and predicting the yield stress of the target paste slurry.
One or more technical schemes provided by the application have at least the following technical effects or advantages:
the application provides a method and a system for predicting yield stress of paste slurry in a mine, relates to the technical field of mine filling, and solves the technical problems that in the prior art, the determination of paste slurry data is inaccurate, so that the prediction accuracy of the paste slurry yield stress is low, the accurate determination of the paste slurry data is realized, and the prediction accuracy of the paste slurry yield stress is improved.
Drawings
FIG. 1 is a schematic flow chart of a method for predicting yield stress of paste slurry of a mine;
fig. 2 is a schematic diagram of a mine paste slurry yield stress prediction system.
Reference numerals illustrate: the system comprises an information entropy acquisition module 1, a training module 2, a decision tree set construction module 3, a model establishment module 4, a data verification module 5 and a prediction module 6.
Description of the embodiments
The application provides a method and a system for predicting yield stress of paste slurry in a mine, which are used for solving the technical problems that in the prior art, the determination of paste slurry data is not accurate enough, so that the accuracy of predicting the yield stress of the paste slurry is low.
Example 1
As shown in fig. 1, the embodiment of the application provides a method for predicting yield stress of paste slurry of a mine, which comprises the following steps:
step S100: acquiring slump characteristic information entropy and concentration characteristic information entropy of target paste slurry, and gray-sand ratio characteristic information entropy;
specifically, the method for predicting the yield stress of the mine paste slurry provided by the embodiment of the application is applied to a system for predicting the yield stress of the mine paste slurry.
In order to predict the yield stress of the target paste slurry, the slump characteristic, the concentration characteristic and the gray-sand ratio characteristic of the target paste slurry are required to be extracted, so that the slump characteristic, the concentration characteristic and the gray-sand ratio characteristic are respectively subjected to information theory coding operation through an information entropy calculation formula in information theory coding, slump characteristic information entropy, concentration characteristic information entropy and gray-sand ratio characteristic information entropy are obtained, and the prediction of the yield stress of the target paste slurry is used as an important reference basis for later realization.
Step S200: inputting the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy into a sequence adjustment model for training, and obtaining first root node characteristic information;
specifically, firstly, based on a BP neural network, the obtained slump characteristic information entropy, concentration characteristic information entropy and gray-sand ratio characteristic information entropy are adopted as construction data of a sequence adjustment model, the sequence adjustment model is subjected to supervised training, verification and testing, so that the sequence adjustment model is constructed, then random selection is carried out on the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy respectively, the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy which are selected randomly are compared with the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy which are trained in the sequence adjustment model, the minimum entropy value is classified preferentially, then the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy are sequentially adjusted according to the sequence from small entropy value to large, and the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy after the sequence adjustment are recorded as first root node characteristic information, so that the yield stress of target slurry is predicted is realized.
Step S300: constructing a decision tree set based on the first root node characteristic information;
specifically, according to the sequence contained in the characteristic information of the first root node, the slump characteristic, the concentration characteristic and the gray-sand ratio characteristic of the target paste slurry are sequentially classified by a regression algorithm, so that the decision tree is built, and the regression algorithm is an algorithm for directly or indirectly calling a function or a method of the algorithm.
The first root node feature information has N feature information samples in total, N is 3, and N feature information samples are randomly selected and replaced. The selected N characteristic information samples are used for training a decision tree, when each characteristic information sample has M attributes, M is a positive integer larger than 1, P attributes are randomly selected from the M attributes when each node of the decision tree needs to be split, P is a positive integer smaller than M, when the condition of P < < M is met, an information gain strategy is adopted from the P attributes, 1 attribute is selected as a splitting attribute of the node, and therefore each node in the characteristic information of the first root node is split according to the method until the node cannot be split again, the construction of the decision tree is completed, and a basis for predicting and tamping yield stress of target paste is realized for the subsequent implementation.
Step S400: combining the decision tree sets to establish a random forest model;
specifically, decision tree combination is carried out on a decision tree set constructed based on the first root node characteristic information, so that establishment of a plurality of decision trees on a random forest is completed, a voting result exists in each decision tree, the category with the largest final voting result serves as a random forest model prediction result, random selection of data is contained in a random forest model, namely replaced samples are adopted from the first root node characteristic information data set, a first root node characteristic information sub-data set is constructed, and the data quantity of the first root node characteristic information sub-data set is identical to that of the first root node characteristic information data set. Elements of different first root node characteristic information sub-data sets may be repeated, as may elements within the same first root node characteristic information sub-data set. And constructing a sub-decision tree by utilizing the first root node characteristic information sub-data set, putting the data into each sub-decision tree, and outputting a result by each sub-decision tree. When new data need to be obtained through the random forest model, the output result of the random forest model can be obtained through voting on the judging result of the sub decision tree, and by way of example, the classification result of 2 sub trees is A class, the classification result of 1 sub tree is B class, the classification result of the random forest is A class, and the prediction of the yield stress of the target paste slurry is realized.
Step S500: inputting target paste slurry data into the random forest model for data verification to obtain a verification result;
specifically, each item of data contained in the target paste slurry is acquired, the target paste slurry data can contain solid material characteristic data of the paste slurry, solid material proportion control data of the paste slurry, concentration proportion control data of the paste slurry and the like, further, the target paste slurry data are input into a constructed random forest model, meanwhile, an empirical tree is used for calculating the error rate of the target paste slurry data, the obtained error rate is compared with a preset range, the preset range can be set to be 10% of the current error data, if the data error rate result is smaller than or equal to the preset range, the current random forest model output data is regarded as correct data, meanwhile, the constructed decision tree is put into use, and the yield stress of the target paste slurry is predicted in a later period.
Step S600: and optimizing the random forest model based on the verification result, inputting the data of the target paste slurry into the optimized random forest model, and predicting the yield stress of the target paste slurry.
Specifically, the target paste slurry data is input into the constructed random forest model, meanwhile, the error rate is calculated by using an empirical tree, the obtained error rate is compared with a preset range, the preset range can be set to be that the current error data accounts for 10% of the total data, if the error rate is larger than the preset range, decision trees in a decision tree set are trimmed based on the complexity of the current decision tree, further, the random forest model is optimized according to the constructed decision tree set, the yield stress of the target paste slurry is predicted by finally inputting the target paste slurry data into the optimized random forest model, accurate determination of the paste slurry data is realized, and further, the prediction accuracy of the paste slurry yield stress is improved.
Further, the step S100 of the present application further includes:
step S110: acquiring slump characteristics, concentration characteristics and gray-sand ratio characteristics of the target paste slurry;
step S120: performing information theory coding operation on the slump characteristics to obtain slump characteristic information entropy;
step S130: performing information theory coding operation on the concentration characteristics to obtain the concentration characteristic information entropy;
step S140: and carrying out information theory coding operation on the gray-sand ratio characteristics to obtain the gray-sand ratio characteristic information entropy.
Specifically, slump characteristics, concentration characteristics, and sand-lime ratio characteristics of the target paste slurry are extracted, wherein slump characteristics refer to characteristics of final deformation of the paste slurry caused by self gravity and stopping of the paste slurry caused by internal resistance, concentration characteristics refer to the percentage of the mass of the paste slurry solid to the total mass, concentration can greatly influence slump, shear stress, viscosity and the like of the paste slurry, and sand-lime ratio characteristics refer to characteristics of mass ratio of cement to tailing in the paste slurry preparation process.
And then carrying out information theory coding operation on slump characteristics, concentration characteristics and gray-sand ratio characteristics through an information entropy calculation formula in information theory coding, wherein the information entropy calculation formula is as follows:
wherein t represents a random variable, corresponding to which is a set of all possible outputs, defined as a set of symbols, the output of the random variable being represented by t,the larger the uncertainty of the variable, the larger the entropy, which represents the output probability function.
The specific calculation of the information entropy value is carried out, so that the corresponding slump characteristic information entropy, concentration characteristic information entropy and gray-sand ratio characteristic information entropy are obtained, and the technical effect of providing important basis for predicting the yield stress of the target paste slurry in the later period is achieved.
Further, step S200 of the present application includes:
step S210: based on a BP neural network, adopting the slump characteristic information entropy, the concentration characteristic information entropy and the gray sand ratio characteristic information entropy as construction data, and performing supervision training, verification and test on the sequence adjustment model to obtain the sequence adjustment model;
step S220: and selecting information entropy in the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy randomly, and inputting the information entropy into the sequence adjustment model to obtain the characteristic information of the first root node.
Specifically, based on a BP neural network, slump characteristic information entropy, concentration characteristic information entropy and gray-sand ratio characteristic information entropy are used as construction data, a sequence adjustment model is subjected to supervision training, verification and test, the sequence adjustment model is obtained through training of a training data set and a supervision data set, each group of training data in the training data set comprises the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy, and the supervision data set is one-to-one supervision data corresponding to the training data set. And inputting each group of training data in the training data set into a sequence adjustment model, outputting and supervising the sequence adjustment model through supervising data corresponding to the group of training data, finishing the current group of training when the output result of the sequence adjustment model is consistent with the supervising data, finishing the training of all the training data in the training data set, and finishing the training of the sequence adjustment model.
In order to ensure the accuracy of the sequence adjustment model, the test processing of the sequence adjustment model may be performed by the test data set, for example, the test accuracy may be set to 85%, and when the test accuracy of the test data set satisfies 85%, the sequence adjustment model is constructed.
Further, the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy which are randomly selected are compared with the entropy values of the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy which are trained in the sequence adjustment model, the minimum entropy values are classified preferentially, then the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy are sequentially adjusted according to the order of the entropy values from small to large, and the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy after the sequence adjustment are recorded as first root node characteristic information, so that the yield stress of the target paste is better predicted in the later period.
Further, the step S300 of the present application further includes:
step S310: randomly selecting feature information of a root node to be selected from the feature information of the first root node, wherein the feature information of the root node to be selected is used as the feature information of the first root node and is used as the feature information of a temporary optimal root node;
step S320: randomly selecting feature information of a root node to be selected from the feature information of the first root node to serve as feature information of a second root node, wherein the feature information of the first root node and the feature information of the second root node are different feature information;
step S330: comparing the first root node characteristic information with the second root node characteristic information based on a climbing algorithm, and iterating to obtain optimal root node characteristic information;
step S340: constructing the optimal root node characteristic information as a leaf node of the decision tree;
step S350: the leaf node is added to the decision tree.
Specifically, the construction of the decision tree can reach the optimum in the later stage, therefore, the first root node characteristic information of the decision tree needs to be optimized, firstly, the first root node characteristic information needs to be randomly selected as the first root node characteristic information, meanwhile, the first root node characteristic information is further, the first root node characteristic information is randomly selected as the second root node characteristic information again, namely, the first root node characteristic information and the second root node characteristic information are different characteristic information, the iteration is carried out until the ith root node characteristic information is obtained, i is a positive integer larger than 3, meanwhile, the second root node characteristic information is optimally compared on the basis of a hill climbing algorithm, the second root node characteristic information is compared with the first root node characteristic information, and the third root node characteristic information, if the second root node characteristic information is the optimal root node characteristic information, namely, the second root node characteristic information is better than the first root node characteristic information, the leaf characteristic information is better than the optimal than the leaf characteristic information, the leaf characteristic information is better than the optimal when the leaf characteristic information is fully constructed in the current root node, and the leaf characteristic information is better than the optimal in the current root node, and the optimal leaf characteristic is better than the optimal node is constructed in the current root node.
Further, the step S500 of the present application further includes:
step S510: acquiring target paste data;
step S520: inputting the target paste slurry data into a random forest model, calculating error rate by using an empirical tree, and obtaining data error rate;
step S530: judging whether the data error rate result is smaller than or equal to a preset range;
step S540: and if the data error rate result is smaller than or equal to the preset range, putting the decision tree into use.
Specifically, each item of data contained in the target paste slurry is acquired, the target paste slurry data may include solid material characteristic data of the paste slurry, solid material proportion control data of the paste slurry, concentration proportion control data of the paste slurry, and the like, the solid material characteristic data of the paste slurry refers to data such as density, granularity, specific surface area, and the like of solid materials of the paste slurry, the solid material proportion control data of the paste slurry refers to proportion coefficients of gangue, fly ash, water and cementing materials, and the concentration proportion control data of the paste slurry refers to important data affecting fluidity and slump of the paste slurry.
Further, the target paste slurry data is input into the constructed random forest model, meanwhile, the error rate is calculated by using an empirical tree, the empirical tree is used for completing acquisition of the data error rate according to the error ratio of the historical data in a test algorithm, so that the data error rate is compared with a preset range, the preset range can be set to be 10% of the total data of the current error data, if the data error rate result is smaller than or equal to the preset range, the output data of the current random forest model can be regarded as correct data, and meanwhile, the constructed decision tree is put into use, so that the technical effect of predicting the yield stress of the target paste slurry is achieved.
Further, the step S600 of the present application further includes:
step S610: judging whether the data error rate result is larger than the preset range;
step S620: if the error rate is greater than the preset range, pruning the decision tree set based on the decision tree complexity to obtain a simplified decision tree set;
step S630: and optimizing the random forest model based on the simplified decision tree set.
Specifically, the data error rate and the preset range are compared, the preset range can be set to be that the current error data accounts for 10% of the total data, if the data error rate result is larger than the preset range, then on the basis of the complexity of the constructed decision tree, the decision tree which does not reach the investigation standard is pruned after the decision tree is concentrated, calculation is carried out upwards from the bottom layer of the decision tree, namely, non-leaf nodes in the decision tree are investigated from the bottom to the top, if the target paste slurry data is obtained, the decision tree can give out output meeting the preset range, the evaluation standard is obtained, if the subtree corresponding to the node is replaced by the leaf node, the leaf node is replaced by the leaf node, and the error is reduced due to the combination of the leaf node of the current decision tree, so that the pruning operation is completed, the decision tree after the pruning and the decision tree which reaches the investigation standard are integrated, the decision tree set is obtained, and finally on the basis of the obtained simplified decision tree set, the decision tree in the purchased random forest model is correspondingly optimized, the forest is based on the simplified tree set, the optimal yield stress is predicted to reach the target paste.
Example 2
Based on the same inventive concept as the method for predicting yield stress of mine paste slurry in the foregoing embodiment, as shown in fig. 2, the present application provides a system for predicting yield stress of mine paste slurry, which comprises:
the information entropy acquisition module 1 is used for acquiring slump characteristic information entropy, concentration characteristic information entropy and gray-sand ratio characteristic information entropy of the target paste slurry;
the training module 2 is used for inputting the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy into a sequence adjustment model for training, and acquiring first root node characteristic information;
the decision tree set constructing module 3 is used for constructing a decision tree set based on the first root node characteristic information;
the model building module 4 is used for combining the decision tree sets and building a random forest model;
the data verification module 5 is used for inputting the target paste slurry data into the random forest model for data verification, and obtaining a verification result;
and the prediction module 6 is used for optimizing the random forest model based on the verification result, inputting the target paste data into the optimized random forest model, and predicting the yield stress of the target paste.
Further, the system further comprises:
the characteristic acquisition module is used for acquiring slump characteristics, concentration characteristics and gray-sand ratio characteristics of the target paste slurry;
the first information theory coding operation module is used for carrying out information theory coding operation on the slump characteristics to obtain slump characteristic information entropy;
the second information theory coding operation module is used for carrying out information theory coding operation on the concentration characteristics to obtain the concentration characteristic information entropy;
and the third information theory coding operation module is used for carrying out information theory coding operation on the gray-sand ratio characteristics to obtain the gray-sand ratio characteristic information entropy.
Further, the system further comprises:
the model obtaining module is used for performing supervision training, verification and test on the sequence adjustment model based on a BP neural network by adopting the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy as construction data to obtain the sequence adjustment model;
the characteristic information obtaining module is used for randomly selecting information entropy in the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy respectively, inputting the information entropy into the sequence adjustment model and obtaining the characteristic information of the first root node.
Further, the system further comprises:
the first random selection module is used for randomly selecting feature information of a root node to be selected from the feature information of the first root node, and is used as the feature information of the first root node and is used as the feature information of a temporary optimal root node;
the second random selection module is used for randomly selecting feature information of a root node to be selected from the first root node feature information again to serve as second root node feature information, wherein the first root node feature information and the second root node feature information are different feature information;
the iteration module is used for comparing the first root node characteristic information with the second root node characteristic information based on a climbing algorithm so as to iterate the first root node characteristic information and obtain optimal root node characteristic information;
the leaf node construction module is used for constructing the optimal root node characteristic information as a leaf node of the decision tree;
and the adding module is used for adding the leaf nodes into the decision tree.
Further, the system further comprises:
the data acquisition module is used for acquiring target paste material data;
the error rate calculation module is used for inputting the target paste slurry data into a random forest model, calculating the error rate by using an empirical tree, and obtaining the data error rate;
the first judging module is used for judging whether the data error rate result is smaller than or equal to a preset range;
and the decision tree use module is used for putting the decision tree into use if the data error rate result is smaller than or equal to the preset range.
Further, the system further comprises:
the second judging module is used for judging whether the data error rate result is larger than the preset range or not;
the pruning module is used for pruning the decision tree set based on the decision tree complexity to obtain a simplified decision tree set if the error rate is larger than the preset range;
and the optimization module is used for optimizing the random forest model based on the simplified decision tree set.
From the foregoing detailed description of a method for predicting yield stress of paste slurry for mines, those skilled in the art can clearly understand that a system for predicting yield stress of paste slurry for mines in this embodiment is relatively simple for the device disclosed in the embodiments, and the relevant points refer to the description of the method.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (7)
1. The method for predicting the yield stress of the paste slurry of the mine is characterized by comprising the following steps of:
acquiring slump characteristic information entropy and concentration characteristic information entropy of target paste slurry, and gray-sand ratio characteristic information entropy;
inputting the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy into a sequence adjustment model for training, and obtaining first root node characteristic information;
constructing a decision tree set based on the first root node characteristic information;
combining the decision tree sets to establish a random forest model;
inputting target paste slurry data into the random forest model for data verification to obtain a verification result;
and optimizing the random forest model based on the verification result, inputting the data of the target paste slurry into the optimized random forest model, and predicting the yield stress of the target paste slurry.
2. The method of claim 1, wherein obtaining the slump characteristic information entropy, the concentration characteristic information entropy, the gray-to-sand ratio characteristic information entropy, comprises:
acquiring slump characteristics, concentration characteristics and gray-sand ratio characteristics of the target paste slurry;
performing information theory coding operation on the slump characteristics to obtain slump characteristic information entropy;
performing information theory coding operation on the concentration characteristics to obtain the concentration characteristic information entropy;
and carrying out information theory coding operation on the gray-sand ratio characteristics to obtain the gray-sand ratio characteristic information entropy.
3. The method of claim 1, wherein obtaining the first root node characteristic information comprises:
based on a BP neural network, adopting the slump characteristic information entropy, the concentration characteristic information entropy and the gray sand ratio characteristic information entropy as construction data, and performing supervision training, verification and test on the sequence adjustment model to obtain the sequence adjustment model;
and selecting information entropy in the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy randomly, and inputting the information entropy into the sequence adjustment model to obtain the characteristic information of the first root node.
4. The method of claim 1, wherein the constructing a decision tree comprises:
randomly selecting feature information of a root node to be selected from the feature information of the first root node, wherein the feature information of the root node to be selected is used as the feature information of the first root node and is used as the feature information of a temporary optimal root node;
randomly selecting feature information of a root node to be selected from the feature information of the first root node to serve as feature information of a second root node, wherein the feature information of the first root node and the feature information of the second root node are different feature information;
comparing the first root node characteristic information with the second root node characteristic information based on a climbing algorithm, and iterating to obtain optimal root node characteristic information;
constructing the optimal root node characteristic information as a leaf node of the decision tree;
the leaf node is added to the decision tree.
5. The method of claim 1, wherein inputting target paste slurry data into the random forest model for data validation comprises:
acquiring target paste data;
inputting the target paste slurry data into a random forest model, calculating error rate by using an empirical tree, and obtaining data error rate;
judging whether the data error rate result is smaller than or equal to a preset range;
and if the data error rate result is smaller than or equal to the preset range, putting the decision tree into use.
6. The method of claim 5, wherein optimizing the random forest model comprises:
judging whether the data error rate result is larger than the preset range;
if the error rate is greater than the preset range, pruning the decision tree set based on the decision tree complexity to obtain a simplified decision tree set;
and optimizing the random forest model based on the simplified decision tree set.
7. A mine paste slurry yield stress prediction system, comprising:
the information entropy acquisition module is used for acquiring slump characteristic information entropy, concentration characteristic information entropy and gray-sand ratio characteristic information entropy of the target paste slurry;
the training module is used for inputting the slump characteristic information entropy, the concentration characteristic information entropy and the gray-sand ratio characteristic information entropy into a sequence adjustment model for training, and acquiring first root node characteristic information;
the decision tree set construction module is used for constructing a decision tree set based on the first root node characteristic information;
the model building module is used for combining the decision tree sets and building a random forest model;
the data verification module is used for inputting the target paste slurry data into the random forest model for data verification, and obtaining a verification result;
and the prediction module is used for optimizing the random forest model based on the verification result, inputting the target paste slurry data into the optimized random forest model, and predicting the yield stress of the target paste slurry.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310399722.8A CN116720139A (en) | 2023-04-14 | 2023-04-14 | Mine paste slurry yield stress prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310399722.8A CN116720139A (en) | 2023-04-14 | 2023-04-14 | Mine paste slurry yield stress prediction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116720139A true CN116720139A (en) | 2023-09-08 |
Family
ID=87864868
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310399722.8A Pending CN116720139A (en) | 2023-04-14 | 2023-04-14 | Mine paste slurry yield stress prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116720139A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117637080A (en) * | 2024-01-26 | 2024-03-01 | 昆明理工大学 | Yield stress prediction method based on filling slurry differential analysis |
-
2023
- 2023-04-14 CN CN202310399722.8A patent/CN116720139A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117637080A (en) * | 2024-01-26 | 2024-03-01 | 昆明理工大学 | Yield stress prediction method based on filling slurry differential analysis |
| CN117637080B (en) * | 2024-01-26 | 2024-04-09 | 昆明理工大学 | Yield stress prediction method based on filling slurry differential analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110084367B (en) | Soil moisture content prediction method based on LSTM deep learning model | |
| CN116720139A (en) | Mine paste slurry yield stress prediction method and system | |
| CN110456026B (en) | Soil moisture content monitoring method and device | |
| CN115061219B (en) | Fracture type reservoir prediction identification method and system based on petroleum and natural gas detection | |
| CN114611778B (en) | Reservoir water level early warning method and system based on warehousing flow | |
| CN111639803A (en) | Prediction method applied to future vegetation index of area under climate change scene | |
| CN113326660A (en) | Tunnel surrounding rock extrusion deformation prediction method based on GA-XGboost model | |
| CN112307536B (en) | Dam seepage parameter inversion method | |
| CN115049124A (en) | Deep and long tunnel water inrush prediction method based on Bayesian network | |
| CN116415481A (en) | Regional landslide hazard risk prediction method and device, computer equipment and storage medium | |
| CN111079783B (en) | Method for identifying stratum lithology parameters based on multi-core ensemble learning | |
| CN115759445A (en) | Machine learning and cloud model-based classified flood random forecasting method | |
| CN114692981A (en) | Medium-and-long-term runoff forecasting method and system based on Seq2Seq model | |
| CN111144636A (en) | Slope deformation prediction method | |
| CN116960962A (en) | Mid-long term area load prediction method for cross-area data fusion | |
| US11927717B2 (en) | Optimized methodology for automatic history matching of a petroleum reservoir model with Ensemble Kalman Filter (EnKF) | |
| CN113378276B (en) | Composite foundation intelligent design method based on genetic algorithm and gene expression programming | |
| CN115563874A (en) | Tunnel health degree analysis method and system based on big data | |
| CN115456260A (en) | Customer service telephone traffic prediction method | |
| CN114575802A (en) | High water cut oil reservoir oil well yield prediction method based on machine learning | |
| CN117575106B (en) | Method, system, electronic equipment and medium for predicting gas production profile of coal-bed gas well | |
| CN117455321B (en) | Intelligent assessment method and system for performance of fluidized solidified soil | |
| Feder | Machine-learning approach determines spatial variation in shale decline curves | |
| Liu et al. | Continuous reservoir simulation model updating and forecasting using a markov chain monte carlo method | |
| Pasupunuri et al. | Use of Machine Learning Algorithms for Predicting the Transverse Cracking in Jointed Plain Concrete Pavements |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |