CN114186478B - Blasting block degree prediction method based on RBF neural network - Google Patents
Blasting block degree prediction method based on RBF neural network Download PDFInfo
- Publication number
- CN114186478B CN114186478B CN202111349702.7A CN202111349702A CN114186478B CN 114186478 B CN114186478 B CN 114186478B CN 202111349702 A CN202111349702 A CN 202111349702A CN 114186478 B CN114186478 B CN 114186478B
- Authority
- CN
- China
- Prior art keywords
- neural network
- rock
- rbf neural
- rbf
- data
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- G06N3/045—Combinations of 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
Abstract
The invention discloses a blasting blockiness prediction method based on a RBF neural network, which comprises the following steps: the method comprises the steps that firstly, a training data set comprising blasting design parameters, rock coefficients and average block sizes is constructed on the basis of historical blasting data, wherein the rock coefficients are obtained through calculation on the basis of multiple groups of data reflecting rock properties; step two, constructing an RBF neural network, training the RBF neural network by taking blasting design parameters and rock coefficients of a training data set as input layers and average block degrees as output layers, and determining a hidden layer of the RBF neural network and a weight coefficient between the hidden layer and the output layers to obtain the trained RBF neural network; and step three, inputting blasting design parameters and rock coefficients at the position to be blasted to the trained RBF neural network, and outputting to obtain the average block degree. The method has the advantages that the RBF neural network is matched with the distribution characteristics of blasting data, the input data dimension is small, and the prediction accuracy is high.
Description
Technical Field
The invention relates to the technical field of engineering blasting. More specifically, the invention relates to a blasting blockiness prediction method based on an RBF neural network.
Background
Blasting is a very important step in ore mining, and the blasting effect is directly related to the mining efficiency of ore and the later shovel loading and transportation, and is an important factor influencing the mining cost of mines. Factors influencing the blasting block size of the rock are many, such as row spacing, hole pitch, step height, hole diameter and the like, and have a very complex nonlinear relation with the blasting block size, and the traditional Kutzlam model is difficult to accurately predict the blasting block size.
The neural network is a machine learning method based on data, has strong nonlinear mapping capability, and can approximate any nonlinear continuous function by a three-layer neural network. The neural network technology is applied to blasting block degree prediction, the accuracy of prediction is greatly improved, the large block rate can be effectively reduced, and the blasting cost is further reduced. However, in a general neural network model or a multilayer perceptron, the structure of the network, i.e., the number of layers and the number of neurons in each layer, is difficult to determine, and the optimal structure in an experiment can be selected only by an experimental method. If the data set changes, the optimal network structure changes and the manually determined structure is easily over-fitted.
Disclosure of Invention
It is an object of the present invention to address at least the above problems and to provide at least the advantages described hereinafter.
The invention also aims to provide a blasting blockiness prediction method based on the RBF neural network, wherein the RBF neural network is adopted to select data with the largest influence on output as hidden layer neurons by an orthogonal least square method, and the network structure can be determined in a self-adaptive manner according to a real data set.
To achieve these objects and other advantages and in accordance with the purpose of the invention, there is provided a method for predicting a blasting block size based on an RBF neural network, comprising the steps of:
the method comprises the steps that firstly, a training data set comprising blasting design parameters, rock coefficients and average block sizes is constructed on the basis of historical blasting data, wherein the rock coefficients are obtained through calculation on the basis of multiple groups of data reflecting rock properties;
step two, constructing an RBF neural network, taking blasting design parameters and rock coefficients of a training data set as an input layer of the RBF neural network, taking average block degree of the training data set as an output layer, training the RBF neural network to determine a hidden layer structure of the RBF neural network and a weight coefficient between the hidden layer and the output layer, and obtaining the trained RBF neural network;
and step three, inputting blasting design parameters and rock coefficients at the position to be blasted to the RBF neural network trained in the step two, and outputting the obtained average block degree, namely the predicted average block degree.
Preferably, the blast design parameters include: row spacing, pitch, step height, aperture, plugging length, single-hole charge, and explosive coefficient.
Preferably, the rock coefficient is calculated using equation 1:
a =0.06 × (RMD + SDI + HF) formula 1
Wherein RMD represents values corresponding to rock mass description and rock joint, SDI represents values corresponding to density influence, and HF represents hardness coefficient;
when the rock mass is powdery or friable rock, RMD =10;
when the rock mass is massive rock, RMD =50;
when the rock mass is a vertical jointed rock, RMD = JF = (JCF × JPS) + JPA;
JCF represents a value corresponding to the joint surface condition, JCF =1 when the joint surface condition is tight, JCF =1.5 when the joint surface condition is loose, and JCF =2 when the joint surface condition is a mud layer;
JPS represents a value corresponding to the pitch of the joint surface, and when the pitch of the joint surface is less than 0.1m, JPS =10;
when the pitch of the joint surfaces is between 0.1 and 0.3m, JPS =20;
when the pitch between joint planes is 0.3 to P, JPS =80, P = (pitch x pitch) 0.5 ；
When the pitch of the joint surface is more than or equal to P, JPS =50;
JPA represents a value corresponding to the joint face angle, and when the joint face angle is horizontal, JPA =10;
when the joint face angles are in the same direction, JPA =20;
JPA =30 when the joint face angle is vertical;
JPA =40 when the joint face angle is inverse;
SDI＝25×RD-50，
wherein RD is rock density;
when the Young modulus E is less than or equal to 50,
when Young's modulus E>At the time of 50 f, the temperature of the alloy is higher,
wherein UCS represents uniaxial compressive strength.
Preferably, before training the RBF neural network, the training data set is normalized by equation 2:
the data' is data after standardization, the data is original data of a training data set, the avr is an average value of parameter data of the same type in the training data set, and the std is a standard deviation of the parameter data of the same type.
Preferably, a gaussian function is used as a radial basis function, and an orthogonal least square method is used for selecting neurons for determining the RBF neural network so as to determine the hidden layer of the RBF neural network, and the specific method comprises the following steps:
(1) Taking all training set data as RBF neurons, and acquiring regression vectors processed by the RBF neurons;
(2) Calculating the influence of each regression vector on the training error, and selecting the regression vector v with the largest influence 1 Corresponding original data is used as a first RBF neuron;
(3) Respectively returning the residual regression vectors v 2 ～v k Orthogonalizing the selected regression vector, and selecting original data corresponding to the regression vector with the largest influence on reducing the training error from the orthogonalized regression vectors as a new RBF neuron;
(4) Repeating the step (3) k-1 times to obtain k RBF neurons, wherein the k RBF neurons form a hidden layer of the current RBF neural network.
Preferably, an LM algorithm is used to determine the weight coefficient between the hidden layer and the output layer, and the specific method is:
(1) Initializing parameters between a hidden layer and an output layer of the RBF neural network by adopting random numbers, and calculating an output value of the average block degree of the RBF neural network under the current parameters;
(2) Obtaining the gradient of the output value of the average block degree to each weight coefficient by adopting a BP algorithm, and further calculating a Jacobian matrix J;
(3) Calculating the error gradient of the current parameter:
g＝J'*residual
wherein J' is the transposition of a Jacobian matrix, residual is the real value of the average block degree in the training data set minus the output value of the average block degree under the current neural network in the step (1), and g is the speed of error reduction, namely the error gradient;
(4) Calculating step length:
Δ＝(J*J'+μI) -1 *g
wherein, Δ is the step length of each iteration parameter update, I is an identity matrix, and μ is an artificially set coefficient;
(5) Updating a weight parameter between an RBF neural network hidden layer and an output layer:
weight′＝weight-Δ
wherein, weight' is a weight parameter after current iteration, and weight is a weight parameter before current iteration;
(6) And (5) repeating the steps (2) to (5) until a preset iteration number is reached.
Provided is an electronic device including: at least one processor, and a memory communicatively coupled to the at least one processor, wherein the memory stores instructions executable by the at least one processor to cause the at least one processor to perform any of the methods described above.
There is provided a storage medium having stored thereon a computer program which, when executed by a processor, implements the method of any of the above.
The invention at least comprises the following beneficial effects: the RBF neural network selects the data with the largest influence on the output as hidden layer neurons by an orthogonal least square method, and the network structure can be determined in a self-adaptive manner according to a real data set. Moreover, the RBF neural network has excellent local approximation performance and is very suitable for the distribution characteristics of blasting data.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention.
Drawings
FIG. 1 is a flow chart of RBF neural network training according to one embodiment of the present invention;
FIG. 2 is a histogram of error distribution according to one embodiment of the present invention;
FIG. 3 is a predicted value-true value scatter plot of one of the aspects of the present invention;
FIG. 4 is a predicted value-true value line graph of one embodiment of the present invention.
Detailed Description
The present invention is further described in detail below with reference to the attached drawings so that those skilled in the art can implement the invention by referring to the description text.
It is to be noted that the experimental methods described in the following embodiments are all conventional methods unless otherwise specified, and the reagents and materials, if not otherwise specified, are commercially available; in the description of the present invention, the terms indicating orientation or positional relationship are based on the orientation or positional relationship shown in the drawings only for the convenience of description and simplification of description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus, should not be construed as limiting the present invention.
As shown in fig. 1 to 4, the present invention provides a method for predicting blasting block size based on RBF neural network, comprising the following steps:
the method comprises the steps that firstly, a training data set comprising blasting design parameters, rock coefficients and average block sizes is constructed on the basis of historical blasting data, wherein the rock coefficients are obtained through calculation on the basis of multiple groups of data reflecting rock properties;
step two, constructing an RBF neural network, taking blasting design parameters and rock coefficients of a training data set as an input layer of the RBF neural network, taking average block degree of the training data set as an output layer, training the RBF neural network to determine a hidden layer structure of the RBF neural network and a weight coefficient between the hidden layer and the output layer, and obtaining the trained RBF neural network;
and step three, inputting blasting design parameters and rock coefficients at the position to be blasted to the RBF neural network trained in the step two, and outputting the obtained average block degree, namely the predicted average block degree.
In the technical scheme, the RBF neural network is analyzed to be composed of an input layer, a hidden layer and an input layer, the RBF neural network has excellent local approximation performance, the RBF neural network is very suitable for the distribution characteristics of blasting data, but a lot of factors influencing the block degree after blasting exist, if the RBF neural network is not screened, the RBF neural network is directly used, parameters are too many, the accuracy of prediction can be improved, more data are needed for training, more computing resources are needed to be consumed, and if only a small number of parameters are selected as input, the accuracy of prediction cannot meet the requirement at all. According to the technical scheme, multiple groups of data reflecting rock properties are combined into one rock coefficient after being calculated, so that the convergence rate and the prediction accuracy of the RBF neural network can be remarkably improved.
In another aspect, the blast design parameters include: row spacing, pitch, step height, aperture, plugging length, single-hole charge, and explosive coefficient. By combining with the calculated rock coefficient, the blasting design parameters only need to select a small quantity, namely row spacing, hole spacing, step height, hole diameter, blocking length, single-hole explosive loading and explosive coefficient, so that the blasting block degree can be accurately predicted.
In another solution, the rock coefficient is calculated by formula 1:
a =0.06 × (RMD + SDI + HF) formula 1
The method comprises the following steps of A, obtaining a rock mass description and rock joint data, wherein RMD represents a value corresponding to rock mass description and rock joint, SDI represents a value corresponding to density influence, and HF represents a hardness coefficient;
when the rock mass is a powdery or friable rock, RMD =10;
when the rock mass is massive rock, RMD =50;
when the rock mass is vertical jointed rock, RMD = JF = (JCF × JPS) + JPA;
JCF represents a value corresponding to the joint surface condition, JCF =1 when the joint surface condition is tight, JCF =1.5 when the joint surface condition is loose, and JCF =2 when the joint surface condition is a mud layer;
JPS represents a value corresponding to the pitch of the joint surfaces, and when the pitch of the joint surfaces is less than 0.1m, JPS =10;
when the pitch of the joint surfaces is between 0.1 and 0.3m, JPS =20;
when the pitch between joint planes is 0.3 to P, JPS =80, P = (pitch x pitch) 0.5 ；
When the pitch of the joint surface is more than or equal to P, JPS =50;
JPA represents a value corresponding to the joint face angle, and when the joint face angle is horizontal, JPA =10;
JPA =20 when the joint face angles are in the same direction;
JPA =30 when the joint face angle is vertical;
JPA =40 when the joint face angle is inverse;
SDI＝25×RD-50，
wherein RD is rock density;
when the Young modulus E is less than or equal to 50,
when Young's modulus E>At the time of 50 f, the temperature of the alloy is higher,
wherein UCS represents uniaxial compressive strength.
In the technical scheme, the rock coefficient is used as the input of the RBF neural network, so that the prediction accuracy is improved. When the same blasting design parameters are used for different mines, the output variability can be very large, and the important reason is that the rock mass properties of the ore are different. But if the dimension of the input is increased, the complexity of the neural network is increased, and more data is needed to train the neural network. The rock coefficient is used as input, the influence of rock joint, density, uniaxial compressive strength and the like on blasting is comprehensively considered, and the accuracy of blasting prediction can be improved under the condition that the input dimension is not increased.
In another technical solution, before training the RBF neural network, a training data set is normalized by using formula 2:
the data' is data after standardization, the data is original data of a training data set, the avr is an average value of parameter data of the same type in the training data set, and the std is a standard deviation of the parameter data of the same type.
Because the size difference of each dimension data of the blasting design parameters may be relatively large, if the original data is directly used for training, the difficulty of network convergence is increased. The above technical problem can be solved by standardizing the data.
In another technical scheme, a Gaussian function is used as a radial basis function, neurons for determining the RBF neural network are selected by adopting an orthogonal least square method to determine a hidden layer of the RBF neural network, and the specific method comprises the following steps:
(1) Taking all training set data as RBF neurons, and acquiring regression vectors processed by the RBF neurons;
(2) Calculating the influence of each regression vector on the training error, and selecting the regression vector v with the largest influence 1 Corresponding original data is used as a first RBF neuron;
(3) Is divided intoRespectively dividing the residual regression vector v 2 ～v k Orthogonalizing the selected regression vector, and selecting original data corresponding to the regression vector with the largest influence on reducing the training error from the orthogonalized regression vectors as a new RBF neuron;
(4) Repeating the step (3) k-1 times to obtain k RBF neurons, wherein the k RBF neurons form a hidden layer of the current RBF neural network.
In the technical scheme, the RBF neural network selects the data with the largest influence on the output as hidden layer neurons by an orthogonal least square method, and the network structure can be determined in a self-adaptive manner according to a real data set.
In another technical scheme, an LM algorithm is adopted to determine a weight coefficient between a hidden layer and an output layer, and the specific method comprises the following steps:
(1) Initializing parameters between a hidden layer and an output layer of the RBF neural network by adopting random numbers, and calculating an output value of the average block degree of the RBF neural network under the current parameters;
(2) Obtaining the gradient of the output value of the average block degree to each weight coefficient by adopting a BP algorithm, and further calculating a Jacobian matrix J;
(3) Calculating the error gradient of the current parameter:
g＝J'*residual
wherein J' is the transposition of a Jacobian matrix, residual is the real value of the average block degree in the training data set minus the output value of the average block degree under the current neural network in the step (1), and g is the speed of error reduction, namely the error gradient;
(4) Calculating step length:
Δ＝(J*J'+μI) -1 *g
wherein, Δ is the step length of each iteration parameter update, I is an identity matrix, and μ is an artificially set coefficient;
(5) Updating a weight parameter between an RBF neural network hidden layer and an output layer:
weight′＝weight-Δ
wherein, weight' is a weight parameter after current iteration, and weight is a weight parameter before current iteration;
(6) And (5) repeating the steps (2) to (5) until a preset iteration number is reached.
In the technical scheme, compared with the traditional BP algorithm, the LM algorithm utilizes not only the first derivative of the error to the neural network parameters, but also the second derivative thereof, and has higher convergence rate. More importantly, the LM algorithm can adaptively adjust the learning rate in the iterative process without spending excessive energy on parameter adjustment.
Provided is an electronic device including: at least one processor, and a memory communicatively coupled to the at least one processor, wherein the memory stores instructions executable by the at least one processor to cause the at least one processor to perform any of the methods described above.
There is provided a storage medium having stored thereon a computer program which, when executed by a processor, implements the method of any of the above.
Wherein OLS represents an orthogonal least squares method, X50 represents an average block size, m represents a number corresponding to a regression vector having the largest influence on an error, and e i Expressing the influence of the ith regression vector on the error, wherein the Jacobian matrix expresses a Jacobian matrix, the number expresses the number of the test data sets within the corresponding error rate range, the error rate expresses the error rate, the actual expresses the true value of the average block degree, the ANN expresses the predicted value of the average block degree, and the error rate is calculated by the following formula:
wherein the prediction represents an output value of the RBF neural network.
In fig. 3, the abscissa represents the predicted value of the average block size, and the ordinate represents the true value of the average block size. The closer the trend of the points is to 45 degrees, the better the representation effect is, and fig. 3 shows that the prediction method of the present application has extremely high accuracy.
In fig. 4, the abscissa indicates the reference numeral of the data, i.e., the several test data; the ordinate represents the average block degree, and it can be seen from fig. 4 that the predicted value is very close to the true value, which indicates that the prediction method of the present application has very high accuracy.
While embodiments of the invention have been described above, it is not limited to the applications set forth in the description and the embodiments, which are fully applicable in various fields of endeavor to which the invention pertains, and further modifications may readily be made by those skilled in the art, it being understood that the invention is not limited to the details shown and described herein without departing from the general concept defined by the appended claims and their equivalents.
Claims (3)
1. The blasting block degree prediction method based on the RBF neural network is characterized by comprising the following steps of:
the method comprises the steps that firstly, a training data set comprising blasting design parameters, rock coefficients and average block sizes is constructed on the basis of historical blasting data, wherein the rock coefficients are obtained through calculation on the basis of multiple groups of data reflecting rock properties;
step two, constructing an RBF neural network, taking blasting design parameters and rock coefficients of a training data set as an input layer of the RBF neural network, taking average block degree of the training data set as an output layer, training the RBF neural network to determine a hidden layer structure of the RBF neural network and a weight coefficient between the hidden layer and the output layer, and obtaining the trained RBF neural network;
inputting blasting design parameters and rock coefficients at the position to be blasted to the RBF neural network trained in the step two, and outputting the obtained average block degree which is the predicted average block degree;
blasting design parameters include: row spacing, pitch, step height, aperture, plugging length, single-hole charge and explosive coefficient;
the rock coefficient is calculated using equation 1:
a =0.06 × (RMD + SDI + HF) formula 1
The method comprises the following steps of A, obtaining a rock mass description and rock joint data, wherein RMD represents a value corresponding to rock mass description and rock joint, SDI represents a value corresponding to density influence, and HF represents a hardness coefficient;
when the rock mass is powdery or friable rock, RMD =10;
when the rock mass is massive rock, RMD =50;
when the rock mass is vertical jointed rock, RMD = JF = (JCF × JPS) + JPA;
JCF represents a value corresponding to the joint surface condition, JCF =1 when the joint surface condition is tight, JCF =1.5 when the joint surface condition is loose, and JCF =2 when the joint surface condition is a mud layer;
JPS represents a value corresponding to the pitch of the joint surface, and when the pitch of the joint surface is less than 0.1m, JPS =10;
when the pitch of the joint surfaces is between 0.1 and 0.3m, JPS =20;
when the pitch between joint planes is 0.3 to P, JPS =80, P = (pitch x pitch) 0.5 ；
When the pitch of the joint surface is more than or equal to P, JPS =50;
JPA represents a value corresponding to the joint face angle, and when the joint face angle is horizontal, JPA =10;
JPA =20 when the joint face angles are in the same direction;
JPA =30 when the joint face angle is vertical;
JPA =40 when the joint face angle is inverse;
SDI＝25×RD-50，
wherein RD is rock density;
when the Young modulus E is less than or equal to 50,
when Young's modulus E>At the time of 50 f, the temperature of the alloy is higher,
wherein UCS represents uniaxial compressive strength;
before training the RBF neural network, adopting formula 2 to standardize and process a training data set:
the data' is data after standardization, the data is original data of a training data set, avr is an average value of parameter data of the same type in the training data set, and std is a standard deviation of the parameter data of the same type;
a Gaussian function is adopted as a radial basis function, neurons for determining the RBF neural network are selected by adopting an orthogonal least square method to determine the hidden layer of the RBF neural network, and the specific method comprises the following steps:
(1) Taking all training set data as RBF neurons, and acquiring regression vectors processed by the RBF neurons;
(2) Calculating the influence of each regression vector on the training error, and selecting the regression vector v with the largest influence 1 Corresponding original data is used as a first RBF neuron;
(3) Respectively dividing the residual regression vectors v 2 ～v k Orthogonalizing the selected regression vector, and selecting original data corresponding to the regression vector with the largest influence on reducing the training error from the orthogonalized regression vectors as a new RBF neuron;
(4) Repeating the step (3) k-1 times to obtain k RBF neurons, wherein the k RBF neurons form a hidden layer of the current RBF neural network;
the LM algorithm is adopted to determine the weight coefficient between the hidden layer and the output layer, and the specific method comprises the following steps:
(1) Initializing parameters between a hidden layer and an output layer of the RBF neural network by adopting random numbers, and calculating an output value of the average block degree of the RBF neural network under the current parameters;
(2) Obtaining the gradient of the output value of the average block degree to each weight coefficient by adopting a BP algorithm, and further calculating a Jacobian matrix J;
(3) Calculating the error gradient of the current parameter:
g＝J'*residual
wherein J' is the transposition of a Jacobian matrix, residual is the real value of the average block degree in the training data set minus the output value of the average block degree under the current neural network in the step (1), and g is the speed of error reduction, namely the error gradient;
(4) Calculating step length:
Δ＝(J*J'+μI) -1 *g
wherein, Δ is the step length of each iteration parameter update, I is an identity matrix, and μ is an artificially set coefficient;
(5) Updating a weight parameter between an RBF neural network hidden layer and an output layer:
weight′＝weight-Δ
wherein, weight' is a weight parameter after current iteration, and weight is a weight parameter before current iteration;
(6) And (6) repeating the steps (2) to (5) until a preset iteration number is reached.
2. An electronic device, comprising: at least one processor, and a memory communicatively coupled to the at least one processor, wherein the memory stores instructions executable by the at least one processor to cause the at least one processor to perform the method of claim 1.
3. Storage medium on which a computer program is stored which, when executed by a processor, carries out the method of claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111349702.7A CN114186478B (en) | 2021-11-15 | 2021-11-15 | Blasting block degree prediction method based on RBF neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111349702.7A CN114186478B (en) | 2021-11-15 | 2021-11-15 | Blasting block degree prediction method based on RBF neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114186478A CN114186478A (en) | 2022-03-15 |
| CN114186478B true CN114186478B (en) | 2023-01-17 |
Family
ID=80540099
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111349702.7A Active CN114186478B (en) | 2021-11-15 | 2021-11-15 | Blasting block degree prediction method based on RBF neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114186478B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107203659A (en) * | 2017-04-26 | 2017-09-26 | 江苏大学 | Subcooled flow boiling heat exchange quality transmission relaxation factor Forecasting Methodology based on radial base neural net |
| CN107528650A (en) * | 2017-08-30 | 2017-12-29 | 华南师范大学 | A kind of Forecasting Methodology of the cognitive radio networks frequency spectrum based on GCV RBF neurals |
| CN111783987A (en) * | 2020-07-14 | 2020-10-16 | 中国水利水电科学研究院 | Farmland reference crop evapotranspiration prediction method based on improved BP neural network |
| CN113340164A (en) * | 2021-06-21 | 2021-09-03 | 中铁十九局集团矿业投资有限公司 | Method for predicting rock blasting block size |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105550744A (en) * | 2015-12-06 | 2016-05-04 | 北京工业大学 | Nerve network clustering method based on iteration |
-
2021
- 2021-11-15 CN CN202111349702.7A patent/CN114186478B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107203659A (en) * | 2017-04-26 | 2017-09-26 | 江苏大学 | Subcooled flow boiling heat exchange quality transmission relaxation factor Forecasting Methodology based on radial base neural net |
| CN107528650A (en) * | 2017-08-30 | 2017-12-29 | 华南师范大学 | A kind of Forecasting Methodology of the cognitive radio networks frequency spectrum based on GCV RBF neurals |
| CN111783987A (en) * | 2020-07-14 | 2020-10-16 | 中国水利水电科学研究院 | Farmland reference crop evapotranspiration prediction method based on improved BP neural network |
| CN113340164A (en) * | 2021-06-21 | 2021-09-03 | 中铁十九局集团矿业投资有限公司 | Method for predicting rock blasting block size |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114186478A (en) | 2022-03-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Reimann et al. | Modeling macroscopic material behavior with machine learning algorithms trained by micromechanical simulations | |
| Zhang et al. | Using neural networks to represent von Mises plasticity with isotropic hardening | |
| US7243056B2 (en) | Strategy parameter adaptation in evolution strategies | |
| Biabangard-Oskouyi et al. | Application of imperialist competitive algorithm for materials property characterization from sharp indentation test | |
| CN109343505A (en) | Gear method for predicting residual useful life based on shot and long term memory network | |
| CN111753467B (en) | Mineral product result prediction method, device, computer equipment and storage medium | |
| CN111324993A (en) | Turbulent flow field updating method, device and related equipment | |
| Qu et al. | Deep learning predicts stress–strain relations of granular materials based on triaxial testing data | |
| CN111783338B (en) | Microstructure metal intensity distribution prediction method and device based on artificial intelligence | |
| Wei et al. | A damage particle method for smeared modeling of brittle fracture | |
| Resk et al. | Adaptive mesh refinement and automatic remeshing in crystal plasticity finite element simulations | |
| CN111880412A (en) | Reconfigurable robot zero and neural optimal control method based on single evaluation network | |
| Nieto-Valeiras et al. | On the effect of slip transfer at grain boundaries on the strength of FCC polycrystals | |
| CN114186478B (en) | Blasting block degree prediction method based on RBF neural network | |
| Liu et al. | Operator learning for predicting mechanical response of hierarchical composites with applications of inverse design | |
| JP6969713B1 (en) | Steel pipe crush strength prediction model generation method, steel pipe crush strength prediction method, steel pipe manufacturing characteristic determination method, and steel pipe manufacturing method | |
| Phalke et al. | Modeling size effects in microwire torsion: A comparison between a Lagrange multiplier-based and a CurlFp gradient crystal plasticity model | |
| Fehlemann et al. | Influence of synthetically generated inclusions on the stress accumulation and concentration in X65 pipeline steel | |
| CN116842853B (en) | Missile aerodynamic characteristic prediction model construction method for uncertainty quantization | |
| KR102120150B1 (en) | Learning method and learning device for variational interference using neural network and test method and test device for variational interference using the same | |
| Bugeda et al. | An integration of a low cost adaptive remeshing strategy in the solution of structural shape optimization problems using evolutionary methods | |
| Kazeruni et al. | Data-driven artificial neural network for elastic plastic stress and strain computation for notched bodies | |
| Liu et al. | An AI-powered approach to improving tunnel blast performance considering geological conditions | |
| Bernacki | Kinetic equations and level-set approach for simulating solid-state microstructure evolutions at the mesoscopic scale: state of the art, limitations, and prospects | |
| Zamora | Machine Learning for the Development of Damage Models in Engineering Applications |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |