CN116881676B - Prediction method for water inflow of open pit - Google Patents
Prediction method for water inflow of open pit Download PDFInfo
- Publication number
- CN116881676B CN116881676B CN202311154516.7A CN202311154516A CN116881676B CN 116881676 B CN116881676 B CN 116881676B CN 202311154516 A CN202311154516 A CN 202311154516A CN 116881676 B CN116881676 B CN 116881676B
- Authority
- CN
- China
- Prior art keywords
- water inflow
- value
- open pit
- vmd
- 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
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 95
- 238000000034 method Methods 0.000 title claims abstract description 71
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 45
- 238000012545 processing Methods 0.000 claims abstract description 27
- 238000005457 optimization Methods 0.000 claims abstract description 15
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims abstract description 12
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims abstract description 12
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 12
- 238000004458 analytical method Methods 0.000 claims abstract description 5
- 230000006870 function Effects 0.000 claims description 29
- 239000002245 particle Substances 0.000 claims description 21
- 238000004364 calculation method Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 8
- 230000001502 supplementing effect Effects 0.000 claims description 8
- 230000008859 change Effects 0.000 claims description 6
- 230000009471 action Effects 0.000 claims description 4
- 230000003044 adaptive effect Effects 0.000 claims description 4
- 238000010219 correlation analysis Methods 0.000 claims description 4
- 239000013589 supplement Substances 0.000 claims description 4
- 239000000047 product Substances 0.000 claims description 2
- 230000008569 process Effects 0.000 description 14
- 239000003673 groundwater Substances 0.000 description 8
- 238000011161 development Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 230000006378 damage Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 230000010355 oscillation Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 240000002262 Litsea cubeba Species 0.000 description 1
- 235000012854 Litsea cubeba Nutrition 0.000 description 1
- 239000005442 atmospheric precipitation Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000008595 infiltration Effects 0.000 description 1
- 238000001764 infiltration Methods 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000036314 physical performance Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a prediction method of water inflow of an open pit, and belongs to the technical field of data processing. According to the method, weather data of a historical time sequence and corresponding water inflow of the open pit are obtained, and an original data set is supplemented with a missing value; obtaining an optimal (k, alpha) value by improving parameters of a northern eagle algorithm (PSRNGO) optimization Variation Modal Decomposition (VMD), inputting the optimal (k, alpha) value into the VMD for decomposition, decomposing the VMD to obtain a plurality of IMF components, substituting the decomposed IMF components into an SSA-BILSTM model to obtain a water inflow data prediction result, and accumulating and summing the water inflow data prediction result to obtain a final prediction result of the water inflow of the open pit; by absolute error MAE, root mean square error RMSE, average relative error MAPE, R 2 And evaluating and carrying out error analysis on the final prediction result of the water inflow of the open pit. According to the method, PSRNGO is adopted to optimize VMD, so that the prediction accuracy of the water inflow of the open pit is greatly improved.
Description
Technical Field
The invention relates to a prediction method of water inflow of an open pit, and belongs to the technical field of data processing.
Background
The water inflow of the open pit refers to accumulated water formed in the open pit, and sources of the accumulated water include atmospheric precipitation, surface runoff, underground water inflow and the like. The open pit water has great influence and harm to the development of mineral exploitation work and the safety of workers, and is mainly reflected in that: the stability of the slope is damaged, the descending speed of mine engineering is reduced, and the efficiency and the service life of equipment are reduced.
In rainy season, the water inflow of the open pit tends to rise rapidly, which can cause damage to mining equipment, production interruption and even endanger life safety of staff, and infiltration of rainwater can reduce physical performance indexes such as internal friction force and cohesive force, so that stability of the side slope is reduced. Through accurate prediction water inflow, the mine enterprise can reasonably arrange production plans, makes flood control preparation work in advance, adopts corresponding flood control measures, avoids production interruption and equipment damage caused by water inflow, and improves production efficiency and economic benefits.
The prediction method for the water inflow of the open pit mostly adopts geological methods such as a large well method, a water balance method, a numerical method, a hydrogeological comparison method and the like. The large well method is used for observing the change of the water level by injecting or extracting a certain amount of water in the well so as to infer the hydraulic characteristic parameters of groundwater around the well, but the large well method needs to manually operate the well, and inaccurate test results can be caused in the operation process; certain requirements are imposed on the size and shape of the well, and if the size of the well is not satisfactory, deviations may occur in the test results; the overall situation of the groundwater system cannot be fully understood. The water balance method is based on the balance relation between water input and output, and the hydraulic characteristic of the groundwater system is deduced by observing the groundwater level and the flow, but the water balance method has higher requirements on the water input and output of the groundwater system, needs to accurately measure and estimate various water quantities, certain errors may exist in the data acquisition and estimation, the water balance method assumes that the groundwater system is stable, however, in actual situations, the water input and output of the groundwater system may be affected by factors such as seasons, climates and the like, so that the water balance relation is not established. The numerical method is a method for performing simulation and calculation by using a computer by establishing a mathematical model, but the method needs a large amount of input data and parameter estimation for establishing a complex mathematical model, has higher requirements on the accuracy of the input data and the parameter of the model, has a complex calculation process, needs a large amount of iterative calculation by using the computer, and has longer calculation time. Hydrogeologic simulation utilizes existing hydrogeologic data and empirical knowledge to apply hydrogeologic features in a geologic-like environment to a target region to infer groundwater hydraulic characteristics, but it relies on the geologic-like environment hydrogeologic data, which may be inaccurate if the target region differs significantly from the geologic environment of the existing data.
In general, these methods have certain applications in predicting water inflow studies, but all suffer from certain limitations and uncertainties.
Disclosure of Invention
Aiming at the problem of predicting the water inflow of the open pit, the invention provides a method for predicting the water inflow of the open pit, namely, an algorithm is adopted to match a neural network, time sequence research is carried out on the water inflow of the open pit, a research thought of optimizing VMD (virtual machine direction model) by adopting an improved northern eagle algorithm is adopted, and the change trend and the water inflow of the open pit can be predicted more accurately by analyzing and modeling time sequence data and combining with the neural network for prediction.
A prediction method for water inflow of an open pit comprises the following specific steps:
s1, acquiring weather data of a historical time sequence and corresponding water inflow of an open pit, and supplementing a missing value to an original data set;
s2, initializing population parameters of the improved northern eagle algorithm, setting a range of VMD processing parameters (k, alpha), wherein tau is set to 0, DC is set to 0, init is set to 1, and tol is set to 1 multiplied by 10 -7 Tau is a frequency penalty factor, DC is whether a direct current component is included, init is an initialization mode, and tol is an iteration tolerance; preferably, k has a value in the range of [2,10 ]]Alpha is within the range of [100,10000 ]]；
S3, adopting an improved hawk algorithm, and optimizing the VMD processing parameter (k, alpha) according to the minimum sample entropy of the open pit water inflow data to obtain an optimal combination value of the VMD processing parameter (k, alpha);
s4, performing VMD signal processing on the open pit water inflow historical data based on the VMD processing parameter (k, alpha) optimal combination value to obtain a plurality of IMF components;
s5, carrying out correlation analysis on the IMF modal function, judging the Pearson correlation coefficient of the adjacent modal function by taking 0.1 as a threshold value, and verifying the VMD decomposition quality;
s6, substituting the decomposed IMF components into an SSA-BILSTM model to obtain a water inflow data prediction result, and accumulating and summing the water inflow data prediction result to obtain a final prediction result of the water inflow of the open pit;
s7, adopting absolute error MAE, root mean square error RMSE and average relative error MAPE, R 2 And evaluating and carrying out error analysis on the final prediction result of the water inflow of the open pit.
The weather data in the step S1 comprise air temperature, air pressure, humidity and rainfall.
The method for supplementing the missing value in the step S1 is to supplement the average value of the data at the time points before and after, and the expression is as follows
The method comprises the steps of carrying out a first treatment on the surface of the In (1) the->Data representing an i-th influence factor of the d-th history time to be preprocessed, the influence factor affecting the water inflow of the pit; />(d=1, 2, …, N) data representing the i-th influence factor affecting pit water inflow at the d-1-th history; />(d=1, 2, …, N) data representing the ith influence factor affecting pit water inflow at the (d+1) th historical time; i is {1,2, …, M }, i.e., M influence factors that influence the water inflow of the open sky pit; n is the original time series data length.
Initializing the north hawk algorithm population parameters to be the ith individual position of the initialized population in the step S2:
X i =lb+r⊙(ub-lb)
wherein X is i To initialize the position of individuals in the ith dimension, lb is the lower boundary vector, ub is the upper boundary vector, r is a random matrix of size (population size x 1), and # is the matrix dot product operation.
The sample entropy calculation formula in the step S3 is as follows:
；
where m is the mode dimension, r is the similarity tolerance threshold,probability of matching m points for two sequences under a similar tolerance r, wherein N is data number; />The ratio of the approximate number to the total number is N-m-1;
adding 1 to the mode dimension to obtain m+1The sample entropy formula is transformed into:
；
in the method, in the process of the invention,for the entropy value of the sample,probability of matching m+1 points for two sequences;
the specific step of optimizing the VMD processing parameter (k, α) value in the step S3 includes:
s31, in a northern hawk algorithm prey identification stage (exploration stage), a particle swarm optimization strategy is adopted to change the exploration stage; the formula of the particle swarm optimization strategy is as follows
；
；
Wherein:a j-th dimensional velocity vector for particle i in the k-th iteration;a position vector of the particle i in the j-th dimension in the k-1 th iteration;,is interval [0,1 ]]A random number within;the historical optimal position of the particle i in the j-th dimension in the k-th iteration;historical optimal positions of the group in the j-th dimension in the k-th iteration; 0.7 is an inertial weight for adjusting the search range for the solution space; 2 is a learning factor;
s32, redefining an attenuation coefficient in a northern eagle pursuit action stage (development stage), and determining R in a mixed attenuation mode to obtain an improved northern eagle algorithm; wherein the mixed attenuation mode comprises exponential attenuation and adaptive attenuation; the attenuation coefficient is given by
；
Wherein t is the current iteration number; t is the maximum iteration number; r is an attenuation coefficient;storing an array of objective function values for each northern hawk game location;storing an objective function value array of each new north eagle position;
s33, optimizing the VMD processing parameter (k, alpha) by using an improved northern hawk algorithm to obtain an optimal parameter combination (k, alpha).
The formula of the adjacent mode function Pearson in the step S5 is as follows:
；
wherein:are respectively toStandard fraction of sample, sample average and sample standard deviation;，are respectively toStandard fraction of sample, sample average and sample standard deviation;for the correlation coefficient value, n is the number of samples. By r 2 Judging correlation coefficients of two adjacent modal functions;
the absolute error MAE formula in step S7 is:
MAE；
in which y i Is a true value of the code,representing the predicted value;
the root mean square error RMSE formula is:
；
in which y i Is a true value of the code,representing the predicted value;
the average relative error MAPE formula is:
；
wherein y is i Is a true value of the code,representing the predicted value;
R 2 the formula is:
；
in which y i Is a true value of the code,the predicted value is represented by a value of the prediction,representing the average of the true values.
The beneficial effects of the invention are as follows:
(1) According to the invention, when VMD decomposition is carried out, an improved northern eagle algorithm is introduced, and k value and alpha value in VMD are optimized, so that errors caused by determining k value and alpha value empirically in the past are changed;
(2) The invention improves the exploration stage in the northern hawk algorithm, adopts the particle swarm optimization variable, has strong global searching capability and high convergence speed compared with the original algorithm, and is easy to realize and adjust;
(3) The invention improves the R attenuation coefficient in the north hawk algorithm development stage, adopts a hybrid attenuation strategy, adopts exponential attenuation in the early stage and adopts self-adaptive attenuation in the later stage; the searching is smoother and more flexible, and sudden jump or oscillation in the searching process is avoided, so that the searching efficiency and the resolution quality are improved;
(4) Compared with the original prediction method, the method for predicting the water inflow of the open pit by deep learning is simpler and more convenient and has higher precision.
Drawings
FIG. 1 is a general flow chart for predicting water inflow of an open pit according to the present invention;
FIG. 2 is a flowchart for optimizing VMD by improving the litsea cubeba algorithm;
FIG. 3 is a graph of the results of the VMD decomposition of the components after optimization;
FIG. 4 is a graph of the correlation coefficient of the eigenmode components at a k value of 7;
FIG. 5 is a graph of the correlation coefficient of the eigenmode components at a k value of 8;
FIG. 6 is a graph of 10 iterations of PSRNGO and NGO;
FIG. 7 is a graph of PSRNGO and NGO iterated 100 times;
FIG. 8 is a graph of the predicted result of the improved northern eagle algorithm optimized VMD substitution model SSA-BILSTM;
FIG. 9 is a graph of model absolute error MAE prediction accuracy versus after optimized VMD and un-optimized VMD;
FIG. 10 is a graph comparing the root mean square error RMSE of the model after optimizing the VMD and the model after not optimizing the VMD;
FIG. 11 is a graph comparing the mean relative error MAPE of the models after optimizing the VMD and un-optimizing the VMD;
FIG. 12 is a post-optimized VMD and un-optimized VMD model R 2 Comparison graph.
Detailed Description
The invention will be described in further detail with reference to specific embodiments, but the scope of the invention is not limited to the description.
Summary of The Invention
A prediction method of water inflow of an open pit (see figures 1 and 2) comprises the following specific steps:
s1, acquiring weather data (the weather data comprises air temperature, air pressure, humidity and rainfall) of a historical time sequence and corresponding water inflow of an open pit, and supplementing a missing value to an original data set; the method for supplementing the missing value is to supplement the data by adopting the average value of the time points before and after, and the expression is as follows
；
In the method, in the process of the invention,data representing an i-th influence factor of the d-th history time to be preprocessed, the influence factor affecting the water inflow of the pit;(d=1, 2, …, N) data representing the i-th influence factor affecting pit water inflow at the d-1-th history;(d=1, 2, …, N) data representing the ith influence factor affecting pit water inflow at the (d+1) th historical time; i is {1,2, …, M }, i.e., M influence factors that influence the water inflow of the open sky pit; n is the original time sequence data length;
s2, initializing population parameters of the improved northern eagle algorithm, setting a range of VMD processing parameters (k, alpha), wherein tau is set to 0, DC is set to 0, init is set to 1, and tol is set to 1 multiplied by 10 -7 Tau is a frequency penalty factor, DC is whether a direct current component is included, init is an initialization mode, and tol is an iteration tolerance; preferably, k has a value in the range of [2,10 ]]Alpha is within the range of [100,10000 ]]；
The initializing and improving northern hawk algorithm population parameters are the ith individual position of the initializing population:
X i =lb+r⊙(ub-lb)
where Xi is the position of the initialized population individual in the ith dimension, lb is the lower boundary vector, ub is the upper boundary vector, r is a random matrix with the size (population scale x 1), and l is the matrix point multiplication operation;
s3, adopting an improved hawk algorithm, and optimizing the VMD processing parameter (k, alpha) according to the minimum sample entropy of the open pit water inflow data to obtain an optimal combination value of the VMD processing parameter (k, alpha);
the VMD processing parameter (k, alpha) value optimizing specific steps include:
s31, in a northern hawk algorithm prey identification stage (exploration stage), a particle swarm optimization strategy is adopted to change the exploration stage; the formula of the particle swarm optimization strategy is as follows
；
；
Wherein:a j-th dimensional velocity vector for particle i in the k-th iteration;a position vector of the particle i in the j-th dimension in the k-1 th iteration;,is interval [0,1 ]]Random numbers in the search module, so that the randomness of the search is increased;the historical optimal position of the particle i in the j-th dimension in the k-th iteration;historical optimal positions of the group in the j-th dimension in the k-th iteration; 0.7 is an inertial weight for adjusting the search range for the solution space; 2 is a learning factor;
s32, redefining an attenuation coefficient in a northern eagle pursuit action stage (development stage), and determining R in a mixed attenuation mode to obtain an improved northern eagle algorithm; wherein the mixed attenuation mode comprises exponential attenuation and adaptive attenuation; the attenuation coefficient is given by
；
Wherein t is the current iteration number; t is the maximum iteration number; r is an attenuation coefficient;storing an array of objective function values for each northern hawk game location;storing an objective function value array of each new north eagle position;
s33, optimizing the VMD processing parameter (k, alpha) value by utilizing an improved northern hawk algorithm to obtain an optimal parameter combination (k, alpha);
the sample entropy calculation formula is as follows:
；
where m is the mode dimension, r is the similarity tolerance threshold,probability of matching m points for two sequences under a similar tolerance r, wherein N is data number;the ratio of the approximate number to the total number is N-m-1;
adding 1 to the mode dimension to obtain m+1The sample entropy formula is transformed into:
；
in the method, in the process of the invention,for the entropy value of the sample,probability of matching m+1 points for two sequences;
s4, performing VMD signal processing on the open pit water inflow historical data based on the VMD processing parameter (k, alpha) optimal combination value to obtain a plurality of IMF components;
s5, carrying out correlation analysis on the IMF modal function, judging the Pearson correlation coefficient of the adjacent modal function by taking 0.1 as a threshold value, and verifying the VMD decomposition quality;
the formula of the adjacent mode function Pearson is:
；
wherein:are respectively toStandard fraction of sample, sample average and sample standard deviation;，are respectively toStandard fraction of sample, sample average and sample standard deviation;for the correlation coefficient value, n is the number of samples. By r 2 Judging correlation coefficients of two adjacent modal functions;
s6, substituting the decomposed IMF components into an SSA-BILSTM model to obtain a water inflow data prediction result, and accumulating and summing the water inflow data prediction result to obtain a final prediction result of the water inflow of the open pit;
s7, adopting absolute error MAE, root mean square error RMSE and average relative error MAPE, R 2 Evaluating and carrying out error analysis on a final prediction result of the water inflow of the open pit;
the absolute error MAE formula is:
MAE；
in which y i Is a true value of the code,representing the predicted value;
the root mean square error RMSE formula is:
；
in which y i Is a true value of the code,representing the predicted value;
the average relative error MAPE formula is:
；
wherein y is i Is a true value of the code,representing the predicted value;
R 2 the formula is:
；
in which y i Is a true value of the code,the predicted value is represented by a value of the prediction,representing the average of the true values.
Example 1: a prediction method of water inflow of an open pit (see figures 1 and 2) comprises the following specific steps:
s1, acquiring weather data (the weather data comprises air temperature, air pressure, humidity and rainfall) of a historical time sequence of a certain open pit 2020 in Yunnan and corresponding open pit water inflow, and supplementing missing values to obtain 360 groups of original data sets (see table 1); the method for supplementing the missing value is to supplement the data by adopting the average value of the time points before and after, and the expression is as follows:
；
in the method, in the process of the invention,data representing an i-th influence factor of the d-th history time to be preprocessed, the influence factor affecting the water inflow of the pit;(d=1, 2, …, N) data representing the i-th influence factor affecting pit water inflow at the d-1-th history;(d=1, 2, …, N) data representing the ith influence factor affecting pit water inflow at the (d+1) th historical time; i is {1,2, …, M }, i.e., M influence factors that influence the water inflow of the open sky pit; n is the original time sequence data length;
table 1 raw dataset
| Rainfall on the same day (mm) | Average air temperature (DEG C) | Air pressure (hpa) | Humidity (%) | Pit water inflow (m) 3 .h -1 ) |
| 0.8 | 9 | 870 | 79.097 | 1195 |
| 2.3 | 7 | 871 | 82.297 | 1254 |
| 3.5 | 5 | 870 | 83.498 | 1355 |
| …… | …… | …… | …… | …… |
| 1.1 | 6 | 851 | 83.699 | 1216 |
| 13.2 | 11 | 851 | 84.697 | 1451 |
S2, initializing algorithm parameters
Improved northern hawk by initializationAlgorithm population parameters, set range of VMD process parameters (k, α) values, VMD process parameters tau set to 0, DC set to 0, init set to 1, tol set to 1×10 -7 Tau is a frequency penalty factor, DC is whether a direct current component is included, init is an initialization mode, and tol is an iteration tolerance; the value range of k is [2,10 ]]Alpha is within the range of [100,10000 ]]；
The initializing and improving northern hawk algorithm population parameters are the ith individual position of the initializing population:
X i =lb+r⊙(ub-lb)
where Xi is the position of the initialized population individual in the ith dimension, lb is the lower boundary vector, ub is the upper boundary vector, r is a random matrix with the size (population scale x 1), and l is the matrix point multiplication operation;
s3, adopting an improved hawk algorithm and optimizing the VMD processing parameter (k, alpha) value according to the minimum sample entropy of the water inflow data of the open pit;
the sample entropy calculation formula is as follows:
；
where m is the mode dimension, r is the similarity tolerance threshold,probability of matching m points for two sequences under a similar tolerance r, wherein N is data number;the ratio of the approximate number to the total number is N-m-1;
adding 1 to the mode dimension to obtain m+1The sample entropy formula is transformed into:
；
in the method, in the process of the invention,for the entropy value of the sample,probability of matching m+1 points for two sequences;
the VMD processing parameter (k, alpha) value optimizing specific steps include:
s31, improving a northern hawk algorithm, as shown in fig. 2, adopting a particle swarm optimization strategy to change an exploration stage in a prey identification stage (exploration stage) of the northern hawk algorithm, optimizing a random prey selection mode, and changing a j-th vinylogous position of the northern hawk in a first stage; the formula of the particle swarm optimization strategy is as follows
；
；
Wherein:a j-th dimensional velocity vector for particle i in the k-th iteration;a position vector of the particle i in the j-th dimension in the k-1 th iteration;,is interval [0,1 ]]Random numbers in the search module, so that the randomness of the search is increased;the historical optimal position of the particle i in the j-th dimension in the k-th iteration;historical optimal positions of the group in the j-th dimension in the k-th iteration; 0.7 is inertiaThe weight is used for adjusting the search range of the solution space; 2 is a learning factor;
s32, updating the northern hawk position in the first stage:
；
in the method, in the process of the invention,is the new position of the 1 st stage of the ith northern hawk,is the objective function value based on the ith northern eagle after the 1 st stage update;
calculating the position of a northern hawk algorithm in the second stage:
；
in the method, in the process of the invention,a new position in the j-th dimension for the ith northern hawk;
redefining an attenuation coefficient in a northern eagle escape action stage (development stage), and determining R in a mixed attenuation mode to obtain an improved northern eagle algorithm; wherein the mixed attenuation mode comprises exponential attenuation and adaptive attenuation; the attenuation coefficient is given by
；
Wherein t is the current iteration number; t is the maximum iteration number; r is an attenuation coefficient;storing an array of objective function values for each northern hawk game location;store the purpose of each new position of the northern hawkA set of standard function values;
when (when)When the algorithm is started, an exponential nonlinear attenuation strategy is adopted in the early stage, so that the early stage convergence of the algorithm is rapid; when (when)When the method is used, the self-adaptive attenuation strategy is adopted in the later period, so that the searching is smoother and more flexible, and sudden jump or oscillation in the searching process is avoided; the mixed attenuation strategy is improved, and the searching efficiency and the resolving quality of the algorithm are improved;
updating the position of the second stage of the northern eagle:
；
in the method, in the process of the invention,is the new position of the ith northern hawk;is the objective function value based on the ith northern eagle after the 2 nd stage update;
storing the optimal (k, alpha) combination value obtained by each iteration of PSRNGO;
s33, reaching the iteration times and outputting an optimal (k, alpha) combined value; the iteration effect diagrams of PSRNGO and NGO are shown in fig. 6 and 7, and as can be seen from fig. 6 and 7, through multiple algorithm tests, compared with the original algorithm NGO and PSRNGO, the iteration speed is faster in searching the VMD optimal parameter, and the optimal solution is calculated within the range of 10 iterations;
s4, performing VMD signal processing on the open pit water inflow historical data based on the VMD processing parameter (k, alpha) optimal combination value to obtain a plurality of IMF components, wherein the decomposition result is shown in figure 3;
s5, carrying out correlation analysis on the IMF modal function, judging the Pearson correlation coefficient of the adjacent modal function by taking 0.1 as a threshold value, and verifying the VMD decomposition quality; specifically, each IMF component is decomposed into a pair of adjacent IMsThe correlation coefficient calculation is carried out on the F component to verify the optimization effect, and the square value r of the Pearson coefficient of the adjacent modal function is used 2 Is a reference;
the formula of the adjacent mode function Pearson is:
；
wherein:are respectively toStandard fraction of sample, sample average and sample standard deviation;，are respectively toStandard fraction of sample, sample average and sample standard deviation;for the correlation coefficient value, n is the number of samples, r 2 Judging correlation coefficients of two adjacent modal functions;
when the square value r of the Pearson coefficient of the adjacent mode function 2 If the amount is more than 0.1, the decomposition effect is considered to be poor; if the square value r of the Pearson coefficient of the adjacent mode function 2 All are smaller than 0.1, the correlation of the adjacent modal components IMF is low, namely the decomposition efficiency is good;
in the embodiment, the optimal k value is 7, namely 7 IMF components exist, the alpha value is 2416, the combination decomposition is carried out, the correlation coefficient diagram of the intrinsic mode components is shown in fig. 4, and the adjacent correlation coefficients of IMF1-IMF7 are all smaller than 0.1; when the k value is 8, the correlation coefficient diagram of the intrinsic mode component is shown in fig. 5, and the correlation between IMF5 and IMF6 is larger than 0.1; so k is 7, α is 2416 is the optimal solution of the PSRNGO algorithm with minimum sample entropy as the objective function;
s4, substituting the decomposed n IMF components into a constructed prediction model SSA-BILSTM to perform prediction to obtain n water inflow data prediction results Y, and accumulating and summing the n water inflow data prediction results Y to obtain a final prediction result of the water inflow of the open pit; the improved northern eagle algorithm optimizing VMD substitutes the model SSA-BILSTM prediction result as shown in fig. 6, the last 68 data of the data set is divided into a test set, and the test set is compared with the model prediction result, and the VMD adopts the improved NGO algorithm optimizing; in the figure, two prediction models of BILSTM and VMD-BILSTM are compared, so that the prediction result of VMD-SSA-BILSTM is more accurate;
s5, adopting absolute error MAE, root mean square error RMSE and average relative error MAPE, R 2 Evaluating and carrying out error analysis on a final prediction result of the water inflow of the open pit;
the absolute error MAE formula is:
MAE；
in which y i Is a true value of the code,representing the predicted value;
the root mean square error RMSE formula is:
；
in which y i Is a true value of the code,representing the predicted value;
the average relative error MAPE formula is:
；
wherein y is i Is a true value of the code,representing the predicted value;
R 2 the formula is:
；
in which y i Is a true value of the code,the predicted value is represented by a value of the prediction,representing an average of the true values;
according to the model predictive evaluation formula, the results are shown in the following table 2:
table 2 predictive evaluation formula results
| RMSE | MAE | MAPE | R 2 | |
| VMD-BILSTM | 35.0763 | 29.1912 | 2.4360% | 0.5320 |
| VMD-SSA-BILSTM | 15.1158 | 12.0025 | 0.9966% | 0.9130 |
| VMD*-BILSTM | 24.4861 | 20.2984 | 1.6966% | 0.6174 |
| VMD*-SSA-BILSTM | 7.5807 | 6.1115 | 0.5084% | 0.9603 |
The model prediction accuracy comparison graph after the optimized VMD and the un-optimized VMD is shown in fig. 9-12, wherein the model absolute error MAE comparison graph, the root mean square error RMSE comparison graph, the average relative error MAPE comparison graph and R after the optimized VMD and the un-optimized VMD 2 The comparison of the graphs is shown in fig. 9, 10, 11 and 12, respectively, and it can be seen that the PSRNGO-optimized VMD of the embodiment has a better prediction effect on the prediction of the water inflow of the open pit, improves the performance of the NGO algorithm, and improves the R before the optimization of the algorithm 2 The accuracy of the predicted result is improved to 0.9603 by the improved algorithm optimization of 0.913, and the PSRNGO-VMD-SSA-BILSTM is adopted to have good performance on the prediction of the water inflow of the open pit, so that a new thought and technical means are provided for the prediction of the water inflow of the open pit.
While the specific embodiments of the present invention have been described in detail, the present invention is not limited to the above embodiments, and various changes may be made without departing from the spirit of the present invention within the knowledge of those skilled in the art.
Claims (7)
1. The method for predicting the water inflow of the open pit is characterized by comprising the following specific steps of:
s1, acquiring weather data of a historical time sequence and corresponding water inflow of an open pit, and supplementing a missing value to an original data set;
s2, initializing population parameters of the improved northern eagle algorithm, setting a range of VMD processing parameters (k, alpha), wherein tau is set to 0, DC is set to 0, init is set to 1, and tol is set to 1 multiplied by 10 -7 ；
S3, adopting an improved northern eagle algorithm, and carrying out VMD processing parameter (k, alpha) value optimization according to the minimum sample entropy of the water inflow data of the open pit to obtain an optimal combination value of the VMD processing parameters (k, alpha);
s4, performing VMD signal processing on the open pit water inflow historical data based on the VMD processing parameter (k, alpha) optimal combination value to obtain a plurality of IMF components;
s5, carrying out correlation analysis on the IMF modal function, judging the Pearson correlation coefficient of the adjacent modal function by taking 0.1 as a threshold value, and verifying the VMD decomposition quality;
s6, substituting the decomposed IMF components into an SSA-BILSTM model to obtain a water inflow data prediction result, and accumulating and summing the water inflow data prediction result to obtain a final prediction result of the water inflow of the open pit;
s7, adopting absolute error MAE, root mean square error RMSE and average relative error MAPE, R 2 Evaluating and carrying out error analysis on a final prediction result of the water inflow of the open pit;
the specific step of optimizing the VMD processing parameter (k, α) value in the step S3 includes:
s31, adopting a particle swarm optimization strategy to change the exploration stage in the northern hawk algorithm prey identification stage; the formula of the particle swarm optimization strategy is as follows
；
；
Wherein:a j-th dimensional velocity vector for particle i in the k-th iteration; />A position vector of the particle i in the j-th dimension in the k-1 th iteration; />,/>Is interval [0,1 ]]A random number within; />The historical optimal position of the particle i in the j-th dimension in the k-th iteration; />Historical optimal positions of the group in the j-th dimension in the k-th iteration; 0.7 is an inertial weight for adjusting the search range for the solution space; 2 is a learning factor;
s32, redefining an attenuation coefficient in a northern hawk escape action stage, and determining R in a mixed attenuation mode to obtain an improved northern hawk algorithm; wherein the mixed attenuation mode comprises exponential attenuation and adaptive attenuation; the attenuation coefficient is given by
The method comprises the steps of carrying out a first treatment on the surface of the Wherein t is the current iteration number; t is the maximum iteration number; r is an attenuation coefficient; />Storing an array of objective function values for each northern hawk game location; />Storing an objective function value array of each new north eagle position;
s33, optimizing the VMD processing parameter (k, alpha) by using an improved northern hawk algorithm to obtain an optimal parameter combination (k, alpha).
2. The method for predicting water inflow in an open pit according to claim 1, wherein: the weather data in step S1 includes air temperature, air pressure, humidity and rainfall.
3. The method for predicting water inflow in an open pit according to claim 2, wherein: the method for supplementing the missing value in the step S1 is to supplement the average value of the data at the time points before and after, and the expression is as follows
The method comprises the steps of carrying out a first treatment on the surface of the In (1) the->Data representing an i-th influence factor of the d-th history time to be preprocessed, the influence factor affecting the water inflow of the pit; />(d=1, 2, …, N) data representing the i-th influence factor affecting pit water inflow at the d-1-th history; />(d=1, 2, …, N) data representing the ith influence factor affecting pit water inflow at the (d+1) th historical time; i is {1,2, …, M }, i.e., M influence factors that influence the water inflow of the open sky pit; n is the original time series data length.
4. The method for predicting water inflow in an open pit according to claim 1, wherein: in the step S2, initializing the population parameters of the northern eagle algorithm to be the ith individual position of the initialized population:
X i =lb+r⊙(ub-lb)
wherein X is i To initialize the position of population individuals in the ith dimension, lb is the lower boundary vector, ub is the upper boundary vector, r is a random matrix of size (population size. Times.1), and c is the matrix dot product operation.
5. The method for predicting water inflow in an open pit according to claim 1, wherein: the sample entropy calculation formula in step S3 is:
the method comprises the steps of carrying out a first treatment on the surface of the Where m is the mode dimension, r is the similarity tolerance threshold, +.>Probability of matching m points for two sequences under a similar tolerance r, wherein N is data number; />The ratio of the approximate number to the total number is N-m-1;
adding 1 to the mode dimension to obtain m+1The sample entropy formula is transformed into:
the method comprises the steps of carrying out a first treatment on the surface of the In (1) the->For the sample entropy value, +.>The probability of matching m+1 points for two sequences.
6. The method for predicting water inflow in an open pit according to claim 1, wherein: the formula of the adjacent mode function Pearson in step S5 is:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein: />Are respectively pair->Standard fraction of sample, sample average and sample standard deviation; />，/>Are respectively pair->Standard fraction of sample, sample average and sample standard deviation; />For the correlation coefficient value, n is the number of samples.
7. The method for predicting water inflow in an open pit according to claim 1, wherein: in step S7, the absolute error MAE formula is:
MAEthe method comprises the steps of carrying out a first treatment on the surface of the In which y i Is a true value, < >>Representing the predicted value;
the root mean square error RMSE formula is:
the method comprises the steps of carrying out a first treatment on the surface of the In which y i Is a true value, < >>Representing pre-emphasisMeasuring a value;
the average relative error MAPE formula is:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein y is i Is a true value, < >>Representing the predicted value;
R 2 the formula is:
the method comprises the steps of carrying out a first treatment on the surface of the In which y i Is a true value, < >>Representing predicted values +.>Representing the average of the true values.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311154516.7A CN116881676B (en) | 2023-09-08 | 2023-09-08 | Prediction method for water inflow of open pit |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311154516.7A CN116881676B (en) | 2023-09-08 | 2023-09-08 | Prediction method for water inflow of open pit |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116881676A CN116881676A (en) | 2023-10-13 |
| CN116881676B true CN116881676B (en) | 2023-11-21 |
Family
ID=88257290
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311154516.7A Active CN116881676B (en) | 2023-09-08 | 2023-09-08 | Prediction method for water inflow of open pit |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116881676B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115511177A (en) * | 2022-09-26 | 2022-12-23 | 三峡大学 | Ultra-short-term wind speed prediction method based on INGO-SWGMN hybrid model |
| CN116309668A (en) * | 2023-03-21 | 2023-06-23 | 江苏理工学院 | Complex image multi-threshold segmentation method based on improved gray wolf optimization algorithm |
| CN116361718A (en) * | 2023-03-23 | 2023-06-30 | 中国人民解放军海军工程大学 | Single-phase earth fault line selection method based on improved variational modal decomposition and dispersion entropy |
| CN116454863A (en) * | 2023-03-16 | 2023-07-18 | 南昌大学 | Optimal weight determining method of wind power combination prediction model based on improved hawk optimization algorithm |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110427916B (en) * | 2019-08-14 | 2020-05-01 | 苏州大学 | Fault diagnosis method under action of central frequency convergence tendency |
| CN112653142B (en) * | 2020-12-18 | 2022-09-02 | 武汉大学 | Wind power prediction method and system for optimizing depth transform network |
-
2023
- 2023-09-08 CN CN202311154516.7A patent/CN116881676B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115511177A (en) * | 2022-09-26 | 2022-12-23 | 三峡大学 | Ultra-short-term wind speed prediction method based on INGO-SWGMN hybrid model |
| CN116454863A (en) * | 2023-03-16 | 2023-07-18 | 南昌大学 | Optimal weight determining method of wind power combination prediction model based on improved hawk optimization algorithm |
| CN116309668A (en) * | 2023-03-21 | 2023-06-23 | 江苏理工学院 | Complex image multi-threshold segmentation method based on improved gray wolf optimization algorithm |
| CN116361718A (en) * | 2023-03-23 | 2023-06-30 | 中国人民解放军海军工程大学 | Single-phase earth fault line selection method based on improved variational modal decomposition and dispersion entropy |
Non-Patent Citations (2)
| Title |
|---|
| "Program design based on parametric of 3D tunnel for mine ventilation network";XiaoDong Wang等;《2011 International Conference on Business Management and Electronic Information》;全文 * |
| "基于VMD-DBN的矿井涌水量预测方法";刘慧等;《煤田地质与勘探》;第51卷卷(第第306期期);第13-21页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116881676A (en) | 2023-10-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114254561B (en) | Waterlogging prediction method, system and storage medium | |
| CN112989708B (en) | Well logging lithology identification method and system based on LSTM neural network | |
| CN109523013B (en) | Air particulate matter pollution degree estimation method based on shallow convolutional neural network | |
| CN111401599B (en) | Water level prediction method based on similarity search and LSTM neural network | |
| CN110942194A (en) | Wind power prediction error interval evaluation method based on TCN | |
| CN113326969A (en) | Short-term wind speed prediction method and system based on improved whale algorithm for optimizing ELM | |
| CN111242351A (en) | Tropical cyclone track prediction method based on self-encoder and GRU neural network | |
| CN116720620A (en) | Grain storage ventilation temperature prediction method based on IPSO algorithm optimization CNN-BiGRU-Attention network model | |
| CN112614021B (en) | Tunnel surrounding rock geological information prediction method based on built tunnel information intelligent identification | |
| CN113486591B (en) | Gravity multi-parameter data density weighted inversion method for convolutional neural network result | |
| CN109754122A (en) | A kind of Numerical Predicting Method of the BP neural network based on random forest feature extraction | |
| CN117170294B (en) | Intelligent control method of satellite thermal control system based on space thermal environment prediction | |
| Ahmadi et al. | Input data selection for solar radiation estimation | |
| CN116345555A (en) | CNN-ISCA-LSTM model-based short-term photovoltaic power generation power prediction method | |
| CN112926251B (en) | Landslide displacement high-precision prediction method based on machine learning | |
| CN110598914A (en) | Mine disaster gas concentration interval prediction method and system under influence of multiple factors | |
| CN116881676B (en) | Prediction method for water inflow of open pit | |
| CN116628886A (en) | Real-time optimization method for shield tunneling machine crossing construction parameters based on limited engineering data | |
| CN114862007A (en) | Short-period gas production rate prediction method and system for carbonate gas well | |
| CN115860165A (en) | Neural network basin rainfall runoff forecasting method and system considering initial loss | |
| CN114282614B (en) | Medium-long runoff prediction method for optimizing CNN-GRU based on random forest and IFDA | |
| CN113361476B (en) | Zhang Heng one-number pre-earthquake abnormal signal identification method based on artificial intelligence technology | |
| CN116187206B (en) | COD spectrum data migration method based on generation countermeasure network | |
| CN113255986B (en) | Multi-step daily runoff forecasting method based on meteorological information and deep learning algorithm | |
| CN116167128A (en) | BP neural network establishment method and tunnel surrounding rock parameter inversion method based on IBGA |
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 |