CN111292124A - Water demand prediction method based on optimized combined neural network - Google Patents
Water demand prediction method based on optimized combined neural network Download PDFInfo
- Publication number
- CN111292124A CN111292124A CN202010056323.8A CN202010056323A CN111292124A CN 111292124 A CN111292124 A CN 111292124A CN 202010056323 A CN202010056323 A CN 202010056323A CN 111292124 A CN111292124 A CN 111292124A
- Authority
- CN
- China
- Prior art keywords
- neural network
- water demand
- value
- influence
- algorithm
- 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.)
- Withdrawn
Links
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 66
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 38
- 238000012216 screening Methods 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims description 22
- 238000003062 neural network model Methods 0.000 claims description 11
- 239000013598 vector Substances 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 10
- 230000008859 change Effects 0.000 claims description 7
- 238000013459 approach Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000004088 simulation Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 4
- 230000002262 irrigation Effects 0.000 abstract description 3
- 238000003973 irrigation Methods 0.000 abstract description 3
- 235000020234 walnut Nutrition 0.000 description 42
- 241000758789 Juglans Species 0.000 description 33
- 235000009496 Juglans regia Nutrition 0.000 description 33
- 241000758791 Juglandaceae Species 0.000 description 9
- 230000006870 function Effects 0.000 description 5
- 238000005457 optimization Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000004069 differentiation Effects 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 239000003621 irrigation water Substances 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 108010014173 Factor X Proteins 0.000 description 1
- 206010036590 Premature baby Diseases 0.000 description 1
- 235000006040 Prunus persica var persica Nutrition 0.000 description 1
- 240000006413 Prunus persica var. persica Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Accounting & Taxation (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Marketing (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Entrepreneurship & Innovation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Genetics & Genomics (AREA)
- Primary Health Care (AREA)
- Physiology (AREA)
- Tourism & Hospitality (AREA)
- Human Resources & Organizations (AREA)
- Mining & Mineral Resources (AREA)
- Marine Sciences & Fisheries (AREA)
- Animal Husbandry (AREA)
- Agronomy & Crop Science (AREA)
- Game Theory and Decision Science (AREA)
Abstract
The invention discloses a water demand prediction method of an Elman neural network based on optimized combination, which comprises the steps of screening input variables based on an average influence value (MIV) algorithm and optimizing weight and threshold values of the Elman neural network by using a thought evolution algorithm (MEA). The MIV algorithm can effectively eliminate the information overlapping of the influencing factors and screen better indexes; the MEA algorithm has good global search capability and can solve the problem that a pure artificial neural network is limited by random selection of initial weight and threshold. Therefore, the method has higher accuracy and better prediction effect, can effectively predict and forecast the water demand of crops, ensures timely and reasonable adjustment of an irrigation system, and has certain application value in the prediction of the water demand of the crops.
Description
Technical Field
The invention relates to a prediction method, which is used for screening variables based on an average influence Value (MIV) algorithm, optimizing the weight and the threshold of an Elman neural network by utilizing a thought evolution algorithm (MEA), and predicting water demand of crops, and belongs to the technical field of water resource management and communication information.
Background
China is a country with quite short water resources and is also a big agricultural country, and the agricultural water consumption accounts for the vast majority of the total water consumption. In recent years, the walnut planting area and total yield of China are increased rapidly, and the walnut planting area is extremely wide. As a main growing crop, if the water demand condition of walnuts in an important stage can be accurately forecasted, and the water demand condition is accurately and reasonably judged, the irrigation water demand can be forecasted, and an irrigation water plan can be better formulated.
The walnut water demand law is complicated and complicated, a complex nonlinear relation exists between the walnut water demand law and each influence factor, the neural network has a strong nonlinear problem processing function and is suitable for solving the problems, but the general neural network generally has the problems of early solution and easy falling into a local minimum value. On the other hand, the influence of main factors (wind speed, air temperature, relative humidity, sunshine hours and the like) influencing the walnut water demand on the walnut water demand is interconnected, the walnut water demand forecasting model has the characteristics of high complexity and multiple dimensions, and the interconnection of each influence index and the walnut water demand is mostly ignored in the existing crop water demand forecasting model.
Disclosure of Invention
The MIV algorithm can effectively eliminate the information overlapping of the influencing factors and screen better indexes; the MEA algorithm has good global search capability, the problem that a pure artificial neural network is limited by random selection of initial weights and threshold values can be solved, the MEA algorithm is used in cooperation with nonlinearity, high dimensionality, wide interconnectivity and adaptivity of neurons of the neural network, and a water demand prediction model adaptive to the water demand characteristics of walnut crops is established.
The application selects walnut crops in the accurate irrigation experimental field of Handan City, Hebei engineering university in Hebei province as experimental objects, and acquires 2015-flavored walnut water demand data of 1-31 days every 7 months in 2017 and meteorological data of Handan City, Hebei province. The effect of the prediction method is verified by taking meteorological data as an input variable and taking the water demand value of the walnut crop every day as an output variable of the network.
The invention adopts the following technical scheme:
an Elman neural network water demand prediction method based on optimized combination is characterized in that: firstly, screening input variables by using an average influence value (MIV) algorithm to find out main factors influencing water demand; and optimizing the weight and the threshold of the Elman neural network by using a thought evolution algorithm (MEA), and obtaining the Elman neural network with the optimal state for prediction.
The step of screening the input variables refers to the step of calculating average influence values of the influence factors by using an MIV algorithm, sorting the average influence values by comparing the average influence values, screening out main factors influencing water demand, and using the main factors as the input of an Elman neural network model.
Further, calculating the average impact value includes:
(1) inputting the crop water demand and the collected influencing factor samples into an Elman model for training;
(2) respectively adding or subtracting 10% to the data of each influencing factor in the sample, keeping the other indexes unchanged, forming a new sample, and respectively recording as Xi(1) And Xi(2) Deriving the prediction Y as a new inputi(1)、Yi(2) (ii) a Wherein, Xi(1) And Xi(2) Are respectively as
i (i ═ 1, 2.., n) denotes the ith influencing factor; x is the number ofmnA value representing the nth influence factor at the time of the mth observation;
the equations (1) and (2) are input as the trained neural network model again, and the output is 2n output vectors corresponding to the ith (i ═ 1, 2, …, n) influencing factor index when the index changes in the sample points:
solving the difference of the two equations to obtain the following vector of the influence change value of the ith influence factor index in each sample point on the output value of the water demand of the cropsTherefore, when the index of the ith influence factor is changed, the average influence value of the m water demand output values is as follows:
in the formula: IVi(j) Representative vector IViJ-th element (j ═ 1, 2.., m), MI bodyiAnd (3) an average influence value of the ith influence factor index in the data sample on the output result of the water demand of the crops, wherein i (i is 1, 2.
The 5 selected influencing factors are average air pressure, average air temperature, wind speed, sunshine hours and relative humidity.
Further, optimizing the weight and threshold of the Elman neural network by using the MEA algorithm comprises the following steps:
(1) dividing the imported data into a training set and a test set;
(2) setting parameters of a thought evolution algorithm;
(3) mapping a solution space to a coding space according to the topological structure of the Elman neural network, wherein each code corresponds to one solution of a problem to form an initial population; setting the reciprocal of the mean square error of the training set as a score function of each individual and the population, calculating the score of each individual according to the score function, selecting a winning individual and a temporary individual, and generating a plurality of new individuals around each individual by taking the winning individual and the temporary individual as centers to form a winning sub-population and a temporary sub-population;
(4) continuously carrying out iteration to approach an optimal path through operations such as convergence, diversification and the like, ending the overall optimal solution searching process when the iteration times set in a thought evolution algorithm or a mature criterion is met, and outputting optimal individuals as optimal initial weights and thresholds of the Elman neural network; the maturity criterion is that all temporal sub-populations score lower than the winning sub-population.
(5) Calculating network errors and feeding back the network errors to the Elman neural network to judge whether training is completed or not; the training is completed under the condition that the maximum iteration number set by the Elman neural network is reached or the error is smaller than the set training error.
(6) After the training is finished, carrying out simulation prediction and outputting a prediction result; otherwise, updating the weight value and the threshold value of the Elman neural network, adding 1 to the iteration number, and returning to the step (5).
Parameters of the thought evolution algorithm are set as follows: the population size was set to 200 populations, 5 winner sub-populations and 5 temporary sub-populations, the sub-populations size was 20, and the number of iterations was 10.
Drawings
FIG. 1 shows the water demand change of 7-month peaches in 2015 and 2016;
FIG. 2 shows a flow chart for screening factors affecting walnut crop water demand using the MIV algorithm;
FIG. 3 illustrates a thought evolution optimized neural network prediction flow diagram;
FIG. 4 shows the scoring of the temporary sub-population convergence process in the thought evolution algorithm;
FIG. 5 shows the scoring of the winning sub-population convergence process in the thought evolution algorithm; and
fig. 6 shows a comparison of predicted results and true value curves for the three models.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to specific embodiments and the accompanying drawings.
The application takes walnut crops as an example. Fig. 1 shows the walnut water demand change rule of 2015 and 2016 from 1 day to 31 days in two years and 7 months. According to the graph, the change of the water demand of the walnuts in the first ten days is stable, the water demand of the walnuts rises from the tenth day, the water demand of the walnuts is increased, the water demand of the walnuts rapidly drops after the peak value is reached, and the water demand of the walnuts is changed periodically in 7 months.
First, walnut water demand prediction process
Selecting five influence factors, namely average air pressure P (hPa), average air temperature T (DEG C), air speed F (m/s), sunshine hours S (h) and relative humidity RH (%), obtaining an average influence value of each influence factor through an MIV algorithm, sorting by comparing the sizes, screening out main factors influencing the water demand of walnuts, taking the main factors as an input layer of the established MEA-Elman neural network model, and generating a prediction result of the water demand of walnuts after numerical action of neurons of each layer on the influence factors through a four-layer structure.
The flow of screening the factors influencing the water demand of the walnut crops by using the MIV algorithm is shown in figure 2, and comprises the following steps:
(1) and inputting factor samples such as walnut water demand and collected weather into an Elman model for training.
(2) Respectively adding or subtracting 10% to the data of each influencing factor in the sample, keeping the other indexes unchanged, forming a new sample, and respectively recording as Xi(1) And Xi(2) Deriving the prediction Y as a new inputi(1)、Yi(2)。
(3) An average influence value of each influence factor is calculated.
(4) According to the above rules, sample construction of all the influence factors and calculation of an average influence value are completed.
(5) Sorting according to the absolute value of each influence factor MIV, screening variables, and screening out main factors influencing the water demand of walnuts.
Secondly, calculating the average influence value of each factor
For n influencing factors, an n-dimensional independent variable X is set, m observations are made, and a vector space X (X) is obtained1,x2,…,xm) T and the walnut water demand Y corresponding to each sample point are written as Y ═ Y1,y2,...,ym) And T. Using the influencing factor X as the input of the neural network and the walnut water demand Y as the output of the neural network, then performing network learning, and storing and trainingThe network parameter of (2). And (3) converting the X by the following steps (the data of each influencing factor is respectively added and subtracted by 10%, and the other indexes are kept unchanged) to obtain 2n input vectors:
where i (i ═ 1, 2.., n) denotes the ith influencing factor. x is the number ofmnThe value of the nth influencing factor at the time of the mth observation is shown.
The equations (1) and (2) are input as the trained neural network model again, and the output is 2n output vectors corresponding to the ith (i ═ 1, 2, …, n) influencing factor index when the index changes in the sample points:
solving the difference of the two formulas to obtain the following vector of the influence change value of the ith influence factor index in each sample point on the walnut water demand output value after the change of the indexTherefore, when the ith influence factor index changes, the average influence value of the m walnut water demand output values is as follows:
in the formula: IVi(j) Representative vector IViJ-th element (j ═ 1, 2.., m), MI bodyiAnd (3) an average influence value of the ith influence factor index in the walnut data sample on a walnut water demand output result, i (i is 1, 2.
TABLE 1 MIV Algorithm screening influence value results
Index of influence factor | Direction of influence | Standardized value of MIV | Sorting |
Wind speed | + | 0.06 | 5 |
Relative humidity | - | 0.21 | 3 |
Mean air temperature | + | 0.32 | 1 |
Mean air pressure | - | 0.23 | 2 |
Sunshine hours | + | 0.17 | 4 |
The importance degree of various influence factors on the walnut water demand is shown in the table 1, wherein the plus and minus signs represent the influence direction of the factors on the walnut water demand, the standardized numerical value represents the relative importance of the influence factors on the walnut water demand, and the following conclusion can be obtained from the table:
(1) wind speed, average air temperature, hours of sunshine have a positive effect on walnut water demand, while average air pressure and relative humidity have a negative effect.
(2) Sorting according to the influence degree: the average air temperature has the largest influence degree, and is arranged at the first place, and then the average air pressure, the relative humidity, the sunshine hours and the wind speed are sequentially arranged. Wherein the wind speed has minimal influence on the water demand in the walnut growth process.
Three, thinking evolution algorithm optimization neural network
The thought of the thought evolution algorithm is derived from a mode of simulating human thought evolution in biological evolution, including convergence and differentiation processes. Within the sub-population range, the process of competition of individuals for becoming winners is called convergence, each mature sub-population in the whole solution space carries out global competition for becoming winners, and new points in the solution space are continuously searched, and the process is called diversification. The two operations are respectively detected and developed, the functions are mutually promoted, a certain independence is kept, and the overall searching efficiency of the algorithm is improved.
The thinking evolution optimization neural network prediction process is shown in fig. 3, and comprises the following steps:
(1) the imported data is divided into a training set and a test set.
(2) And setting parameters of the thought evolution algorithm.
(3) And mapping the solution space to a coding space according to the topological structure of the Elman neural network, wherein each code corresponds to one solution of the problem to form an initial population. Selecting the reciprocal of the mean square error of the training set as a score function of each individual and each population, calculating the score of each individual, selecting a winning individual and a temporary individual, and generating a plurality of new individuals around each individual by taking the winning individual and the temporary individual as the center to form a plurality of winning sub-populations and temporary sub-populations.
(4) And continuously carrying out iteration to approach the optimal path through operations such as convergence, diversification and the like, finishing the searching process of the global optimal solution when the finishing condition is met, and outputting the optimal individual as the optimal initial weight and threshold of the Elman neural network. The end condition is to reach the iteration number set in the thought evolution algorithm or meet the maturity criterion.
(5) And calculating the network error and feeding back to the network to judge whether the training is finished.
(6) And after the training is finished, carrying out simulation prediction and outputting a prediction result. Otherwise, updating the weight value and the threshold value of the network, adding 1 to the iteration number, and returning to the step (5).
Preferably, the training is completed if the maximum number of iterations is reached or if the error is less than the set training error.
As shown in fig. 4 and 5, after several convergence operations, the score of each sub-population is not changed, indicating that they are mature. After the population is mature, a situation may occur in which the sub-population score of a temporary sub-population may be higher than the score of a sub-population in the winning sub-population, and at this time, a dissimilarity operation needs to be performed. Through the convergence and differentiation process, the optimization of the weight and the threshold of the neural network is finally completed when the maturity criterion is reached (the scores of all the temporary sub-populations are lower than the score of the winning sub-population).
The parameters of the thought evolution algorithm are set as follows: the population size is set to 200 populations, 5 winner sub-populations and 5 temporary sub-populations, the sub-populations being: and (2) changing 200/(5+5) to 20 groups, wherein the iteration times are 10, the 5 winning sub-groups and the 5 temporary sub-groups are respectively differentiated, and repeated and cyclic elimination is carried out continuously to obtain an optimal group serving as an initial weight and a threshold of the Elman neural network.
Setting initial parameters of an Elman neural network: the maximum iteration number is 500, the learning rate is 0.1, and the training error is 0.05.
An unimptimized Elman neural network model and a genetic algorithm optimized neural network model were added for error analysis for comparison, and the predicted results are shown in fig. 6. And analyzing and evaluating the walnut water demand prediction performance of each model by using two evaluation indexes, namely average absolute error MAE (mean absolute error) and root mean square error RMSE (root mean square error).
In the formula: q represents the number of predictions, μbkRepresents the predicted value of walnut water demand, muakAnd (4) representing the actual water demand value of the walnut.
MEA-Elman neural network: MAE 0.075 and RMSE 0.317
GA-Elman neural network: MAE 0.094 and RMSE 0.352
Elman neural network: MAE 0.139 and RMSE 0.517
The calculation result shows that the result predicted by the optimized neural network model is closer to the measured value than the result predicted by the common Elman neural network model. Compared with the model optimized by MEA, the neural network model optimized by the genetic algorithm has larger deviation between the predicted value and the true value, so that the thinking evolution algorithm has obvious superiority in solving the problems of prematurity and local minimum value existing in the solution of the Elman neural network. Therefore, the model provided by the invention has higher accuracy and better prediction effect, can effectively predict and forecast the water demand of crops, ensures timely and reasonable adjustment of an irrigation system, and has certain application value in the prediction of the water demand of walnut crops.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are only exemplary embodiments of the present invention, and are not intended to limit the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. An Elman neural network water demand prediction method based on optimized combination is characterized in that: firstly, screening input variables by using an average Impact Value (MIV) algorithm to find out main factors influencing water demand; and optimizing the weight and the threshold of the Elman neural network by using a thought evolution Algorithm (MEA), and obtaining the Elman neural network with the optimal state for prediction.
2. The method of claim 1, wherein the screening of the input variables means that an average influence value is calculated for the influence factors by using an MIV algorithm, and the influence factors are ranked by comparing the magnitudes to screen out main factors influencing water demand and serve as the input of the Elman neural network model.
3. The method of claim 2, the calculating the average impact value comprising:
(1) inputting the crop water demand and the collected influencing factor samples into an Elman model for training;
(2) respectively adding or subtracting 10% to the data of each influencing factor in the sample, keeping the other indexes unchanged, forming a new sample, and respectively recording as Xi (1)And Xi (2)Deriving the prediction Y as a new inputi (1)、Yi (2);
(3) An average influence value of each influence factor is calculated.
4. The method according to claim 3, further Xi (1)And Xi (2)Are respectively as
Wherein i (i ═ 1, 2.., n) denotes the ith influencing factor; x is the number ofmnA value representing the nth influence factor at the time of the mth observation;
the equations (1) and (2) are input as the trained neural network model again, and the output is 2n output vectors corresponding to the ith (i ═ 1, 2, …, n) influencing factor index when the index changes in the sample points:
solving the difference of the two equations to obtain the following vector of the influence change value of the ith influence factor index in each sample point on the output value of the water demand of the cropsThereby obtaining the first
When the index of the i influence factors is changed, the average influence value of the m water demand output values is as follows:
in the formula: IVi(j) Representative vector IViJ-th element (j ═ 1, 2.., m), MIViAnd (3) an average influence value of the ith influence factor index in the data sample on the output result of the water demand of the crops, wherein i (i is 1, 2.
5. The method of claim 4, selecting 5 influencing factors: average air pressure, average air temperature, wind speed, hours of sunshine, and relative humidity.
6. The method according to claim 1, wherein the optimizing the weights and thresholds of the Elman neural network using the MEA algorithm comprises the steps of:
(1) dividing the imported data into a training set and a test set;
(2) setting parameters of a thought evolution algorithm;
(3) mapping a solution space to a coding space according to the topological structure of the Elman neural network, wherein each code corresponds to one solution of a problem to form an initial population; calculating the score of each individual according to a score function, selecting a winning individual and a temporary individual, and generating a plurality of new individuals around each individual by taking the winning individual and the temporary individual as the center to form a winning sub-population and a temporary sub-population;
(4) continuously carrying out iteration to approach an optimal path through operations such as convergence, diversification and the like, finishing the searching process of the global optimal solution when the finishing condition is met, and outputting an optimal individual as an optimal initial weight and a threshold of the Elman neural network;
(5) calculating network errors and feeding back the network errors to the Elman neural network to judge whether training is completed or not;
(6) after the training is finished, carrying out simulation prediction and outputting a prediction result; otherwise, updating the weight value and the threshold value of the Elman neural network, adding 1 to the iteration number, and returning to the step (5).
7. The method of claim 6, setting the inverse of the mean square error of the training set as a function of the scores of the individuals and populations.
8. The method according to claim 7, wherein the ending condition in the step (4) is that the number of iterations set in the thought evolution algorithm is reached or that the scores of all the temporary sub-populations are lower than the score of the winning sub-population.
9. The method according to claim 8, wherein the training in step (5) is performed on the condition that the maximum number of iterations set by the Elman neural network is reached or the error is smaller than the set training error.
10. The method of claim 9, setting parameters of the thought evolution algorithm comprising: the population size was set to 200 populations, 5 winner sub-populations and 5 temporary sub-populations, the sub-populations size was 20, and the number of iterations was 10.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010056323.8A CN111292124A (en) | 2020-01-18 | 2020-01-18 | Water demand prediction method based on optimized combined neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010056323.8A CN111292124A (en) | 2020-01-18 | 2020-01-18 | Water demand prediction method based on optimized combined neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111292124A true CN111292124A (en) | 2020-06-16 |
Family
ID=71022277
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010056323.8A Withdrawn CN111292124A (en) | 2020-01-18 | 2020-01-18 | Water demand prediction method based on optimized combined neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111292124A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113052379A (en) * | 2021-03-17 | 2021-06-29 | 河北工程大学 | Tomato water demand prediction method based on BAS-RBF |
CN113469440A (en) * | 2021-07-02 | 2021-10-01 | 河海大学 | Method for predicting monthly-scale water demand for life |
CN114201923A (en) * | 2022-02-16 | 2022-03-18 | 中国地质调查局水文地质环境地质调查中心 | Processing method and device of landslide displacement prediction model |
CN117114450A (en) * | 2023-10-19 | 2023-11-24 | 江西农业大学 | Irrigation water decision-making method and system |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105631517A (en) * | 2015-12-17 | 2016-06-01 | 河海大学 | Photovoltaic power generation power short term prediction method based on mind evolution Elman neural network |
CN107609774A (en) * | 2017-09-11 | 2018-01-19 | 华北电力大学 | A kind of photovoltaic power Forecasting Methodology based on mind evolutionary Optimization of Wavelet neutral net |
CN109376863A (en) * | 2018-11-02 | 2019-02-22 | 国网浙江省电力有限公司宁波供电公司 | Photovoltaic power based on MIV-BP neural network is classified prediction technique in short term |
CN110363344A (en) * | 2019-07-11 | 2019-10-22 | 安徽理工大学 | Probability integral parameter prediction method based on MIV-GP algorithm optimization BP neural network |
CN110598947A (en) * | 2019-09-20 | 2019-12-20 | 东莞理工学院 | Load prediction method based on improved cuckoo-neural network algorithm |
-
2020
- 2020-01-18 CN CN202010056323.8A patent/CN111292124A/en not_active Withdrawn
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105631517A (en) * | 2015-12-17 | 2016-06-01 | 河海大学 | Photovoltaic power generation power short term prediction method based on mind evolution Elman neural network |
CN107609774A (en) * | 2017-09-11 | 2018-01-19 | 华北电力大学 | A kind of photovoltaic power Forecasting Methodology based on mind evolutionary Optimization of Wavelet neutral net |
CN109376863A (en) * | 2018-11-02 | 2019-02-22 | 国网浙江省电力有限公司宁波供电公司 | Photovoltaic power based on MIV-BP neural network is classified prediction technique in short term |
CN110363344A (en) * | 2019-07-11 | 2019-10-22 | 安徽理工大学 | Probability integral parameter prediction method based on MIV-GP algorithm optimization BP neural network |
CN110598947A (en) * | 2019-09-20 | 2019-12-20 | 东莞理工学院 | Load prediction method based on improved cuckoo-neural network algorithm |
Non-Patent Citations (3)
Title |
---|
孙丽娜,等.: "基于MIV-Elman神经网络的海洋生物酶发酵软测量", 《传感器与微系统》 * |
李志新,等.: "基于GA-Elman神经网络的参考作物需水量预测", 《节水灌溉》 * |
艾格林,等.: "基于MEA-ELMAN神经网络的光伏发电功率短期预测", 《电网与清洁能源》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113052379A (en) * | 2021-03-17 | 2021-06-29 | 河北工程大学 | Tomato water demand prediction method based on BAS-RBF |
CN113469440A (en) * | 2021-07-02 | 2021-10-01 | 河海大学 | Method for predicting monthly-scale water demand for life |
CN113469440B (en) * | 2021-07-02 | 2023-11-14 | 河海大学 | Method for predicting monthly scale living water demand |
CN114201923A (en) * | 2022-02-16 | 2022-03-18 | 中国地质调查局水文地质环境地质调查中心 | Processing method and device of landslide displacement prediction model |
CN117114450A (en) * | 2023-10-19 | 2023-11-24 | 江西农业大学 | Irrigation water decision-making method and system |
CN117114450B (en) * | 2023-10-19 | 2024-02-02 | 江西农业大学 | Irrigation water decision-making method and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111292124A (en) | Water demand prediction method based on optimized combined neural network | |
CN106971240A (en) | The short-term load forecasting method that a kind of variables choice is returned with Gaussian process | |
CN113468803B (en) | WOA-GRU flood flow prediction method and system based on improvement | |
CN112288164B (en) | Wind power combined prediction method considering spatial correlation and correcting numerical weather forecast | |
CN107609774B (en) | Photovoltaic power prediction method for optimizing wavelet neural network based on thought evolution algorithm | |
CN110175416A (en) | Three Gorges Reservoir water temperature prediction method based on principal component analysis and neural network | |
CN113344288B (en) | Cascade hydropower station group water level prediction method and device and computer readable storage medium | |
CN114444378A (en) | Short-term power prediction method for regional wind power cluster | |
CN109934422A (en) | Neural network wind speed prediction method based on time series data analysis | |
CN115689008A (en) | CNN-BilSTM short-term photovoltaic power prediction method and system based on ensemble empirical mode decomposition | |
CN113592132B (en) | Rainfall objective forecasting method based on numerical weather forecast and artificial intelligence | |
CN109754122A (en) | A kind of Numerical Predicting Method of the BP neural network based on random forest feature extraction | |
CN114462718A (en) | CNN-GRU wind power prediction method based on time sliding window | |
CN113988426A (en) | Electric vehicle charging load prediction method and system based on FCM clustering and LSTM | |
CN114169434A (en) | Load prediction method | |
Shang et al. | Research on intelligent pest prediction of based on improved artificial neural network | |
CN116702937A (en) | Photovoltaic output day-ahead prediction method based on K-means mean value clustering and BP neural network optimization | |
CN113221447A (en) | Soil humidity prediction method for optimizing BP neural network based on improved genetic algorithm | |
CN113379116A (en) | Cluster and convolutional neural network-based line loss prediction method for transformer area | |
CN114580762A (en) | Hydrological forecast error correction method based on XGboost | |
CN115995810A (en) | Wind power prediction method and system considering weather fluctuation self-adaptive matching | |
CN114897204A (en) | Method and device for predicting short-term wind speed of offshore wind farm | |
CN113962819A (en) | Method for predicting dissolved oxygen in industrial aquaculture based on extreme learning machine | |
CN117458480A (en) | Photovoltaic power generation power short-term prediction method and system based on improved LOF | |
CN111126827A (en) | Input-output accounting model construction method based on BP artificial neural network |
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 | ||
WW01 | Invention patent application withdrawn after publication | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20200616 |