CN112862168A - Neural network multi-model combination-based population density prediction method and system - Google Patents
Neural network multi-model combination-based population density prediction method and system Download PDFInfo
- Publication number
- CN112862168A CN112862168A CN202110116420.6A CN202110116420A CN112862168A CN 112862168 A CN112862168 A CN 112862168A CN 202110116420 A CN202110116420 A CN 202110116420A CN 112862168 A CN112862168 A CN 112862168A
- Authority
- CN
- China
- Prior art keywords
- neural network
- model
- data
- network model
- data set
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 60
- 241001414720 Cicadellidae Species 0.000 claims abstract description 59
- 238000012795 verification Methods 0.000 claims abstract description 30
- 238000012544 monitoring process Methods 0.000 claims abstract description 27
- 238000012545 processing Methods 0.000 claims abstract description 21
- 241000238631 Hexapoda Species 0.000 claims abstract description 14
- 230000010354 integration Effects 0.000 claims abstract description 9
- 241001122767 Theaceae Species 0.000 claims abstract 19
- 238000003062 neural network model Methods 0.000 claims description 55
- 238000013527 convolutional neural network Methods 0.000 claims description 49
- 230000007787 long-term memory Effects 0.000 claims description 29
- 230000006403 short-term memory Effects 0.000 claims description 29
- 230000015654 memory Effects 0.000 claims description 17
- 238000001514 detection method Methods 0.000 claims description 6
- 235000009024 Ceanothus sanguineus Nutrition 0.000 claims description 5
- 240000003553 Leptospermum scoparium Species 0.000 claims description 5
- 235000015459 Lycium barbarum Nutrition 0.000 claims description 5
- 238000011176 pooling Methods 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000000638 solvent extraction Methods 0.000 claims description 2
- 238000010200 validation analysis Methods 0.000 claims 1
- 241000607479 Yersinia pestis Species 0.000 abstract description 7
- 244000269722 Thea sinensis Species 0.000 description 42
- 230000006870 function Effects 0.000 description 6
- 230000000694 effects Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 2
- 235000006468 Thea sinensis Nutrition 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 240000009206 Hemerocallis fulva Species 0.000 description 1
- 235000002941 Hemerocallis fulva Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- YOBAEOGBNPPUQV-UHFFFAOYSA-N iron;trihydrate Chemical compound O.O.O.[Fe].[Fe] YOBAEOGBNPPUQV-UHFFFAOYSA-N 0.000 description 1
- 230000000306 recurrent 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Human Resources & Organizations (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Evolutionary Biology (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Agronomy & Crop Science (AREA)
- Animal Husbandry (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a population density prediction method and system based on neural network multi-model combination, wherein the method comprises the following steps: monitoring tea lesser leafhopper data and climate data, and performing data processing and data division to obtain a training data set and a verification data set; obtaining a plurality of trained models based on a training data set and a verification data set; integrating the models according to preset weight coefficients to obtain integrated models; and acquiring tea lesser leafhopper data and climatic data within a period of time, processing the data and inputting the processed data into the integrated model, and outputting a prediction result of the population density of the tea lesser leafhopper. The system comprises: the device comprises a data processing module, a training module, an integration module and a prediction module. By using the method, the population density of the tea lesser leafhopper can be accurately predicted, so that the insect pest occurrence condition is predicted. The population density prediction method and system based on neural network multi-model combination can be widely applied to the field of insect pest prediction.
Description
Technical Field
The invention relates to the field of insect pest prediction, in particular to a population density prediction method and system based on neural network multi-model combination.
Background
The method for predicting the population density of the tea lesser leafhopper has guiding significance on insect damage of a tea tree, the population density of the tea lesser leafhopper on the tea tree is related to climate factors such as air temperature, humidity, rainfall and the like, the current prediction method belongs to a traditional prediction method, for example, a traditional regression model is constructed for prediction, but the traditional regression model needs data to be a stable non-white noise sequence, the climate data such as air temperature, humidity, rainfall and the like have large variation fluctuation and do not have stable characteristics, the analysis capability on the non-stable climate data is weak, a traditional expert system prediction method is also used, and the obtained prediction result is only one range of the population density or insect damage outbreak time instead of an accurate population density value or outbreak time point. Therefore, the accuracy of the prediction result obtained by the conventional prediction method is low.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a population density prediction method and system based on neural network multi-model combination, which can accurately predict the population density of tea lesser leafhoppers so as to accurately predict the occurrence condition of insect pests.
The first technical scheme adopted by the invention is as follows: a population density prediction method based on neural network multi-model combination comprises the following steps:
monitoring tea lesser leafhopper data and climate data, and performing data processing and data division to obtain a training data set and a verification data set;
respectively training and optimizing parameters of a pre-constructed BP neural network model, a convolutional neural network model and a long-short term memory model based on a training data set and a verification data set to obtain the trained BP neural network model, convolutional neural network model and long-short term memory model;
integrating the trained BP neural network model, the convolutional neural network model and the long-term and short-term memory model according to a preset weight coefficient to obtain an integrated model;
and acquiring tea lesser leafhopper data and climatic data within a period of time, processing the data and inputting the processed data into the integrated model, and outputting a prediction result of the population density of the tea lesser leafhopper.
Further, the step of monitoring tea lesser leafhopper data and climate data, performing data processing and data division to obtain a training data set and a verification data set specifically comprises:
counting the temperature, the rainfall and the sunshine duration of each day according to a detection instrument in the tea garden to obtain climate data;
counting the number of insect mouths of each louver of the tea lesser leafhoppers on the tea trees based on a 5-point sampling method according to the selected interval time to obtain the data of the tea lesser leafhoppers;
adding the climate data and the tea lesser leafhopper data into a data set according to corresponding dates, and dividing the data set into a training data set and a verification data set according to the proportion of 3: 1.
Further, the step of training and optimizing parameters of the pre-constructed BP neural network model, the convolutional neural network model and the long and short term memory model based on the training data set and the verification data set to obtain the trained BP neural network model, convolutional neural network model and long and short term memory model specifically includes:
training the pre-constructed BP neural network model by taking the population density of the tea lesser leafhopper at each detection time point in the training data set and each climate factor as input variables of the pre-constructed BP neural network model and taking the population density of the corresponding tea lesser leafhopper at the next time point as output variables to obtain the pre-trained BP neural network model;
using a sliding window method, taking the population density of the tea lesser leafhoppers at n continuous time points in the training data set and each climate factor as input variables of a pre-constructed convolutional neural network model, and taking the population density of the tea lesser leafhoppers at the corresponding next time point as output variables to train the pre-constructed convolutional neural network model to obtain a pre-trained convolutional neural network model;
taking the last monitoring time point in the training data set as an end point, sequentially selecting time points in the training data set from the first time point as an initial point to construct an input variable and an output variable, and training a pre-constructed long-short term memory model to obtain a pre-trained long-short term memory model;
and verifying and adjusting parameters of the pre-trained BP neural network model, the pre-trained convolutional neural network model and the pre-trained long and short term memory model based on a verification data set to obtain the trained BP neural network model, the trained convolutional neural network model and the trained long and short term memory model.
Further, the pre-constructed BP neural network model comprises an input layer, a hidden layer and an output layer, wherein the size of the input layer is 8, the size of the hidden layer is 16, and the size of the output layer is 1.
Further, the pre-constructed convolutional neural network model comprises a first input layer, a second convolutional layer, a third pooling layer, a fourth full-link layer and a fifth output layer.
Further, the long-term and short-term memory model comprises two hidden layers, the size of each hidden layer is 8, and two fully-connected layers are connected behind the hidden layers.
Further, the step of integrating the trained BP neural network model, convolutional neural network model and long-short term memory model according to a preset weight coefficient to obtain an integrated model specifically includes:
distributing weight coefficients to the trained BP neural network model, the trained convolutional neural network and the trained long and short term memory model to obtain an integrated model under the combination of a plurality of different weight coefficients;
and evaluating the integrated model under different weight coefficient combinations based on the verification data set, and obtaining the optimal integrated model according to the evaluation result.
The second technical scheme adopted by the invention is as follows: a population density prediction system based on neural network multi-model combination comprises:
the data processing module is used for monitoring tea lesser leafhopper data and climate data and performing data processing and data division to obtain a training data set and a verification data set;
the training module is used for respectively training and optimizing parameters of the pre-constructed BP neural network model, the convolutional neural network model and the long and short term memory model based on the training data set and the verification data set to obtain the trained BP neural network model, convolutional neural network model and the long and short term memory model;
the integration module is used for integrating the trained BP neural network model, the convolutional neural network model and the long-term and short-term memory model according to a preset weight coefficient to obtain an integration model;
and the prediction module is used for acquiring tea lesser leafhopper data and climate data within a period of time, processing the data and inputting the processed data into the integrated model, and outputting a prediction result of the population density of the tea lesser leafhopper.
The method and the system have the beneficial effects that: the invention solves the problem of limited prediction capability of the traditional method, can analyze unstable and nonlinear data such as temperature, humidity, rainfall and the like by predicting through a plurality of neural network models, respectively captures different characteristics in the data through each neural network model, and improves the prediction effect of the neural network models according to the relationship among the characteristics.
Drawings
FIG. 1 is a flow chart of the steps of a population density prediction method based on neural network multi-model combination according to the present invention;
FIG. 2 is a block diagram of a population density prediction system based on neural network multi-model combination according to the present invention;
FIG. 3 is a data processing diagram integrated in accordance with an embodiment of the present invention;
FIG. 4 is a block diagram of an LSTM unit in accordance with an embodiment of the present invention;
FIG. 5 is a comparison of insect pests predicted by the method of the present invention on tea plant, Yinghong Jiu;
FIG. 6 is a comparison of insect pests predicted by applying the method of the present invention to Huang 26858;
FIG. 7 is a comparison of insect pests predicted by the method of the present invention on tea plant of Hemerocallis fulva;
FIG. 8 is a MSE comparison plot of the predicted results for the single model and the integrated model at Yinghong No. nine, Huang 26858, and Jinxuan, respectively;
FIG. 9 is a comparison of MAE for single and integrated models in English Red nine, Huang 26858, and Jinxuan, respectively.
Detailed Description
The invention is described in further detail below with reference to the figures and the specific embodiments. The step numbers in the following embodiments are provided only for convenience of illustration, the order between the steps is not limited at all, and the execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.
The invention solves the problem of limited prediction capability of the traditional method, predicts the population density of the lesser leafhopper by integrating learning and combining the advantages of a multi-neural network model, and improves the prediction accuracy.
Referring to fig. 1, the invention provides a population density prediction method based on neural network multi-model combination, which comprises the following steps:
s1, monitoring tea lesser leafhopper data and climate data, and performing data processing and data division to obtain a training data set and a verification data set;
s2, respectively training and optimizing parameters of the pre-constructed BP neural network model, the convolutional neural network model and the long and short term memory model based on the training data set and the verification data set to obtain the trained BP neural network model, the convolutional neural network model and the long and short term memory model;
s3, integrating the trained BP neural network model, the convolutional neural network model and the long-term and short-term memory model according to a preset weight coefficient to obtain an integrated model;
s4, obtaining tea lesser leafhopper data and climate data within a period of time, inputting the processed data into the integrated model, and outputting the prediction result of the insect population density of the tea lesser leafhopper.
Further as a preferred embodiment of the method, the step of monitoring tea lesser leafhopper data and climate data, performing data processing and data partitioning to obtain a training data set and a verification data set specifically comprises:
counting the temperature, the rainfall and the sunshine duration of each day according to a detection instrument in the tea garden to obtain climate data;
counting the number of insect mouths of each louver of the tea lesser leafhoppers on the tea trees based on a 5-point sampling method according to the selected interval time to obtain the data of the tea lesser leafhoppers;
specifically, the number of the leaf openings of the tea lesser leafhoppers is manually monitored by using a 5-point sampling method, and the leaf openings of the lesser leafhoppers are monitored once every 7 days because the monitoring cost of the leaf openings of the lesser leafhoppers is high. And acquiring the data of temperature, humidity, rainfall and sunshine duration by using a monitoring instrument, and inputting the monitored data into the system.
In addition, as the climate data is obtained by monitoring every day, and the population density of the tea lesser leafhoppers is obtained by monitoring every 7 days, in order to facilitate model training, the average value of the climate factors 7 days (including the day) before the population density monitoring time point of each tea lesser leafhopper is taken as the climate data of the population density monitoring time point.
Adding the climate data and the tea lesser leafhopper data into a data set according to corresponding dates, and dividing the data set into a training data set and a verification data set according to the proportion of 3: 1.
Specifically, because the density variation trend of the insect population has periodicity with a period of years, three characteristics of a monitoring month, a monitoring day and a time interval with the next monitoring time point are derived based on the monitoring time point, data are divided into a training data set and a verification data set according to the proportion of 3:1, and the data are subjected to min-max standardization treatment, wherein the formula is as follows:
where min is the minimum of the sequence, max is the maximum of the sequence, x*After normalizing for xThe value of (c).
Further, as a preferred embodiment of the method, the step of training and optimizing parameters of the pre-constructed BP neural network model, the convolutional neural network model and the long-short term memory model based on the training data set and the verification data set to obtain the trained BP neural network model, the convolutional neural network model and the long-short term memory model specifically includes:
training the pre-constructed BP neural network model by taking the population density of the tea lesser leafhopper at each detection time point in the training data set and each climate factor as input variables of the pre-constructed BP neural network model and taking the population density of the corresponding tea lesser leafhopper at the next time point as output variables to obtain the pre-trained BP neural network model;
specifically, the BP neural network model takes the population density of the tea lesser leafhopper and each climate factor at each monitoring time point as input variables, and the population density of the tea lesser leafhopper at the next time point as output variables to construct training data of the BP neural network model.
Using a sliding window method, taking the population density of the tea lesser leafhoppers at n continuous time points in the training data set and each climate factor as input variables of a pre-constructed convolutional neural network model, and taking the population density of the tea lesser leafhoppers at the corresponding next time point as output variables to train the pre-constructed convolutional neural network model to obtain a pre-trained convolutional neural network model;
specifically, the size of the sliding window is selected to be 4, the population density of the tea lesser leafhoppers at each 4 continuous time points in the training data set and all climate factors are used as input variables, and the population density of the tea lesser leafhoppers at the next time point is used as an output variable to construct training data of the convolutional neural network model.
Taking the last monitoring time point in the training data set as an end point, sequentially selecting time points in the training data set from the first time point as an initial point to construct an input variable and an output variable, and training a pre-constructed long-short term memory model to obtain a pre-trained long-short term memory model;
specifically, in order to fully utilize the monitored data, if the training data set comprises n monitoring time points, 1-n-1 monitoring time points in the sequence are respectively used as starting point sites, the nth monitoring time point is used as an ending point site to construct n-1 sequences as input variables, and each input sequence is translated backwards by a worm density sequence corresponding to one time point to form an output variable.
And verifying and adjusting parameters of the pre-trained BP neural network model, the pre-trained convolutional neural network model and the pre-trained long and short term memory model based on a verification data set to obtain the trained BP neural network model, the trained convolutional neural network model and the trained long and short term memory model.
Further as a preferred embodiment of the method, the pre-constructed BP neural network model includes an input layer, a hidden layer, and an output layer, the size of the input layer is 8, the size of the hidden layer is 16, and the size of the output layer is 1.
Specifically, a tanh activation function is between the hidden layer and the output layer, the formula of the tanh activation function is as in formula (2), and the loss function is as in formula (3):
further as a preferred embodiment of the method, the pre-constructed convolutional neural network model includes a first input layer, a second convolutional layer, a third pooling layer, a fourth fully-connected layer, and a fifth output layer.
Specifically, the size of the selected sliding window is 4, and since 8 features are included, the size of the first layer input layer is 4 × 8, and the second layer convolutional layer uses two one-dimensional convolution kernels with different sizes, 2 and 3 respectively. The third layer of the pooling layer is a maximum pooling layer, and the fourth layer is a full-connection layer. The model is optimized by an Adam optimizer and the loss function formula is as in formula (3).
Further as a preferred embodiment of the method, the long-term and short-term memory model comprises two hidden layers, the size of each hidden layer is 8, and two fully-connected layers are connected behind the hidden layers.
Specifically, the long-term and short-term memory model is an improved recurrent neural network model, the structure of two fully-connected layers is connected after a double-layer LSTM is used in the invention, the size of a hidden layer of the LSTM is 8, the model is optimized through an Adam optimizer, a loss function formula is shown as a formula (3), and the structure of an LSTM unit refers to FIG. 4.
Further, as a preferred embodiment of the method, the step of integrating the trained BP neural network model, convolutional neural network model and long-short term memory model according to a preset weight coefficient to obtain an integrated model specifically includes:
distributing weight coefficients to the trained BP neural network model, the trained convolutional neural network and the trained long and short term memory model to obtain an integrated model under the combination of a plurality of different weight coefficients;
and evaluating the integrated model under different weight coefficient combinations based on the verification data set, and obtaining the optimal integrated model according to the evaluation result.
Specifically, in order to integrate the advantages of the extracted features of each model and improve the prediction effect, a weighted average method of ensemble learning is used to distribute weight coefficients to each model, the prediction effect of the ensemble model under different weight combinations is tested on a verification set, and the ensemble model refers to fig. 3.
Constructing the existing historical data into the format of input data of each model by using Yinghong Jiu, Huang 26858 and Jinxuan, forming the input of an integration model, obtaining a prediction result after inputting the integration model, and referring to fig. 5, 6 and 7 for the prediction result and an actual result.
The Mean Square Error (MSE) and the Mean Absolute Error (MAE) of the model prediction result are respectively shown in fig. 8 and fig. 9, and it can be seen that the MSE and the MAE of the integrated model are minimum, which indicates that the prediction effect of the integrated model is optimal.
As shown in fig. 2, a population density prediction system based on neural network multi-model combination includes:
the data processing module is used for monitoring tea lesser leafhopper data and climate data and performing data processing and data division to obtain a training data set and a verification data set;
the training module is used for respectively training and optimizing parameters of the pre-constructed BP neural network model, the convolutional neural network model and the long and short term memory model based on the training data set and the verification data set to obtain the trained BP neural network model, convolutional neural network model and the long and short term memory model;
the integration module is used for integrating the trained BP neural network model, the convolutional neural network model and the long-term and short-term memory model according to a preset weight coefficient to obtain an integration model;
and the prediction module is used for acquiring tea lesser leafhopper data and climate data within a period of time, processing the data and inputting the processed data into the integrated model, and outputting a prediction result of the population density of the tea lesser leafhopper.
The contents in the above method embodiments are all applicable to the present system embodiment, the functions specifically implemented by the present system embodiment are the same as those in the above method embodiment, and the beneficial effects achieved by the present system embodiment are also the same as those achieved by the above method embodiment.
While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (8)
1. A population density prediction method based on neural network multi-model combination is characterized by comprising the following steps:
monitoring tea lesser leafhopper data and climate data, and performing data processing and data division to obtain a training data set and a verification data set;
respectively training and optimizing parameters of a pre-constructed BP neural network model, a convolutional neural network model and a long-short term memory model based on a training data set and a verification data set to obtain the trained BP neural network model, convolutional neural network model and long-short term memory model;
integrating the trained BP neural network model, the convolutional neural network model and the long-term and short-term memory model according to a preset weight coefficient to obtain an integrated model;
and acquiring tea lesser leafhopper data and climatic data within a period of time, processing the data and inputting the processed data into the integrated model, and outputting a prediction result of the population density of the tea lesser leafhopper.
2. The population density prediction method based on neural network multi-model combination as claimed in claim 1, wherein the step of monitoring tea lesser leafhopper data and climate data and performing data processing and data partitioning to obtain a training data set and a verification data set specifically comprises:
counting the temperature, the rainfall and the sunshine duration of each day according to a detection instrument in the tea garden to obtain climate data;
counting the number of insect mouths of each louver of the tea lesser leafhoppers on the tea trees based on a 5-point sampling method according to the selected interval time to obtain the data of the tea lesser leafhoppers;
adding the climate data and the tea lesser leafhopper data into a data set according to corresponding dates, and dividing the data set into a training data set and a verification data set according to the proportion of 3: 1.
3. The population density prediction method based on neural network multi-model combination as claimed in claim 2, wherein the step of training and optimizing parameters of the pre-constructed BP neural network model, convolutional neural network model and long-short term memory model based on the training data set and validation data set respectively to obtain the trained BP neural network model, convolutional neural network model and long-short term memory model specifically comprises:
training the pre-constructed BP neural network model by taking the population density of the tea lesser leafhopper at each detection time point in the training data set and each climate factor as input variables of the pre-constructed BP neural network model and taking the population density of the corresponding tea lesser leafhopper at the next time point as output variables to obtain the pre-trained BP neural network model;
using a sliding window method, taking the population density of the tea lesser leafhoppers at n continuous time points in the training data set and each climate factor as input variables of a pre-constructed convolutional neural network model, and taking the population density of the tea lesser leafhoppers at the corresponding next time point as output variables to train the pre-constructed convolutional neural network model to obtain a pre-trained convolutional neural network model;
taking the last monitoring time point in the training data set as an end point, sequentially selecting time points in the training data set from the first time point as an initial point to construct an input variable and an output variable, and training a pre-constructed long-short term memory model to obtain a pre-trained long-short term memory model;
and verifying and adjusting parameters of the pre-trained BP neural network model, the pre-trained convolutional neural network model and the pre-trained long and short term memory model based on a verification data set to obtain the trained BP neural network model, the trained convolutional neural network model and the trained long and short term memory model.
4. The population density prediction method based on neural network multi-model combination as claimed in claim 3, wherein the pre-constructed BP neural network model comprises an input layer, a hidden layer and an output layer, the size of the input layer is 8, the size of the hidden layer is 16, and the size of the output layer is 1.
5. The population density prediction method based on neural network multi-model combination as claimed in claim 4, wherein the pre-constructed convolutional neural network model comprises a first input layer, a second convolutional layer, a third pooling layer, a fourth fully-connected layer and a fifth output layer.
6. The method as claimed in claim 5, wherein the long-term and short-term memory model includes two hidden layers, each hidden layer has a size of 8, and two fully-connected layers are connected after the hidden layer.
7. The population density prediction method based on neural network multi-model combination as claimed in claim 6, wherein the step of integrating the trained BP neural network model, convolutional neural network model and long-short term memory model according to preset weight coefficients to obtain an integrated model specifically comprises:
distributing weight coefficients to the trained BP neural network model, the trained convolutional neural network and the trained long and short term memory model to obtain an integrated model under the combination of a plurality of different weight coefficients;
and evaluating the integrated model under different weight coefficient combinations based on the verification data set, and obtaining the optimal integrated model according to the evaluation result.
8. A population density prediction system based on neural network multi-model combination is characterized by comprising:
the data processing module is used for monitoring tea lesser leafhopper data and climate data, processing and dividing the data to obtain a training data set and a verification data set;
the training module is used for respectively training and optimizing parameters of the pre-constructed BP neural network model, the convolutional neural network model and the long and short term memory model based on the training data set and the verification data set to obtain the trained BP neural network model, convolutional neural network model and the long and short term memory model;
the integration module is used for integrating the trained BP neural network model, the convolutional neural network model and the long-term and short-term memory model according to a preset weight coefficient to obtain an integration model;
and the prediction module is used for acquiring tea lesser leafhopper data and climate data within a period of time, processing the data and inputting the processed data into the integrated model, and outputting a prediction result of the population density of the tea lesser leafhopper.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110116420.6A CN112862168A (en) | 2021-01-28 | 2021-01-28 | Neural network multi-model combination-based population density prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110116420.6A CN112862168A (en) | 2021-01-28 | 2021-01-28 | Neural network multi-model combination-based population density prediction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112862168A true CN112862168A (en) | 2021-05-28 |
Family
ID=75987059
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110116420.6A Pending CN112862168A (en) | 2021-01-28 | 2021-01-28 | Neural network multi-model combination-based population density prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112862168A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109214592A (en) * | 2018-10-17 | 2019-01-15 | 北京工商大学 | A kind of Air Quality Forecast method of the deep learning of multi-model fusion |
| CN110245801A (en) * | 2019-06-19 | 2019-09-17 | 中国电力科学研究院有限公司 | A kind of Methods of electric load forecasting and system based on combination mining model |
| CN110837926A (en) * | 2019-11-04 | 2020-02-25 | 四川省烟草公司广元市公司 | Tobacco main pest and disease damage prediction method based on big data |
| CN111931973A (en) * | 2020-06-16 | 2020-11-13 | 广东电网有限责任公司 | Method for improving short-time load prediction precision of cable line |
-
2021
- 2021-01-28 CN CN202110116420.6A patent/CN112862168A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109214592A (en) * | 2018-10-17 | 2019-01-15 | 北京工商大学 | A kind of Air Quality Forecast method of the deep learning of multi-model fusion |
| CN110245801A (en) * | 2019-06-19 | 2019-09-17 | 中国电力科学研究院有限公司 | A kind of Methods of electric load forecasting and system based on combination mining model |
| CN110837926A (en) * | 2019-11-04 | 2020-02-25 | 四川省烟草公司广元市公司 | Tobacco main pest and disease damage prediction method based on big data |
| CN111931973A (en) * | 2020-06-16 | 2020-11-13 | 广东电网有限责任公司 | Method for improving short-time load prediction precision of cable line |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110674604B (en) | Transformer DGA data prediction method based on multi-dimensional time sequence frame convolution LSTM | |
| CN108256697B (en) | Prediction method for short-term load of power system | |
| CN109816221B (en) | Project risk decision method, apparatus, computer device and storage medium | |
| Mehrkanoon et al. | Approximate solutions to ordinary differential equations using least squares support vector machines | |
| CN112766549A (en) | Air pollutant concentration forecasting method and device and storage medium | |
| CN116757534B (en) | Intelligent refrigerator reliability analysis method based on neural training network | |
| CN112712209B (en) | Reservoir warehousing flow prediction method and device, computer equipment and storage medium | |
| CN111832222A (en) | Pollutant concentration prediction model training method, prediction method and device | |
| CN114363195B (en) | Network flow prediction and early warning method for time and frequency spectrum residual convolution network | |
| CN111950805B (en) | Medium-and-long-term power load prediction method and device, computer equipment and storage medium | |
| US20220284345A1 (en) | Device for optimizing training indicator of environment prediction model, and method for operating same | |
| CN111931983B (en) | Precipitation prediction method and system | |
| CN110579186B (en) | Crop growth monitoring method based on inversion of leaf area index by inverse Gaussian process | |
| CN112311488A (en) | Spectrum sensing method based on wavelet decomposition | |
| CN114330647A (en) | Model training method and device and silicon rod weight prediction method | |
| CN114169649A (en) | Traffic flow prediction method based on space-time data embedding | |
| CN115098999A (en) | Multi-mode fusion fuel cell system performance attenuation prediction method | |
| CN116128141A (en) | Storm surge prediction method and device, storage medium and electronic equipment | |
| CN114120637A (en) | Intelligent high-speed traffic flow prediction method based on continuous monitor | |
| CN114118634B (en) | Soil moisture prediction method and device | |
| CN112989548A (en) | Urban road traffic flow simulation deduction method based on multi-model combination | |
| CN110866672B (en) | Data processing method, device, terminal and medium | |
| Huang et al. | A decomposition‐based multi‐time dimension long short‐term memory model for short‐term electric load forecasting | |
| CN112949947A (en) | Power failure early warning method and system based on convolution long-term and short-term memory network | |
| CN112862168A (en) | Neural network multi-model combination-based population density prediction method and system |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210528 |
|
| RJ01 | Rejection of invention patent application after publication |