CN107466816A - A kind of irrigation method based on dynamic multilayer extreme learning machine - Google Patents
A kind of irrigation method based on dynamic multilayer extreme learning machine Download PDFInfo
- Publication number
- CN107466816A CN107466816A CN201710605810.3A CN201710605810A CN107466816A CN 107466816 A CN107466816 A CN 107466816A CN 201710605810 A CN201710605810 A CN 201710605810A CN 107466816 A CN107466816 A CN 107466816A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- mtd
- mtr
- learning machine
- 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.)
- Granted
Links
- 230000002262 irrigation Effects 0.000 title claims abstract description 76
- 238000003973 irrigation Methods 0.000 title claims abstract description 75
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000012549 training Methods 0.000 claims abstract description 45
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 35
- 239000002689 soil Substances 0.000 claims abstract description 23
- 238000004364 calculation method Methods 0.000 claims abstract description 15
- 238000012360 testing method Methods 0.000 claims abstract description 11
- 239000011159 matrix material Substances 0.000 claims description 49
- 238000012545 processing Methods 0.000 claims description 13
- 230000004913 activation Effects 0.000 claims description 12
- 238000010606 normalization Methods 0.000 claims description 11
- 230000005855 radiation Effects 0.000 claims description 9
- 230000007246 mechanism Effects 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 241000894007 species Species 0.000 claims description 6
- 241001269238 Data Species 0.000 claims description 5
- 230000004907 flux Effects 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 2
- 238000004519 manufacturing process Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000013439 planning Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 210000004556 brain Anatomy 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 239000004576 sand Substances 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000009395 breeding Methods 0.000 description 1
- 230000001488 breeding effect Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 230000006735 deficit Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012958 reprocessing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000009331 sowing Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000005068 transpiration Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01G—HORTICULTURE; CULTIVATION OF VEGETABLES, FLOWERS, RICE, FRUIT, VINES, HOPS OR SEAWEED; FORESTRY; WATERING
- A01G25/00—Watering gardens, fields, sports grounds or the like
- A01G25/16—Control of watering
- A01G25/167—Control by humidity of the soil itself or of devices simulating soil or of the atmosphere; Soil humidity sensors
-
- 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
- G06F18/2155—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling
Landscapes
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Soil Sciences (AREA)
- Water Supply & Treatment (AREA)
- Environmental Sciences (AREA)
- Breeding Of Plants And Reproduction By Means Of Culturing (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of irrigation method based on dynamic multilayer extreme learning machine, gathers multigroup training data first, and training data includes soil environment data, meteorological data and the crop coefficient of irrigated crop；Then composing training collection is normalized to each group training data；Final mask is obtained using training set training multilayer extreme learning machine again；Final collecting test data are simultaneously input to final mask after it is normalized, and obtain predicting irrigation requirement, are irrigated according to prediction irrigation requirement.The present invention is according to computational accuracy using " seeking common ground while reserving difference " strategy, i.e. if model meets accuracy requirement to input data result of calculation again, then export the model, otherwise incremental learning training will be carried out on the basis of existing model, obtain the model that dynamic adjusts, the present invention improves the computational accuracy of irrigation requirement, reduces the loss of irrigation requirement predicted time and calculates cost, has reached rational utilization of water resources, the purpose of Rational Irrigation crops.
Description
Technical field
The invention belongs to agriculture Internet of Things intelligent irrigation field, is related to a kind of irrigation based on dynamic multilayer extreme learning machine
Method.
Background technology
Agriculture Internet of Things is highly integrated and integrated use of the generation information technology in agriculture field, and China's agricultural is believed
Breathization development there is important leading action, change conventional agriculture, promote agricultural to intellectuality, become more meticulous direction
Transformation.Production estimation environment real time information is gathered using substantial amounts of sensor node, monitoring system is formed by network technology
System, help peasant to pinpoint the problems in time, and accurately determine the position of generation problem.Make originally dependent on the life for isolating machinery
Production pattern turns to the intelligent production model centered on information and software, so as to reach the purpose of agricultural high-effiency production.
By calculating irrigation requirement exactly in agriculture Internet of Things intelligent irrigation field, understand Study on Crop Water Requirement Rules, be
Scientific and rational irrigation program is formulated, determines lrrigation Area amount, implements the basis finely irrigated；It is to reach water saving, height
Production, efficient purpose, realize the effective means and basic guarantee of Water Resources Irrigation sustainable development；It is to formulate river basin planning, area
Planning for water resources development, water resource arrange the basic foundation in the fields such as the planning, design, management of engineering using planning and filling.By accurate
Ground is calculated to determine water consumption of the irrigation method for reduction Crop growing stage, improves efficiency of water application, Developing Water-saving Agriculture
There is highly important meaning.
At present in the existing substantial amounts of achievement in research in agricultural irrigation method field, to be reached for irrigation management layer and policymaker
Intuitively Visual decision-making foundation is provided, instructs irrigated area to accomplish that timely and appropriate discovery is irrigated, improves the reasonable profit of Irrigation Project Design water resource
With rate.Patent of invention CN 201610951658.X disclose crop irrigation requirement computational methods under the conditions of a kind of Future Climate,
This method builds crop sowing date and Crop growing stage length according to the breeding time data and accumulated temperature formula of crop field test
To the response model of temperature；Calculate crop using Penman formula combination single crop coefficient and soil moisture stress coefficient needs day by day
Water；Crop irrigation requirement day by day is calculated based on deficit irrigation schedule and principle of water balance again.Although this method processing knot
Fruit is relatively accurate, but this method needs the information content that gathers more, such as the soil heat flux that Penman formula is related to, applicable surface compared with
It is small, it is difficult to promote.Patent of invention CN 201710020805.6 provides a kind of field irrigation watermeter and calculates method and apparatus, carries
Go out the Crop Evapotranspiration that the target area according to mechanism model calculates target area is irrigated, then calculate water requirement.Although this method
Result is more accurate, but this method equally exists soil parameters such as soil moisture content, the problem of obtaining costly, it is necessary to
Monitored using special equipment, use cost is higher.Patent of invention CN 201611093504.8 provides a kind of based on the improvement limit
The crop transpirstion amount Forecasting Methodology of learning machine, this method using particle swarm optimization algorithm optimization extreme learning machine network input layer with
Input weights and threshold value between hidden layer, this improves the computational accuracy of transpiration rate, but is taken in iteration searching process
Between relative this patent methods described it is relatively long, can not realize quick processing for large-scale data amount.
Therefore, a kind of data processing is quick, data processing scale is big while in the less situation of given data parameters species
The lower great application prospect of irrigation method that can accurately calculate the water demand of crop short time.
The content of the invention
The purpose of the present invention is to overcome in the prior art that data processing small scale, data processing speed are slow and to training number
According to species it is less the problems such as, there is provided a kind of data processing is quick, data processing scale is big at the same given data parameters species compared with
The irrigation method of the water demand of crop can accurately be calculated in the case of few the short time.
In order to achieve the above object, the technical solution adopted by the present invention is：
A kind of irrigation method based on dynamic multilayer extreme learning machine, the collection data related to known irrigation requirement are simultaneously
Multilayer extreme learning machine is trained to obtain final forecast model after being normalized, by the number related to irrigation requirement to be asked
Obtain being irrigated after predicting irrigation requirement according to final forecast model is input to after normalized, wherein, the multilayer limit
In habit machine last layer input and output simultaneously as next layer of input, simulate human brain consolidates learning mechanic again, dynamic
State multilayer extreme learning machine refers to that node in hidden layer dynamic updates multilayer extreme learning machine in the training process, hidden layer node
Several dynamic renewals, which is mainly based upon, has certain relation between precision of prediction and hidden layer node, thus can be according to current
The prediction result dynamic of data adjusts, and step is as follows：
(1) multigroup training data is gathered, every group of training data includes soil environment data, meteorological data and irrigated crop
Crop coefficient；
(2) every group of training data is normalized, all groups of training data composing training collection；
(3) final forecast model is obtained using training set training multilayer extreme learning machine；
(4) multigroup test data is gathered, every group of test data includes soil environment data, meteorological data and irrigated crop
Crop coefficient；
(5) final forecast model is input to after every group of test data being normalized, obtains predicting crop structure
Amount, irrigated according to prediction irrigation requirement；
It is described that comprising the following steps that for multilayer extreme learning machine is trained using training set：
1) all groups of training datas are divided into by n data block and order by acquisition time order according to sliding window size
Numbering, the unit of sliding window is group；
2) multilayer extreme learning machine (DELM) is trained to obtain model M 1 using data block 1, first by data in training process
Block 1 is input to first layer extreme learning machine and obtains first layer prediction result, then by data block 1 and first layer prediction result simultaneously
It is input to as input parameter in second layer extreme learning machine and obtains second layer prediction result, it is then that data block 1, first layer is pre-
Survey result and second layer prediction result is input in third layer extreme learning machine as input parameter and obtains third layer prediction simultaneously
As a result, the like obtain model M 1；
3) j=2 is made；
4) data block j input integrated predictive model C (j-2) are obtained predicting irrigation requirement and calculates precision of prediction piWith
Prediction result coefficient of determination R2, integrated predictive model C (j-2) refers to the obtained model of data block (j-1) training, integrated prediction mould
Type C0 is model M 1；
5) precision of prediction p is judgediWhether setting precision of prediction E1 is more than or equal to, if it is, defeated using " seeking common ground " strategy
It is integrated predictive model C (j-1) to go out model M (j-1), otherwise, using " depositing different " strategy, into next step；
6) according to prediction result coefficient of determination R2Update the node in hidden layer of multilayer extreme learning machine (DELM)；
7) the output weight matrix of incremental learning new mechanism multilayer extreme learning machine (DELM) is used according to data block j
Increment type multilayer extreme learning machine (IDELM) is obtained, whenever newly-increased data, and all knowledge bases need not be rebuild, but
On the basis of original knowledge base, the renewal caused by increasing data newly is only done, meets the thinking principle of human brain；
8) model M j is obtained using data block j training increment type multilayer extreme learning machines (IDELM), model M j is integrated
Forecast model C (j-1)；
9) j=j+1 is made；
10) circulation step 4)~8) to j=n, obtained integrated predictive model C (n-1) is final forecast model.
As preferable technical scheme：
A kind of irrigation method based on dynamic multilayer extreme learning machine as described above, the soil environment data and meteorology
Data are specially：It is the per day humidity of soil mean daily temperature, soil, air mean daily temperature, the per day humidity of air, per day
Total solar radiation, the wind speed and atmospheric pressure of 2m eminences, the unit of temperature for DEG C, the unit of humidity is %, it is described it is per day too
The unit of positive global radiation is MJm-2day-1, the unit of the wind speed is ms-1, the unit of the pressure is KPa；The crop
Coefficient refers to inquire about the empirical value that irrigated crop type provides each growth period according to different growing stages expert.
A kind of irrigation method based on dynamic multilayer extreme learning machine as described above, returning in step (2) and step (5)
One change processing refers to all data normalizations of collection to [- 1,1], normalization formula is as follows：
In formula, X is the data after normalization, and X* is to be currently needed for normalized data, xminTo need normalized data
In minimum value, xmaxTo need the maximum in normalized data.
A kind of irrigation method based on dynamic multilayer extreme learning machine as described above, in the training set containing 4000~
5000 groups of training datas, the sliding window size are 500 groups.
A kind of irrigation method based on dynamic multilayer extreme learning machine as described above, the multilayer extreme learning machine is double
Layer extreme learning machine.
A kind of irrigation method based on dynamic multilayer extreme learning machine as described above, the structure of the multilayer extreme learning machine
It is as follows to build step：
(1) input parameter of multilayer extreme learning machine network is initialized, and selects the activation primitive g (x) of hidden layer；It is described
The input layer of multilayer extreme learning machine network is soil environment data, meteorological data and crop coefficient, and output layer is crop structure
Amount；The input parameter includes the input number of plies, node in hidden layer and the output number of plies, wherein the input number of plies is input data species
Number, output the number of plies be 1, the initialization formula of node in hidden layer is as follows：
In formula,For the node in hidden layer of kth layer extreme learning machine,For the defeated of kth layer extreme learning machine
Enter node layer number,For the output layer nodes of kth layer extreme learning machine；
The activation primitive g (x) is used to calculate the output weights between hidden layer and output layer, the activation primitive g (x)
Here the equation for choosing sigmod functions is as follows：
In formula, x is independent variable, and in extreme learning machine network, x refers specifically to Wv·Xs+bv, WvFor input layer and hidden layer
Between connection weight, bvFor threshold value,XsFor input vector, e is natural constant,For node in hidden layer；
(2) random initializtion multilayer extreme learning machine network inputs are distinguished according to input layer number and node in hidden layer
Connection weight W between layer and hidden layervWith threshold value bv, initialisation range is [- 0.5,0.5]；
(3) according to the connection weight W between input layer and hidden layerv, threshold value bvIt is defeated that hidden layer is calculated with activation primitive g (x)
Go out matrix H, be specially：Given N group samples { (X is assumed in extreme learning machine networks, ts), s=1 ... N, it is assumed that basic pole
Limit learning machine (ELM) node in hidden layer beThen extreme learning machine network structure is as follows：
In formula, Xs=[Xs1, Xs2..., Xsn]T∈Rn, ts=[ts1, ts2..., tsm]T∈Rm, XsAnd tsRepresent respectively defeated
Enter variable and corresponding output variable, Wv=[Wv1,Wv2,...,Wvn]TIt is between connection v hidden layer nodes and n input layer
Input weight vector, bvIt is the threshold vector of v hidden layers,βv=[βv1,βv2,...,βvm]TIt is connection
Weight vector between v hidden layer nodes and m output layer, Wv·XsInner product is sought in expression；
Above-mentioned equation transform obtains：
H β=T；
In formula, H is hidden layer output matrix, and β is output weight matrix, and T is it is expected output matrix, and the element in T is served as reasons
FAO-56PM equations are multiplied by the water demand of crop that crop coefficient is calculated, wherein：
Because the input weights between input layer and hidden layer and threshold value are determination value, then WvAnd bvValue determination, Ke Yiji
Calculate hidden layer output matrix H；
(4) multilayer extreme learning machine output weight matrix β is calculated according to hidden layer output matrix H, formula is as follows：
In formula,Represent to ask the Moore-Penrose generalized inverses of matrix H, λ is regularization coefficient, and I is unit matrix, T
It is expected output matrix, the element in T is to be multiplied by the water demand of crop that crop coefficient is calculated by FAO-56PM equations.
A kind of irrigation method based on dynamic multilayer extreme learning machine as described above, the precision of prediction piCalculating it is public
Formula is as follows：
In formula, ETPM56(i) it is multiplied by with phase by FAO-56PM equations to the calculated value of i-th group of data water demand of crop
The water demand of crop that the crop coefficient answered obtains, unit mmday-1, ETpredicted(i) it is the prediction of i-th group of data crop
Irrigation requirement, unit mmday-1, N is the group number of input sample；
The span for setting precision of prediction E1 is, more than 0.90, requires to determine according to actual conditions precision of prediction；
The coefficient of determination R2Calculation formula it is as follows：
In formula, meanETpredictedTo seek the average of N group data prediction values；
The ETPM56(i) calculation formula is as follows：
In formula, Δ is the function relation curve slope of saturation vapour pressure-temperature, and unit is KPa DEG C-1, RnIt is preced with for input
The net radiation of layer, unit MJm-2day-1, G is soil heat flux, is ignored here, unit MJm-2day-1, esFor saturation
Vapour pressure, unit KPa, eaFor actual water vapor pressure, unit KPa, γ represent thermometer constant, and unit is KPa DEG C-1, Q is
Daily mean temperature, unit for DEG C, u is the wind speed of 2 meters of eminences, unit ms-1, Crop is crop coefficient；The calculation formula of Δ
It is as follows：
A kind of irrigation method based on dynamic multilayer extreme learning machine as described above, it is described that system is determined according to prediction result
Number R2The node in hidden layer of renewal multilayer extreme learning machine (DELM) refers to according to prediction result coefficient of determination R2Define renewal machine
System, specific formula are as follows：
In formula, c is positive integer, takes 5；ε, close to 0 constant, takes 0.01, it is 0 to prevent divisor for one；R2To determine to be
Number, coefficient of determination R2Closer to 1, precision of prediction is higher.
A kind of irrigation method based on dynamic multilayer extreme learning machine as described above, the increment type multilayer limit study
Machine refers to according to model M (j-1) and data block j that from new data learning new knowledge the data treated in the past need not
Reprocessing, compared with most basic extreme learning machine, often increases new data block, the output weight matrix of multilayer extreme learning machine
Need to be updated according to new data block：
Understand that the 1st data block output weight matrix calculation formula is according to basic limit learning machine：
The output weight matrix calculation formula of 2nd data block of incremental learning mechanism is：
In formula,
By that analogy, the output weight matrix β of increment type multilayer extreme learning machine when can obtain j-th of data block inputjSuch as
Under：
In formula, j >=2, HjThe hidden layer output matrix of increment type multilayer extreme learning machine, T when being inputted for j-th of data blockj
The desired output matrix of increment type multilayer extreme learning machine, β when being inputted for j-th of data block1It is more when being inputted for the 1st data block
The output weight matrix of layer extreme learning machine, H1The hidden layer of multilayer extreme learning machine exports square when being inputted for the 1st data block
Battle array, T1The desired output matrix of multilayer extreme learning machine when being inputted for the 1st data block.
Beneficial effect：
(1) a kind of irrigation method based on dynamic multilayer extreme learning machine of the invention, using the Intelligent Dynamic multilayer limit
Learning machine algorithm calculates irrigation requirement, it is proposed that multilayer extreme learning machine algorithm, the algorithm is for large-scale data processing effect
Fruit can be good than standard limit learning machine algorithm Generalization Capability and data processing stability；
(2) a kind of irrigation method based on dynamic multilayer extreme learning machine of the invention, according to calculating result dynamic
Hidden layer node number is adjusted, improving extreme learning machine algorithm, not only quickly, time loss is low for pace of learning, and is needed to irrigating
The computational accuracy of water is high；
(3) data of collection are divided into data by a kind of irrigation method based on dynamic multilayer extreme learning machine of the invention
Block, according to computational accuracy, using " seeking common ground while reserving difference " strategy, i.e., if model meets precision need to input data result of calculation again
Ask, just directly export the model, incremental learning training otherwise will be carried out on the basis of existing model, obtain what dynamic adjusted
Model, it is possible to achieve from new data learning new knowledge, and the data treated in the past do not need repetition training.
Brief description of the drawings
Fig. 1 is the step block diagram that one kind of the present invention is based on the irrigation method of dynamic multilayer extreme learning machine (DELM)；
Fig. 2 is the construction step block diagram of the multilayer extreme learning machine of the present invention；
Fig. 3 is the prediction crop irrigation requirement calculating network structural representation of the present invention；
Fig. 4 is the schematic diagram of the sigmoid activation primitives of the present invention；
Fig. 5 is that true crop water value, basic limit learning machine (ELM) the prediction water demand of crop and DELM predict crop
The comparison diagram of water requirement；
Fig. 6 is that ELM and DELM of the present invention predicts water demand of crop absolute value error curve map.
Embodiment
The invention will be further elucidated with reference to specific embodiments.It should be understood that these embodiments are merely to illustrate this hair
Bright rather than limitation the scope of the present invention.In addition, it is to be understood that after the content of the invention lectured has been read, art technology
Personnel can make various changes or modifications to the present invention, and these equivalent form of values equally fall within the application appended claims and limited
Fixed scope.
Embodiment 1
A kind of irrigation method based on the double-deck extreme learning machine of dynamic is as shown in figure 1, collection and known irrigation requirement phase
The data of pass simultaneously train double-deck layer extreme learning machine to obtain final forecast model after being normalized, will be with waiting to ask irrigation to need
Final forecast model is input to after the related data normalization processing of water to obtain being irrigated after predicting irrigation requirement, its
In, comprise the following steps that：
(1) the soil environment data and meteorological data in farmland are gathered, the soil environment data and meteorological data of collection come from
The agriculture internet of things equipment of Shanghai City information centre of agricultural commission research and development, including soil mean daily temperature and humidity, air day
Mean temperature and humidity, per day total solar radiation, the wind speed of 2m eminences, atmospheric pressure and corresponding crop coefficient, wherein
The unit of temperature for DEG C, the unit of humidity is %, and the unit of per day total solar radiation is MJm-2day-1, the unit of wind speed is
m·s-1, the unit of pressure is KPa；Inquiry irrigated crop type provides the experience in each growth period according to different growing stages expert
Value；
(2) all data of collection in step (1) are normalized to obtain training set；Normalization refers to the institute of collection
There is data normalization as follows to [- 1,1], normalization formula：
In formula, x is the data after normalization, and x* is to be currently needed for normalized data, xminTo need normalized data
In minimum value, xmaxTo need the maximum in normalized data；
(3) double-deck extreme learning machine is trained to obtain final forecast model using training set：
1) 4000 groups of data are chosen, data are arranged sequentially in time, according to 500 groups of data as a data block point
Block, split data into 8 data blocks and serial number；
2) train double-deck extreme learning machine to obtain model M 1 using data block 1, first input data block 1 in training process
First layer prediction result is obtained to first layer extreme learning machine, then using data block 1 and first layer prediction result simultaneously as defeated
Enter parameter and be input in second layer extreme learning machine to obtain model M 1, basic step such as Fig. 2 of double-deck extreme learning machine structure
It is shown：
I) input parameter of double-deck extreme learning machine network is initialized, and selects the activation primitive g (x) of hidden layer；Input
Parameter includes the input number of plies, node in hidden layer and the output number of plies, wherein the input number of plies is the number of input data species, output
The number of plies is 1, and the initialization formula of node in hidden layer is as follows：
In formula,For the node in hidden layer of kth layer extreme learning machine,For the defeated of kth layer extreme learning machine
Enter node layer number,For the output layer nodes of kth layer extreme learning machine；
Activation primitive g (x) is as shown in figure 4, its equation is as follows：
In formula, x is independent variable, and given N group samples { (X is assumed in extreme learning machine networks, ts), s=1 ... N, then
Here x refers specifically to W in extreme learning machinev·Xs+bv, WvConnection weight between input layer and hidden layer, bvFor threshold value,XsFor input vector, e is natural constant；
Ii random initializtion bilayer extreme learning machine network inputs) are distinguished according to input layer number and node in hidden layer
Connection weight W between layer and hidden layervWith threshold value bv, initialisation range is [- 0.5,0.5]；
Iii hidden layer output matrix H) is calculated according to input weights and threshold value, activation primitive g (x), is specially：Assuming that
N group samples { (X is given in extreme learning machine networks, ts), s=1 ... N, hidden layer node areThen extreme learning machine network
Structure is as follows：
In formula, Xs=[Xs1, Xs2..., Xsn]T∈Rn, ts=[ts1, ts2..., tsm]T∈Rm, XsAnd tsRepresent respectively defeated
Enter variable and corresponding output variable, Wv=[Wv1, Wv2..., Wvn]TIt is between connection v hidden layer nodes and n input layer
Input weight vector, bvIt is the threshold vector of v hidden layers, βv=[βv1, βv2..., βvm]TIt is connection v hidden layer nodes
With the weight vector between m output layer, Wv·XsInner product is sought in expression；
Above-mentioned equation transform obtains：
H β=T；
In formula, H is hidden layer output matrix, and β is output weight matrix, and T is it is expected output matrix, and the element in T is served as reasons
FAO-56PM equations are multiplied by the water demand of crop that crop coefficient is calculated, wherein：
Because the input weights between input layer and hidden layer and threshold value are determination value, then WvAnd bvValue determination, Ke Yiji
Calculate hidden layer output matrix H；
Iv hidden layer output matrix H, calculating limit) are calculated according to the activation primitive g (x) of extreme learning machine network hidden layer
Learning machine output weight matrix β, the formula of the calculating limit learning machine output weight matrix β are as follows：
In formula,The Moore-Penrose generalized inverses of matrix H are asked in expression, and λ is regularization coefficient, and I is unit matrix；
3) the input model M1 of data block 2 is obtained predicting irrigation requirement and calculates precision of prediction piDetermined with prediction result
Coefficients R2, its calculating process is as shown in figure 3, wherein
In formula, ETPM56(i) be by FAO-56PM equations to the calculated value of the i-th particle water demand of crop multiplied by with it is corresponding,
The obtained crop irrigation requirement of crop coefficient, unit mmday-1, ETpredicted(i) for extreme learning machine to i-th
The training predicted value of the sub- water demand of crop, unit mmday-1, N is the group number of input sample；
Crop irrigation requirement ETPM56Calculation formula it is as follows：
In formula, Δ is the function relation curve slope of saturation vapour pressure-temperature, and unit is KPa DEG C-1, RnIt is preced with for input
The net radiation of layer, unit MJm-2day-1, G is soil heat flux, is ignored here, unit MJm-2day-1, esFor saturation
Vapour pressure, unit KPa, eaFor actual water vapor pressure, unit KPa, γ represent thermometer constant, and unit is KPa DEG C-1, Q is
Daily mean temperature, unit for DEG C, u is the wind speed of 2 meters of eminences, unit ms-1；Crop is crop coefficient；
The calculation formula of Δ is as follows：
Coefficient of determination R2Closer to 1, precision of prediction is higher, coefficient of determination R2Calculation formula be：
4) precision of prediction p is judgediWhether setting precision of prediction E1 (0.93) is more than or equal to, if it is, output model M1
As integrated predictive model C1, otherwise, into next step；
5) according to prediction result coefficient of determination R2The node in hidden layer of double-deck extreme learning machine is updated, wherein
In formula, c=5, ε=0.01；
6) increased according to data block 2 using the output weight matrix of incremental learning new mechanism bilayer extreme learning machine
Amount formula bilayer extreme learning machine, the output weight matrix β of increment type bilayer extreme learning machine when j-th of data block inputsjIt is as follows：
In formula, j >=2, HjThe hidden layer output matrix of increment type bilayer extreme learning machine, T when being inputted for j-th of data blockj
The desired output matrix of increment type bilayer extreme learning machine when being inputted for j-th of data block, as j=2, βj-1For the 1st data
The output weight matrix of bilayer extreme learning machine, H when block inputs1Bilayer extreme learning machine is hidden when being inputted for the 1st data block
Output matrix containing layer, T1The desired output matrix of bilayer extreme learning machine when being inputted for the 1st data block；
7) increment type bilayer extreme learning machine is trained to obtain model M 2 using data block 2, model M 2 is integrated prediction mould
Type C1, the same step 2 of training step；
8) data block 3 is inputted into integrated predictive model C1 to obtain prediction irrigation requirement and calculate precision of prediction and predict to tie
It is decisive and resolute to determine coefficients R2, the same step 3) of calculation procedure；
9) circulation step 3)~7) after the input of data block 8, obtained integrated predictive model C7 is final prediction mould
Type；
(4) five groups of test datas are gathered, every group of test data includes soil environment data, meteorological data and irrigated crop
Crop coefficient；Final forecast model C7 is input to after carrying out the normalized as shown in step (2) to every group of test data, is obtained
Final prediction irrigation requirement, is irrigated according to prediction irrigation requirement, predicts irrigation requirement and its mistake with actual value
Difference as shown in Figure 5 and Figure 6, its prediction index coefficient of determination R2=0.98244.
Comparative example 1
A kind of irrigation method based on standard limit learning machine, this example data are same as Example 1, and final gained prediction fills
Irrigate water requirement and its with the error of actual value as shown in Figure 5 and Figure 6, its prediction index coefficient of determination R2=0.9473, by Fig. 5 and
The dynamic multilayer extreme learning machine that Fig. 6 can be seen that the present invention can be calculated more accurately relative to standard limit learning machine
Irrigation requirement, realize Rational Irrigation.
Embodiment 2
A kind of irrigation method based on dynamic multilayer extreme learning machine, its method is substantially the same manner as Example 1, difference
It is three layers of extreme learning machine to be its multilayer extreme learning machine, totally 5000 groups of the data of selection, is divided into 10 data blocks, final meter
Calculate the prediction index coefficient of determination R of gained prediction irrigation requirement and actual demand value2=0.97405, the prediction result is all
It is that program runs 20 average values taken.
Claims (9)
1. a kind of irrigation method based on dynamic multilayer extreme learning machine, it is characterized in that, collection is related to known irrigation requirement
Data and training multilayer extreme learning machine obtains final forecast model after being normalized, will be with irrigation requirement to be asked
Final forecast model is input to after related data normalization processing to obtain being irrigated after predicting irrigation requirement, wherein, it is more
In layer extreme learning machine last layer input and output simultaneously as next layer of input, dynamic multilayer extreme learning machine refers to more
Node in hidden layer dynamic updates layer extreme learning machine in the training process, and step is as follows：
(1) multigroup training data is gathered, every group of training data includes soil environment data, meteorological data and the crop of irrigated crop
Coefficient；
(2) every group of training data is normalized, all groups of training data composing training collection；
(3) final forecast model is obtained using training set training multilayer extreme learning machine；
(4) multigroup test data is gathered, every group of test data includes soil environment data, meteorological data and the crop of irrigated crop
Coefficient；
(5) final forecast model is input to after every group of test data being normalized, obtains predicting irrigation requirement, root
It is predicted that irrigation requirement is irrigated；
It is described that comprising the following steps that for multilayer extreme learning machine is trained using training set：
1) all groups of training datas are divided into by acquisition time order by n data block according to sliding window size and order is compiled
Number, the unit of sliding window is group；
2) multilayer extreme learning machine is trained to obtain model M 1 using data block 1；
3) j=2 is made；
4) data block j input integrated predictive model C (j-2) are obtained predicting irrigation requirement and calculates precision of prediction piAnd prediction
As a result coefficient of determination R2, integrated predictive model C (j-2) refers to the obtained model of data block (j-1) training, integrated predictive model C0
It is model M 1；
5) precision of prediction p is judgediWhether setting precision of prediction E1 is more than or equal to, if it is, output model M (j-1) is integrated
Forecast model C (j-1), otherwise, into next step；
6) according to prediction result coefficient of determination R2Update the node in hidden layer of multilayer extreme learning machine；
7) increment type is obtained using the output weight matrix of incremental learning new mechanism multilayer extreme learning machine according to data block j
Multilayer extreme learning machine；
8) model M j is obtained using data block j training increment type multilayer extreme learning machines, model M j is integrated predictive model C
(j-1)；
9) j=j+1 is made；
10) circulation step 4)~8) to j=n, obtained integrated predictive model C (n-1) is final forecast model.
2. a kind of irrigation method based on dynamic multilayer extreme learning machine according to claim 1, it is characterised in that described
Soil environment data and meteorological data are specially：The per day humidity of soil mean daily temperature, soil, air mean daily temperature, sky
The per day humidity of gas, per day total solar radiation, the wind speed and atmospheric pressure of 2m eminences, the unit of temperature for DEG C, the list of humidity
Position is %, and the unit of the per day total solar radiation is MJm-2day-1, the unit of the wind speed is ms-1, the pressure
Unit is KPa；The crop coefficient refers to that inquire about irrigated crop type provides each growth period according to different growing stages expert
Empirical value.
A kind of 3. irrigation method based on dynamic multilayer extreme learning machine according to claim 1, it is characterised in that step
(2) and the normalized in step (5) refers to, by all data normalizations of collection to [- 1,1], it is as follows to normalize formula：
<mrow>
<mi>x</mi>
<mo>=</mo>
<mn>2</mn>
<mfrac>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>*</mo>
<mo>-</mo>
<msub>
<mi>x</mi>
<mi>min</mi>
</msub>
<mo>)</mo>
</mrow>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>max</mi>
</msub>
<mo>-</mo>
<msub>
<mi>x</mi>
<mi>min</mi>
</msub>
<mo>)</mo>
</mrow>
</mfrac>
<mo>-</mo>
<mn>1</mn>
<mo>;</mo>
</mrow>
In formula, X is the data after normalization, and X* is to be currently needed for normalized data, xminTo need in normalized data
Minimum value, xmaxTo need the maximum in normalized data.
4. a kind of irrigation method based on dynamic multilayer extreme learning machine according to claim 1, it is characterised in that described
Contain 4000~5000 groups of training datas in training set, the sliding window size is 500 groups.
5. a kind of irrigation method based on dynamic multilayer extreme learning machine according to claim 1, it is characterised in that described
Multilayer extreme learning machine is double-deck extreme learning machine.
6. a kind of irrigation method based on dynamic multilayer extreme learning machine according to claim 1, it is characterised in that described
The construction step of multilayer extreme learning machine is as follows：
(1) input parameter of multilayer extreme learning machine network is initialized, and selects the activation primitive g (x) of hidden layer；The input
Parameter includes the input number of plies, node in hidden layer and the output number of plies, wherein the input number of plies is the number of input data species, output
The number of plies is 1, and the initialization formula of node in hidden layer is as follows：
<mrow>
<msubsup>
<mi>N</mi>
<mi>k</mi>
<mrow>
<mi>h</mi>
<mi>i</mi>
<mi>d</mi>
<mi>d</mi>
<mi>e</mi>
<mi>n</mi>
</mrow>
</msubsup>
<mo>=</mo>
<mn>2</mn>
<mo>*</mo>
<mrow>
<mo>(</mo>
<msubsup>
<mi>N</mi>
<mi>k</mi>
<mrow>
<mi>i</mi>
<mi>n</mi>
</mrow>
</msubsup>
<mo>+</mo>
<msubsup>
<mi>N</mi>
<mi>k</mi>
<mrow>
<mi>o</mi>
<mi>u</mi>
<mi>t</mi>
</mrow>
</msubsup>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
<mo>,</mo>
<mn>...</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
In formula,For the node in hidden layer of kth layer extreme learning machine,For the input layer section of kth layer extreme learning machine
Points,For the output layer nodes of kth layer extreme learning machine；
The equation of the activation primitive g (x) is as follows：
<mrow>
<mi>g</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>1</mn>
<mo>+</mo>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<mi>x</mi>
</mrow>
</msup>
</mrow>
</mfrac>
<mo>;</mo>
</mrow>
In formula, x is independent variable, and given N group samples { (X is assumed in extreme learning machine networks, ts), s=1 ... N, it is assumed that base
The node in hidden layer of this extreme learning machine isThen x here refers specifically to W in extreme learning machinev·Xs+bv, WvFor input
Connection weight between layer and hidden layer, bvFor threshold value, v=1,2......XsFor input vector, e is natural constant；
(2) according to input layer number and node in hidden layer distinguish random initializtion multilayer extreme learning machine network input layer with
Connection weight W between hidden layervWith threshold value bv, initialisation range is [- 0.5,0.5]；
(3) according to the connection weight W between input layer and hidden layerv, threshold value bvHidden layer output square is calculated with activation primitive g (x)
Battle array H, hidden layer output matrix H expression formula are as follows：
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<mi>H</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>W</mi>
<mn>1</mn>
</msub>
<mo>,</mo>
<mn>...</mn>
<mo>,</mo>
<msub>
<mi>W</mi>
<mover>
<mi>N</mi>
<mo>~</mo>
</mover>
</msub>
<mo>,</mo>
<msub>
<mi>b</mi>
<mn>1</mn>
</msub>
<mo>,</mo>
<mn>...</mn>
<mo>,</mo>
<msub>
<mi>b</mi>
<mover>
<mi>N</mi>
<mo>~</mo>
</mover>
</msub>
<mo>,</mo>
<msub>
<mi>X</mi>
<mn>1</mn>
</msub>
<mo>,</mo>
<mn>...</mn>
<mo>,</mo>
<msub>
<mi>X</mi>
<mi>N</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>=</mo>
<msub>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<mi>g</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>W</mi>
<mn>1</mn>
</msub>
<mo>&CenterDot;</mo>
<msub>
<mi>X</mi>
<mn>1</mn>
</msub>
<mo>+</mo>
<msub>
<mi>b</mi>
<mn>1</mn>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<mrow>
<mi>g</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>W</mi>
<mover>
<mi>N</mi>
<mo>~</mo>
</mover>
</msub>
<mo>&CenterDot;</mo>
<msub>
<mi>X</mi>
<mn>1</mn>
</msub>
<mo>+</mo>
<msub>
<mi>b</mi>
<mover>
<mi>N</mi>
<mo>~</mo>
</mover>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mrow></mrow>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mrow></mrow>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>g</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>W</mi>
<mn>1</mn>
</msub>
<mo>&CenterDot;</mo>
<msub>
<mi>X</mi>
<mi>N</mi>
</msub>
<mo>+</mo>
<msub>
<mi>b</mi>
<mn>1</mn>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<mrow>
<mi>g</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>W</mi>
<mover>
<mi>N</mi>
<mo>~</mo>
</mover>
</msub>
<mo>&CenterDot;</mo>
<msub>
<mi>X</mi>
<mi>N</mi>
</msub>
<mo>+</mo>
<msub>
<mi>b</mi>
<mover>
<mi>N</mi>
<mo>~</mo>
</mover>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mrow>
<mi>N</mi>
<mo>*</mo>
<mover>
<mi>N</mi>
<mo>~</mo>
</mover>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>;</mo>
</mrow>
(4) multilayer extreme learning machine output weight matrix β is calculated according to hidden layer output matrix H, formula is as follows：
In formula,The Moore-Penrose generalized inverses of matrix H are asked in expression, and λ is regularization coefficient, and I is unit matrix, and T is expectation
Output matrix, the element in T are to be multiplied by the water demand of crop that crop coefficient is calculated by FAO-56PM equations.
7. a kind of irrigation method based on dynamic multilayer extreme learning machine according to claim 1, it is characterised in that described
Precision of prediction piCalculation formula it is as follows：
<mrow>
<msub>
<mi>p</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<msqrt>
<mfrac>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>ET</mi>
<mrow>
<mi>P</mi>
<mi>M</mi>
<mn>56</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
<mo>-</mo>
<msub>
<mi>ET</mi>
<mrow>
<mi>p</mi>
<mi>r</mi>
<mi>e</mi>
<mi>d</mi>
<mi>i</mi>
<mi>c</mi>
<mi>t</mi>
<mi>e</mi>
<mi>d</mi>
</mrow>
</msub>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
<mi>N</mi>
</mfrac>
</msqrt>
<mo>;</mo>
</mrow>
In formula, ETPM56(i) it is multiplied by make accordingly by FAO-56PM equations to the calculated value of i-th group of data water demand of crop
The water demand of crop that thing coefficient obtains, unit mmday-1, ETpredicted(i) prediction for i-th group of data crop is irrigated and needed
Water, unit mmday-1, N is the group number of input sample；
The span for setting precision of prediction E1 is more than 0.90；
The coefficient of determination R2Calculation formula it is as follows：
<mrow>
<msup>
<mi>R</mi>
<mn>2</mn>
</msup>
<mo>=</mo>
<mn>1</mn>
<mo>-</mo>
<mfrac>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>ET</mi>
<mrow>
<mi>p</mi>
<mi>r</mi>
<mi>e</mi>
<mi>d</mi>
<mi>i</mi>
<mi>c</mi>
<mi>t</mi>
<mi>e</mi>
<mi>d</mi>
</mrow>
</msub>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
<mo>-</mo>
<msub>
<mi>ET</mi>
<mrow>
<mi>P</mi>
<mi>M</mi>
<mn>56</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>ET</mi>
<mrow>
<mi>P</mi>
<mi>M</mi>
<mn>56</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
<mo>-</mo>
<msub>
<mi>meanET</mi>
<mrow>
<mi>p</mi>
<mi>r</mi>
<mi>e</mi>
<mi>d</mi>
<mi>i</mi>
<mi>c</mi>
<mi>t</mi>
<mi>e</mi>
<mi>d</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mo>;</mo>
</mrow>
In formula, meanETpredictedTo seek the average of N group data prediction values；
The ETPM56(i) calculation formula is as follows：
<mrow>
<msub>
<mi>ET</mi>
<mrow>
<mi>P</mi>
<mi>M</mi>
<mi>F</mi>
<mn>56</mn>
</mrow>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<mn>0.408</mn>
<mi>&Delta;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>R</mi>
<mi>n</mi>
</msub>
<mo>-</mo>
<mi>G</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>&gamma;</mi>
<mfrac>
<mn>900</mn>
<mrow>
<mi>Q</mi>
<mo>+</mo>
<mn>273</mn>
</mrow>
</mfrac>
<mi>u</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>e</mi>
<mi>s</mi>
</msub>
<mo>-</mo>
<msub>
<mi>e</mi>
<mi>a</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>&Delta;</mi>
<mo>+</mo>
<mi>&gamma;</mi>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>+</mo>
<mn>0.34</mn>
<mi>u</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>*</mo>
<mi>C</mi>
<mi>r</mi>
<mi>o</mi>
<mi>p</mi>
<mo>;</mo>
</mrow>
In formula, Δ is the function relation curve slope of saturation vapour pressure-temperature, and unit is KPa DEG C-1, RnFor input canopy
Net radiation, unit MJm-2day-1, G is soil heat flux, is ignored here, unit MJm-2day-1, esFor saturation vapour
Pressure, unit KPa, eaFor actual water vapor pressure, unit KPa, γ represent thermometer constant, and unit is KPa DEG C-1, Q puts down for day
Equal temperature, unit for DEG C, u is the wind speed of 2 meters of eminences, unit ms-1, Crop is crop coefficient；The calculation formula of Δ is as follows：
<mrow>
<mi>&Delta;</mi>
<mo>=</mo>
<mfrac>
<mrow>
<mn>4.098</mn>
<mrow>
<mo>(</mo>
<mn>0.6108</mn>
<msup>
<mi>e</mi>
<mfrac>
<mrow>
<mn>17.27</mn>
<mi>Q</mi>
</mrow>
<mrow>
<mi>Q</mi>
<mo>+</mo>
<mn>237.3</mn>
</mrow>
</mfrac>
</msup>
<mo>)</mo>
</mrow>
</mrow>
<msup>
<mrow>
<mo>(</mo>
<mi>Q</mi>
<mo>+</mo>
<mn>237.3</mn>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mfrac>
<mo>.</mo>
</mrow>
8. a kind of irrigation method based on dynamic multilayer extreme learning machine according to claim 1, it is characterised in that described
According to prediction result coefficient of determination R2The node in hidden layer of renewal multilayer extreme learning machine refers to be determined according to prediction result
Number R2Update mechanism is defined, specific formula is as follows：
<mrow>
<msup>
<msub>
<mi>N</mi>
<mi>k</mi>
</msub>
<mrow>
<mi>h</mi>
<mi>i</mi>
<mi>d</mi>
<mi>d</mi>
<mi>e</mi>
<mi>n</mi>
</mrow>
</msup>
<mo>=</mo>
<msup>
<msub>
<mi>N</mi>
<mi>k</mi>
</msub>
<mrow>
<mi>h</mi>
<mi>i</mi>
<mi>d</mi>
<mi>d</mi>
<mi>e</mi>
<mi>n</mi>
</mrow>
</msup>
<mo>+</mo>
<mi>f</mi>
<mi>l</mi>
<mi>o</mi>
<mi>o</mi>
<mi>r</mi>
<mrow>
<mo>(</mo>
<mfrac>
<mi>c</mi>
<mrow>
<msup>
<mi>R</mi>
<mn>2</mn>
</msup>
<mo>+</mo>
<mi>&epsiv;</mi>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
<mo>,</mo>
<mo>...</mo>
</mrow>
In formula, c=5, ε=0.01.
A kind of 9. irrigation method based on dynamic multilayer extreme learning machine according to claim 1, it is characterised in that jth
The output weight matrix β of increment type multilayer extreme learning machine during individual data block inputjIt is as follows：
<mrow>
<msub>
<mi>&beta;</mi>
<mi>j</mi>
</msub>
<mo>=</mo>
<msub>
<mi>&beta;</mi>
<mrow>
<mi>j</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>+</mo>
<msubsup>
<mi>K</mi>
<mi>j</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<msup>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mi>H</mi>
<mn>1</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>H</mi>
<mi>j</mi>
</msub>
</mtd>
</mtr>
</mtable>
</mfenced>
<mi>T</mi>
</msup>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mi>T</mi>
<mn>1</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>=</mo>
<msub>
<mi>&beta;</mi>
<mrow>
<mi>j</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>+</mo>
<msup>
<msub>
<mi>K</mi>
<mi>j</mi>
</msub>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msup>
<msubsup>
<mi>H</mi>
<mi>j</mi>
<mi>T</mi>
</msubsup>
<mrow>
<mo>(</mo>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<msub>
<mi>H</mi>
<mi>j</mi>
</msub>
<msub>
<mi>&beta;</mi>
<mrow>
<mi>j</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
3
<mrow>
<msub>
<mi>K</mi>
<mi>j</mi>
</msub>
<mo>=</mo>
<msup>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mi>H</mi>
<mn>1</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>H</mi>
<mi>j</mi>
</msub>
</mtd>
</mtr>
</mtable>
</mfenced>
<mi>T</mi>
</msup>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mi>H</mi>
<mn>1</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>H</mi>
<mi>j</mi>
</msub>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>+</mo>
<mfrac>
<mi>I</mi>
<mi>&lambda;</mi>
</mfrac>
<mo>=</mo>
<msub>
<mi>K</mi>
<mrow>
<mi>j</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>+</mo>
<mo>+</mo>
<msubsup>
<mi>H</mi>
<mi>j</mi>
<mi>T</mi>
</msubsup>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
<mo>;</mo>
</mrow>
<mrow>
<msub>
<mi>K</mi>
<mn>1</mn>
</msub>
<mo>=</mo>
<msup>
<msub>
<mi>H</mi>
<mn>1</mn>
</msub>
<mi>T</mi>
</msup>
<msub>
<mi>H</mi>
<mn>1</mn>
</msub>
<mo>+</mo>
<mfrac>
<mi>I</mi>
<mi>&lambda;</mi>
</mfrac>
<mo>;</mo>
</mrow>
In formula, j >=2, HjThe hidden layer output matrix of increment type multilayer extreme learning machine, T when being inputted for j-th of data blockjFor
The desired output matrix of increment type multilayer extreme learning machine during j data block input, as j=2, βj-1It is defeated for the 1st data block
The output weight matrix of fashionable multilayer extreme learning machine, H1The hidden layer of multilayer extreme learning machine when being inputted for the 1st data block
Output matrix, T1The desired output matrix of multilayer extreme learning machine when being inputted for the 1st data block.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710605810.3A CN107466816B (en) | 2017-07-24 | 2017-07-24 | Irrigation method based on dynamic multilayer extreme learning machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710605810.3A CN107466816B (en) | 2017-07-24 | 2017-07-24 | Irrigation method based on dynamic multilayer extreme learning machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107466816A true CN107466816A (en) | 2017-12-15 |
| CN107466816B CN107466816B (en) | 2020-04-07 |
Family
ID=60595891
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710605810.3A Expired - Fee Related CN107466816B (en) | 2017-07-24 | 2017-07-24 | Irrigation method based on dynamic multilayer extreme learning machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107466816B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110428106A (en) * | 2019-08-05 | 2019-11-08 | 山东农业大学 | A kind of crop water requirement prediction technique based on machine learning |
| CN110999766A (en) * | 2019-12-09 | 2020-04-14 | 怀化学院 | Irrigation decision method, device, computer equipment and storage medium |
| CN111369093A (en) * | 2018-12-26 | 2020-07-03 | 天云融创数据科技（北京）有限公司 | Irrigation method and device based on machine learning |
| CN111880489A (en) * | 2020-07-07 | 2020-11-03 | 北京理工大学 | Regression scheduling method for complex manufacturing system |
| CN112100920A (en) * | 2020-09-15 | 2020-12-18 | 东南大学 | Power distribution network three-phase voltage calculation method, device, equipment and storage medium |
| CN112136667A (en) * | 2020-11-26 | 2020-12-29 | 江苏久智环境科技服务有限公司 | Intelligent sprinkling irrigation method and system based on edge machine learning |
| CN113050567A (en) * | 2021-03-17 | 2021-06-29 | 北京理工大学 | Dynamic scheduling method for intelligent manufacturing system |
| CN113497785A (en) * | 2020-03-20 | 2021-10-12 | 深信服科技股份有限公司 | Malicious encrypted flow detection method and system, storage medium and cloud server |
| CN113994868A (en) * | 2021-09-27 | 2022-02-01 | 上海易航海芯农业科技有限公司 | Automatic irrigation method and system based on plant growth period |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106570627A (en) * | 2016-11-02 | 2017-04-19 | 河海大学 | Crop irrigation water requirement calculation method on future climatic conditions |
| CN106651012A (en) * | 2016-12-02 | 2017-05-10 | 东华大学 | Crop transpiration prediction method based on improved extreme learning machine |
| CN106718695A (en) * | 2017-01-04 | 2017-05-31 | 吉林省沃特管业有限公司 | A kind of intelligent water-saving irrigates Internet of Things network control system |
-
2017
- 2017-07-24 CN CN201710605810.3A patent/CN107466816B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106570627A (en) * | 2016-11-02 | 2017-04-19 | 河海大学 | Crop irrigation water requirement calculation method on future climatic conditions |
| CN106651012A (en) * | 2016-12-02 | 2017-05-10 | 东华大学 | Crop transpiration prediction method based on improved extreme learning machine |
| CN106718695A (en) * | 2017-01-04 | 2017-05-31 | 吉林省沃特管业有限公司 | A kind of intelligent water-saving irrigates Internet of Things network control system |
Non-Patent Citations (3)
| Title |
|---|
| SHAFIKA SULTAN ABDULLAH ETC.: "Extreme Learning Machines: A new approach for prediction of reference evapotranspiration", 《JOURNAL OF HYDROLOGY》 * |
| ZHANG RUI ETC.: "Universal Approximation of Extreme Learning Machine with Adaptive Growth of Hidden Nodes", 《IEEE TRANSACTION ON NEURAL NETWORKS AND LEARNING SYSTEMS》 * |
| 赵志勇 等: "基于极限学习的深度学习算法", 《计算机工程与设计》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111369093A (en) * | 2018-12-26 | 2020-07-03 | 天云融创数据科技（北京）有限公司 | Irrigation method and device based on machine learning |
| CN111369093B (en) * | 2018-12-26 | 2023-09-29 | 天云融创数据科技（北京）有限公司 | Irrigation method and device based on machine learning |
| CN110428106A (en) * | 2019-08-05 | 2019-11-08 | 山东农业大学 | A kind of crop water requirement prediction technique based on machine learning |
| CN110999766A (en) * | 2019-12-09 | 2020-04-14 | 怀化学院 | Irrigation decision method, device, computer equipment and storage medium |
| CN113497785B (en) * | 2020-03-20 | 2023-05-12 | 深信服科技股份有限公司 | Malicious encryption traffic detection method, system, storage medium and cloud server |
| CN113497785A (en) * | 2020-03-20 | 2021-10-12 | 深信服科技股份有限公司 | Malicious encrypted flow detection method and system, storage medium and cloud server |
| CN111880489B (en) * | 2020-07-07 | 2022-12-09 | 北京理工大学 | Regression scheduling method for complex manufacturing system |
| CN111880489A (en) * | 2020-07-07 | 2020-11-03 | 北京理工大学 | Regression scheduling method for complex manufacturing system |
| CN112100920A (en) * | 2020-09-15 | 2020-12-18 | 东南大学 | Power distribution network three-phase voltage calculation method, device, equipment and storage medium |
| CN112136667B (en) * | 2020-11-26 | 2021-02-12 | 江苏久智环境科技服务有限公司 | Intelligent sprinkling irrigation method and system based on edge machine learning |
| CN112136667A (en) * | 2020-11-26 | 2020-12-29 | 江苏久智环境科技服务有限公司 | Intelligent sprinkling irrigation method and system based on edge machine learning |
| CN113050567B (en) * | 2021-03-17 | 2022-02-01 | 北京理工大学 | Dynamic scheduling method for intelligent manufacturing system |
| CN113050567A (en) * | 2021-03-17 | 2021-06-29 | 北京理工大学 | Dynamic scheduling method for intelligent manufacturing system |
| CN113994868A (en) * | 2021-09-27 | 2022-02-01 | 上海易航海芯农业科技有限公司 | Automatic irrigation method and system based on plant growth period |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107466816B (en) | 2020-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107466816A (en) | A kind of irrigation method based on dynamic multilayer extreme learning machine | |
| CN106651012A (en) | Crop transpiration prediction method based on improved extreme learning machine | |
| Yu et al. | Prediction of the temperature in a Chinese solar greenhouse based on LSSVM optimized by improved PSO | |
| CN104460582B (en) | A kind of Internet of Things intelligent irrigation fertilising control method and system based on fuzzy control | |
| Wang et al. | An adaptive fuzzy hierarchical control for maintaining solar greenhouse temperature | |
| CN103390116B (en) | Use the photovoltaic power station power generation power forecasting method of stepping mode | |
| CN110084367A (en) | A kind of Forecast of Soil Moisture Content method based on LSTM deep learning model | |
| CN104521699A (en) | Field intelligent irrigation on-line control management method | |
| CN106650784A (en) | Feature clustering comparison-based power prediction method and device for photovoltaic power station | |
| CN110084417A (en) | A kind of strawberry greenhouse environment parameter intelligent monitor system based on GRNN neural network | |
| CN102736596A (en) | Multi-scale greenhouse environment control system based on crop information fusion | |
| CN108781926A (en) | Greenhouse irrigation system based on neural network prediction and method | |
| CN101315544A (en) | Greenhouse intelligent control method | |
| CN106842923A (en) | Greenhouse multiple-factor control method for coordinating based on plant physiology and energy optimization | |
| CN110119766A (en) | A kind of multiple groups close the green pepper greenhouse temperature intelligence prior-warning device of intelligent model | |
| CN110119169A (en) | A kind of tomato greenhouse temperature intelligent early warning system based on minimum vector machine | |
| CN108762084A (en) | Irrigation system of rice field based on fuzzy control decision and method | |
| CN115530054A (en) | Irrigation control method and device, electronic equipment and storage medium | |
| CN110119086A (en) | A kind of tomato greenhouse environmental parameter intelligent monitoring device based on ANFIS neural network | |
| CN105588930A (en) | Method for measuring parameters of soil in greenhouse | |
| CN110119767A (en) | A kind of cucumber green house temperature intelligent detection device based on LVQ neural network | |
| CN108984995A (en) | A kind of ecology garden landscape design method of evaluation simulation | |
| CN107578116A (en) | Irrigating water quality Optimal Configuration Method based on Bi-objective immunity particle cluster algorithm | |
| CN110472281A (en) | A kind of data assimilation method for estimating space and time continuous earth's surface water and heat | |
| CN109063388A (en) | The micro climate architecture design addressing design method of wind environment simulation |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200407 |