CN112990585A - Hen laying rate prediction method based on LSTM-Kalman model - Google Patents
Hen laying rate prediction method based on LSTM-Kalman model Download PDFInfo
- Publication number
- CN112990585A CN112990585A CN202110299013.3A CN202110299013A CN112990585A CN 112990585 A CN112990585 A CN 112990585A CN 202110299013 A CN202110299013 A CN 202110299013A CN 112990585 A CN112990585 A CN 112990585A
- Authority
- CN
- China
- Prior art keywords
- lstm
- prediction
- laying rate
- input
- laying
- 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 29
- 241000287828 Gallus gallus Species 0.000 claims abstract description 40
- 244000144992 flock Species 0.000 claims abstract description 40
- 230000017448 oviposition Effects 0.000 claims abstract description 33
- 238000001914 filtration Methods 0.000 claims abstract description 16
- 230000036541 health Effects 0.000 claims abstract description 11
- 230000003068 static effect Effects 0.000 claims abstract description 7
- 238000013528 artificial neural network Methods 0.000 claims description 26
- 210000004027 cell Anatomy 0.000 claims description 15
- 230000000694 effects Effects 0.000 claims description 14
- 238000012360 testing method Methods 0.000 claims description 12
- 230000015654 memory Effects 0.000 claims description 11
- 210000002569 neuron Anatomy 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000007613 environmental effect Effects 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 7
- 230000003862 health status Effects 0.000 claims description 7
- 238000012937 correction Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 235000016709 nutrition Nutrition 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000000306 recurrent effect Effects 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 230000007704 transition Effects 0.000 claims description 6
- 238000010586 diagram Methods 0.000 claims description 5
- 230000037406 food intake Effects 0.000 claims description 4
- 235000012631 food intake Nutrition 0.000 claims description 4
- 238000012544 monitoring process Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 238000002474 experimental method Methods 0.000 claims description 3
- 238000005286 illumination Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 238000012847 principal component analysis method Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000000717 retained effect Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000011084 recovery Methods 0.000 claims 2
- 238000004519 manufacturing process Methods 0.000 description 6
- 230000008901 benefit Effects 0.000 description 2
- 230000008034 disappearance Effects 0.000 description 2
- 238000004880 explosion Methods 0.000 description 2
- 235000021050 feed intake Nutrition 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 230000036528 appetite Effects 0.000 description 1
- 235000019789 appetite Nutrition 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Agronomy & Crop Science (AREA)
- Animal Husbandry (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Primary Health Care (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention belongs to the technical field of egg laying rate prediction of laying hens, in particular to an LSTM-Kalman model-based egg laying rate prediction method, which aims at solving the problem that the prediction is difficult due to numerous influencing variables and time sequence of the egg laying rate in the prior art, and provides the following scheme, which comprises the following steps: s1: firstly, carrying out static prediction on the laying rate by adopting LSTM; s2: the result of the prediction output is used as the input of Kalman filtering for dynamic adjustment; s3: taking the adjustment result as a final prediction result; s4: analyzing the environment and health of the chicken flock; s5: collecting historical chicken flock egg laying data; s6: compared with historical data. According to the invention, the LSTM model is selected as the static model, and the Kalman filtering is combined as the dynamic model to predict the laying rate of the laying hens, so that the accuracy of egg laying prediction is improved.
Description
Technical Field
The invention relates to the technical field of egg laying rate prediction of laying hens, in particular to an LSTM-Kalman model-based egg laying rate prediction method.
Background
With the development of digital agriculture, for laying hen farmers, laying hen production can be reasonably arranged by accurately predicting the laying rate trend in advance, and the method plays an important role in improving the economic benefit of the farmers. Laying rate prediction is a time-series problem and also a complex non-linear problem. Factors influencing the laying rate comprise illumination, water intake, feed intake, temperature, humidity and the like, multiple collinearity exists among the factors, for example, the temperature and the humidity, the temperature and the water intake, the feed intake and the like have large correlation, and therefore the difficulty is increased for predicting the laying rate of the laying hens. In recent years, Machine Learning has been increasingly applied to the prediction of the laying rate of laying hens, and among them, Extreme Learning Machines (ELMs) and BP (Back Propagation) neural networks are widely used.
RNN (current Neural Network, RNN) is widely considered as a suitable method for capturing temporal flow and spatial evolution, and has a good prediction effect. The method can not only obtain multilayer characteristics through basic neural network model learning, but also combine low-level characteristics to form a high-level, and the performance characteristics are derived from the memory capacity of neurons. However, previous studies have shown that conventional RNNs have the disadvantages of gradient disappearance, gradient explosion, and the like, and cannot perform long-term memory. To address this problem, Hochreiter and Schmidhuber proposed the LSTM (Long Short Term Memory) model and was improved and promoted by many in later work. Compared with the traditional RNN, the LSTM network can learn a shallow nonlinear network structure, extract basic characteristics of input sample data, realize approximation of complex functions, avoid the problems of gradient disappearance and gradient explosion and realize long-term storage of information.
The prior art has the problems that the influencing variables are numerous and the laying rate is time-ordered, so that the prediction is difficult.
Disclosure of Invention
The invention aims to solve the problem that the egg laying rate is difficult to predict due to numerous influencing variables and time sequence of the egg laying rate in the prior art, and provides an egg laying rate prediction method based on an LSTM-Kalman model.
In order to achieve the purpose, the invention adopts the following technical scheme:
a laying hen egg laying rate prediction method based on an LSTM-Kalman model comprises the following steps:
s1: firstly, carrying out static prediction on the laying rate by adopting LSTM;
s2: the result of the prediction output is used as the input of Kalman filtering for dynamic adjustment;
s3: taking the adjustment result as a final prediction result;
s4: analyzing the environment and health of the chicken flock;
s5: collecting historical chicken flock egg laying data;
s6: and comparing with historical data to judge the egg laying quality.
Preferably, in S1, environmental variable data of the henhouse are acquired by the henhouse temperature and humidity sensors, and data restoration and maximum and minimum normalization preprocessing are performed on the environmental variable data; the method comprises the steps of screening out key variables influencing the laying rate of the laying hens by using a principal component analysis method, reducing the input dimensionality of a model, eliminating the correlation among the variables, and dividing a training set and a test set.
Preferably, the LSTM model parameters are initialized in S2, the training set is input, the model parameters are continuously adjusted until the expected target accuracy is obtained, and the LSTM-Kalman-based hen laying rate prediction model is constructed in combination with Kalman filtering:
the LSTM neural network replaces hidden layer neurons of a general recurrent neural network with a special LSTM recurrent neural network structure, and the figure is a schematic diagram of a memory block of the LSTM neural network.
Wherein x istFor the current input, ht、ht-1Respectively the current output and the output at the previous moment, Ct、Ct-1Respectively, the memory content at the current time and the memory content at the previous time.
The internal calculation of the LSTM neural network is as follows:
first, the forgetting gate functions to keep history information, which determines which information is discarded from the LSTM unit state and which information is retained, the output of which can be expressed as:
ft＝σ(ωfht-1+Ufxt+bf)
\*MERGEFORMAT(1)
where σ () is a sigmoid function. Omegaf、Uf、bfThe weights and bias vector representing the linear relationship.
Secondly, the input of the LSTM input gate consists of three parts, namely the output vector of the input layer neuron, the output vector of the previous layer hidden layer unit and the reserved information of the previous time unit. The input gate comprises two parts: a sigmoid layer of an input gate is used as information needing to be updated, and a tanh layer creates a new post-addressing vector CtAnd added to the cell state, multiplying the new state of the cell by the old state Ct-1And ftTo obtain forgotten information, add
To effect the update.
The input gate variables are defined as:
it＝σ(ωiht-1+Uixt+bi)
\*MERGEFORMAT(2)
third, the state of the LSTM cell is updated by the output gate. The input of the output gate consists of three parts, namely an output vector of an input layer neuron, an output vector of a previous hidden layer unit and reserved information of a current time unit. And determining which parts of the current cell states need to be output by the sigmoid layer, processing the cell states by the tanh layer to obtain a value between-1 and 1, and multiplying the value by the sigmoid output to obtain an output result, specifically represented by formulas (5) and (6).
Ot＝σ(ωoht-1+Uoxt+bo)
\*MERGEFORMAT(5)
ht＝Ottanh(Ct) \*MERGEFORMAT(6)
Under the influence of environmental factors such as temperature, humidity and the like, the LSTM neural network has certain uncertainty on the prediction result of the laying rate, the influence of the uncertainty can be reduced through Kalman filtering, and the prediction precision of the laying rate is improved.
The Kalman filtering mainly comprises two processes of state variable estimation and state variable correction.
And (3) state variable estimation:
and predicting the value of the k moment according to the real value of the k-1 moment and the controllable input of the system.
X′k＝AXk-1+BUk+ωk
\*MERGEFORMAT(7)
Wherein, Xk-1Representing the true value at the previous time instant, in this context the predicted value of the LSTM neural network. X'kAs an estimate, UkIs the control output of the system, set to 1 herein. A is the state transition matrix from time k-1 to time k and B represents the conversion factor between the control input and the state quantity. In this experiment, a ═ I and B ═ I (I is an identity matrix) were set. OmegakIs white gaussian noise with a mean value of 0.
And (3) state variable correction:
the difference between the observed value and the predicted value at the time k is used to correct the predicted value at the time k.
Zk＝HXK+vk \*MERGEFORMAT(8)
X＇k＝X＇k+K(Z(k)-HX＇k)
\*MERGEFORMAT(9)
Wherein, XkIs the true value at that time, i.e., the predicted value of the LSTM neural network, Z (k) is the observed value at time k, vkIs the measurement noise with the mean value of 0 and the gaussian distribution, H is the state transition matrix, H is set to I in this test, K is the gain factor of the system,
is the prediction result of the laying rate finally obtained.
Preferably, the test set is adopted in the S3 to test the prediction performance of the LSTM-Kalman model, and the comparison and analysis are performed with other traditional prediction models to realize the accurate prediction of the laying rate of the laying hens.
Preferably, in S4, the temperature, humidity, and illuminance of the living environment of the chicken flocks are monitored at any time, and data restoration and maximum and minimum normalization preprocessing are performed on the environment, so as to monitor the health status of the chicken flocks at any time.
Preferably, the method for monitoring and judging the health state of the chicken flocks comprises the following steps: firstly, judging by calculating the food intake of the chicken flocks; secondly, the judgment is carried out by observing the activity degree of the chicken flocks.
Preferably, in S5, the egg production quantity, quality and nutritional value of the chicken flocks in different living environments and different health states are collected, and the collected data are stored in a database.
Preferably, in the step S6, the living environment and health status of the chicken flocks are compared with historical data in a database to predict the egg production quantity, quality and nutritional value of the chicken flocks.
Compared with the prior art, the invention has the advantages that:
according to the invention, the static prediction result can be dynamically adjusted according to historical data and real-time data through Kalman filtering so as to obtain higher prediction precision, an LSTM model is selected as a static model, and the egg laying rate of the laying hen is predicted by combining the Kalman filtering as a dynamic model, so that the egg laying prediction accuracy is improved.
Drawings
FIG. 1 is a schematic diagram of an LSTM neural network memory block according to the present invention;
FIG. 2 is a diagram of the egg production prediction process based on the LSTM-Kalman model proposed by the present invention;
FIG. 3 is a flow chart of the egg laying rate prediction method based on the LSTM-Kalman model according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.
Referring to fig. 1-3, a method for predicting an egg laying rate of a laying hen based on an LSTM-Kalman model includes the steps of:
s1: firstly, carrying out static prediction on the laying rate by adopting LSTM;
s2: the result of the prediction output is used as the input of Kalman filtering for dynamic adjustment;
s3: taking the adjustment result as a final prediction result;
s4: analyzing the environment and health of the chicken flock;
s5: collecting historical chicken flock egg laying data;
s6: and comparing with historical data to judge the egg laying quality.
In this embodiment, in S1, environmental variable data of the henhouse is obtained by the henhouse temperature and humidity sensors, and data restoration and maximum and minimum normalization preprocessing are performed on the environmental variable data; the method comprises the steps of screening out key variables influencing the laying rate of the laying hens by using a principal component analysis method, reducing the input dimensionality of a model, eliminating the correlation among the variables, and dividing a training set and a test set.
In this embodiment, the LSTM model parameters are initialized in S2, the training set is input, the model parameters are continuously adjusted until the expected target accuracy is obtained, and an LSTM-Kalman-based egg laying rate prediction model is constructed in combination with Kalman filtering:
the LSTM neural network replaces hidden layer neurons of a general recurrent neural network with a special LSTM recurrent neural network structure, and the figure is a schematic diagram of a memory block of the LSTM neural network.
Wherein x istFor the current input, ht、ht-1Respectively the current output and the output at the previous moment, Ct、Ct-1Respectively, the memory content at the current time and the memory content at the previous time.
The internal calculation of the LSTM neural network is as follows:
first, the forgetting gate functions to keep history information, which determines which information is discarded from the LSTM unit state and which information is retained, the output of which can be expressed as:
ft＝σ(ωfht-1+Ufxt+bf)
\*MERGEFORMAT(1)
where σ () is a sigmoid function. Omegaf、Uf、bfThe weights and bias vector representing the linear relationship.
Secondly, the input of the LSTM input gate consists of three parts, namely the output vector of the input layer neuron, the output vector of the previous layer hidden layer unit and the reserved information of the previous time unit. The input gate comprises two parts: a sigmoid layer of an input gate is used as information needing to be updated, and a tanh layer creates a new post-addressing vector CtAnd added to the cell state, multiplying the new state of the cell by the old state Ct-1And ftTo obtain forgotten information, add
To effect the update.
The input gate variables are defined as:
it＝σ(ωiht-1+Uixt+bi)
\*MERGEFORMAT(2)
third, the state of the LSTM cell is updated by the output gate. The input of the output gate consists of three parts, namely an output vector of an input layer neuron, an output vector of a previous hidden layer unit and reserved information of a current time unit. And determining which parts of the current cell states need to be output by the sigmoid layer, processing the cell states by the tanh layer to obtain a value between-1 and 1, and multiplying the value by the sigmoid output to obtain an output result, specifically represented by formulas (5) and (6).
Ot＝σ(ωoht-1+Uoxt+bo)
\*MERGEFORMAT(5)
ht＝Ot tanh(Ct) \*MERGEFORMAT(6)
Under the influence of environmental factors such as temperature, humidity and the like, the LSTM neural network has certain uncertainty on the prediction result of the laying rate, the influence of the uncertainty can be reduced through Kalman filtering, and the prediction precision of the laying rate is improved.
The Kalman filtering mainly comprises two processes of state variable estimation and state variable correction.
And (3) state variable estimation:
and predicting the value of the k moment according to the real value of the k-1 moment and the controllable input of the system.
X′k＝AXk-1+BUk+ωk
\*MERGEFORMAT(7)
Wherein, Xk-1Representing the true value at the previous time instant, in this context the predicted value of the LSTM neural network. X'kAs an estimate, UkIs the control output of the system, set to 1 herein. A is the state transition matrix from time k-1 to time k and B represents the conversion factor between the control input and the state quantity. In this experiment, a ═ I and B ═ I (I is an identity matrix) were set. OmegakIs white gaussian noise with a mean value of 0.
And (3) state variable correction:
the difference between the observed value and the predicted value at the time k is used to correct the predicted value at the time k.
Zk＝HXK+vk \*MERGEFORMAT(8)
X＇k＝X＇k+K(Z(k)-HX＇k)
\*MERGEFORMAT(9)
Wherein, XkIs the true value of the instant, i.e. the LSTM nerveThe predicted value of the network, Z (k) is the observed value at time k, vkIs the measurement noise with the mean value of 0 and the gaussian distribution, H is the state transition matrix, H is set to I in this test, K is the gain factor of the system,
is the prediction result of the laying rate finally obtained.
In this embodiment, the prediction performance of the LSTM-Kalman model is tested by using the test set in S3, and the test set is compared with other conventional prediction models to perform analysis, thereby realizing accurate prediction of the laying rate of the laying hens.
In this embodiment, the temperature, the humidity, and the illuminance of the living environment of the chicken flocks are constantly monitored in S4, and data restoration and maximum and minimum normalization preprocessing are performed on the living environment, so that the health state of the chicken flocks is constantly monitored.
In this embodiment, the method for monitoring and determining the health status of a chicken flock comprises: firstly, the food intake of the chicken flocks is calculated and judged, and the food intake of the chicken flocks is reduced, which indicates that the appetite is reduced; secondly, the activity degree of the chicken flocks is observed for judgment, the activity degree of the chicken flocks directly reflects the health of the chicken flocks, and the chicken flocks can be monitored through a monitor to pay attention to the activity degree of the chicken flocks.
In this embodiment, the monitoring time period and the activity of the monitor are set, and the set time period is as follows: and during the low activity period of 20-05 and the high activity period of 05-16, judging the activity degree of the chicken flocks in a certain time period, comparing according to historical data, and reminding workers to check the health of the chicken flocks in an alarm mode if the activity degree of the chicken flocks cannot reach the historical data.
In this embodiment, in S5, the egg production quantity, quality, and nutritional value of the chicken flocks in different living environments and different health states are collected, and the collected data are stored in a database.
In this embodiment, in S6, the living environment and health status of the chicken flocks are compared with historical data in the database to predict the egg production number, quality, and nutritional value of the chicken flocks.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.
Claims (8)
1. A laying hen egg laying rate prediction method based on an LSTM-Kalman model is characterized by comprising the following steps:
s1: firstly, carrying out static prediction on the laying rate by adopting LSTM;
s2: the result of the prediction output is used as the input of Kalman filtering for dynamic adjustment;
s3: taking the adjustment result as a final prediction result;
s4: analyzing the environment and health of the chicken flock;
s5: collecting historical chicken flock egg laying data;
s6: and comparing with historical data to judge the egg laying quality.
2. The LSTM-Kalman model-based laying hen egg laying rate prediction method according to claim 1, wherein in S1, the henhouse temperature and humidity sensors acquire the henhouse environment variable data, and perform data recovery and max-min normalization preprocessing on the henhouse environment variable data; the method comprises the steps of screening out key variables influencing the laying rate of the laying hens by using a principal component analysis method, reducing the input dimensionality of a model, eliminating the correlation among the variables, and dividing a training set and a test set.
3. The LSTM-Kalman model-based laying hen egg laying rate prediction method according to claim 1, wherein the LSTM model parameters are initialized in S2, a training set is input, model parameters are continuously adjusted until an expected target accuracy is obtained, and the LSTM-Kalman model-based laying hen egg laying rate prediction model is constructed by combining Kalman filtering:
the LSTM neural network replaces hidden layer neurons of a general recurrent neural network with a special LSTM recurrent neural network structure, and the figure is a schematic diagram of a memory block of the LSTM neural network.
Wherein x istFor the current input, ht，ht-1Respectively the current output and the output at the previous moment, Ct、Ct-1Respectively, the memory content at the current time and the memory content at the previous time.
The internal calculation of the LSTM neural network is as follows:
first, the forgetting gate functions to keep history information, which determines which information is discarded from the LSTM unit state and which information is retained, the output of which can be expressed as:
ft＝σ(ωfht-1+Ufxt+bf)\*MERGEFORMAT (1)
where σ () is a sigmoid function. Omegaf、Uf、bfThe weights and bias vector representing the linear relationship.
Secondly, the input of the LSTM input gate consists of three parts, namely the output vector of the input layer neuron, the output vector of the previous layer hidden layer unit and the reserved information of the previous time unit. The input gate comprises two parts: a sigmoid layer of an input gate is used as information needing to be updated, and a tanh layer creates a new post-addressing vector CtAnd added to the cell state, multiplying the new state of the cell by the old state Ct-1And ftTo obtain forgotten information, add
To effect the update.
The input gate variables are defined as:
it＝σ(ωiht-1+Uixt+bi)\*MERGEFORMAT (2)
third, the state of the LSTM cell is updated by the output gate. The input of the output gate consists of three parts, namely an output vector of an input layer neuron, an output vector of a previous hidden layer unit and reserved information of a current time unit. And determining which parts of the current cell states need to be output by the sigmoid layer, processing the cell states by the tanh layer to obtain a value between-1 and 1, and multiplying the value by the sigmoid output to obtain an output result, specifically represented by formulas (5) and (6).
Ot＝σ(ωoht-1+Uoxt+bo)\*MERGEFORMAT (5)
ht＝Ottanh(Ct)\*MERGEFORMAT (6)
Under the influence of environmental factors such as temperature, humidity and the like, the LSTM neural network has certain uncertainty on the prediction result of the laying rate, the influence of the uncertainty can be reduced through Kalman filtering, and the prediction precision of the laying rate is improved.
The Kalman filtering mainly comprises two processes of state variable estimation and state variable correction.
And (3) state variable estimation:
and predicting the value of the k moment according to the real value of the k-1 moment and the controllable input of the system.
X′k＝AXk-1+BUk+ωk\*MERGEFORMAT (7)
Wherein, Xk-1Representing the true value at the previous time instant, in this context the predicted value of the LSTM neural network. X'kAs an estimate, UkIs the control output of the system, set to 1 herein. A is the state transition matrix from time k-1 to time k and B represents the conversion factor between the control input and the state quantity. In this experiment, a ═ I and B ═ I (I is an identity matrix) were set.
ωkIs white gaussian noise with a mean value of 0.
And (3) state variable correction:
the difference between the observed value and the predicted value at the time k is used to correct the predicted value at the time k.
Zk＝HXK+vk\*MERGEFORMAT (8)
X＇k＝X＇k+K(Z(k)-HX＇k)\*MERGEFORMAT (9)
Wherein, XkIs the true value at that time, i.e., the predicted value of the LSTM neural network, Z (k) is the observed value at time k, vkIs the measurement noise with the mean value of 0 and the gaussian distribution, H is the state transition matrix, H is set to I in this test, K is the gain factor of the system,
is the prediction result of the laying rate finally obtained.
4. The LSTM-Kalman model-based laying hen egg laying rate prediction method as claimed in claim 1, wherein the test set is adopted in S3 to test the prediction performance of the LSTM-Kalman model, and the comparison analysis is performed with other traditional prediction models to realize the accurate prediction of the egg laying rate of the laying hen.
5. The LSTM-Kalman model based laying hen egg laying rate prediction method as claimed in claim 1, wherein the temperature, humidity and illumination of the living environment of the chicken flock are monitored at time S4, and data recovery and maximum and minimum normalization preprocessing are performed on the temperature, humidity and illumination, and the health status of the chicken flock is monitored at time.
6. The LSTM-Kalman model-based laying hen laying rate prediction method as claimed in claim 5, wherein the monitoring and judgment method for the health status of the chicken flock is as follows: firstly, judging by calculating the food intake of the chicken flocks; secondly, the judgment is carried out by observing the activity degree of the chicken flocks.
7. The LSTM-Kalman model-based laying hen egg laying rate prediction method according to claim 1, wherein the egg laying quantity, quality and nutritional value of the chicken flocks in different living environments and different health states are collected in S5, and the collected data are stored in a database.
8. The LSTM-Kalman model based laying hen egg laying rate prediction method according to claim 1, characterized in that the living environment and health status of the chicken flocks are compared with historical data in a database at present in S6 to predict egg laying amount, quality and nutritional value of the chicken flocks.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110299013.3A CN112990585A (en) | 2021-03-20 | 2021-03-20 | Hen laying rate prediction method based on LSTM-Kalman model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110299013.3A CN112990585A (en) | 2021-03-20 | 2021-03-20 | Hen laying rate prediction method based on LSTM-Kalman model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112990585A true CN112990585A (en) | 2021-06-18 |
Family
ID=76334220
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110299013.3A Pending CN112990585A (en) | 2021-03-20 | 2021-03-20 | Hen laying rate prediction method based on LSTM-Kalman model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112990585A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115688972A (en) * | 2022-09-23 | 2023-02-03 | 江苏天成科技集团有限公司 | Laying hen feed intake prediction method based on ANFIS model |
| CN116579508A (en) * | 2023-07-13 | 2023-08-11 | 海煜(福州)生物科技有限公司 | Fish prediction method, device, equipment and storage medium |
| CN117192063A (en) * | 2023-11-06 | 2023-12-08 | 山东大学 | Water quality prediction method and system based on coupled Kalman filtering data assimilation |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110232169A (en) * | 2019-05-09 | 2019-09-13 | 北京航空航天大学 | Track denoising method based on two-way length memory models and Kalman filtering in short-term |
-
2021
- 2021-03-20 CN CN202110299013.3A patent/CN112990585A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110232169A (en) * | 2019-05-09 | 2019-09-13 | 北京航空航天大学 | Track denoising method based on two-way length memory models and Kalman filtering in short-term |
Non-Patent Citations (2)
| Title |
|---|
| XIJUAN SONG: "Air Quality Prediction based on LSTM-Kalman Model", 《2019 IEEE 8TH JOINT INTERNATIONAL INFORMATION TECHNOLOGY AND ARTIFICIAL INTELLIGENCE CONFERENCE (ITAIC)》, pages 695 - 699 * |
| 蒋敏兰: "基于多层LSTM 的海兰褐蛋鸡产蛋率回归模型", 《江苏农业科学》, vol. 48, no. 22, pages 248 - 251 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115688972A (en) * | 2022-09-23 | 2023-02-03 | 江苏天成科技集团有限公司 | Laying hen feed intake prediction method based on ANFIS model |
| CN116579508A (en) * | 2023-07-13 | 2023-08-11 | 海煜(福州)生物科技有限公司 | Fish prediction method, device, equipment and storage medium |
| CN116579508B (en) * | 2023-07-13 | 2023-10-13 | 海煜(福州)生物科技有限公司 | Fish prediction method, device, equipment and storage medium |
| CN117192063A (en) * | 2023-11-06 | 2023-12-08 | 山东大学 | Water quality prediction method and system based on coupled Kalman filtering data assimilation |
| CN117192063B (en) * | 2023-11-06 | 2024-03-15 | 山东大学 | Water quality prediction method and system based on coupled Kalman filtering data assimilation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109978228B (en) | PM2.5 concentration prediction method, device and medium | |
| CN112990585A (en) | Hen laying rate prediction method based on LSTM-Kalman model | |
| CN107544904B (en) | Software reliability prediction method based on deep CG-LSTM neural network | |
| Murphy et al. | Comparison of modelling techniques for milk-production forecasting | |
| Paras et al. | A feature based neural network model for weather forecasting | |
| CN111563706A (en) | Multivariable logistics freight volume prediction method based on LSTM network | |
| CN112131212A (en) | Hybrid cloud scene-oriented time sequence data anomaly prediction method based on ensemble learning technology | |
| CN110686633B (en) | Landslide displacement prediction method and device and electronic equipment | |
| CN112819136A (en) | Time sequence prediction method and system based on CNN-LSTM neural network model and ARIMA model | |
| CN110110794B (en) | Image classification method for updating neural network parameters based on feature function filtering | |
| CN115587666A (en) | Load prediction method and system based on seasonal trend decomposition and hybrid neural network | |
| CN113139605A (en) | Power load prediction method based on principal component analysis and LSTM neural network | |
| CN115456245A (en) | Prediction method for dissolved oxygen in tidal river network area | |
| CN116822920A (en) | Flow prediction method based on cyclic neural network | |
| Narejo et al. | TIME SERIES FORECASTING FOR OUTDOOR TEMPERATURE USING NONLINEAR AUTOREGRESSIVE NEURAL NETWORK MODELS. | |
| CN114862032A (en) | XGboost-LSTM-based power grid load prediction method and device | |
| Kalaiarasi et al. | Crop yield prediction using multi-parametric deep neural networks | |
| Miok | Estimation of prediction intervals in neural network-based regression models | |
| CN113255739A (en) | Fish feed detection and formula system | |
| CN116303786B (en) | Block chain financial big data management system based on multidimensional data fusion algorithm | |
| CN112381213A (en) | Industrial equipment residual life prediction method based on bidirectional long-term and short-term memory network | |
| Shah et al. | A neoteric technique using ARIMA-LSTM for time series analysis on stock market forecasting | |
| Surige et al. | IoT-based monitoring system for oyster mushroom farming | |
| CN114970745B (en) | Intelligent security and environment big data system of Internet of things | |
| CN115062764B (en) | Intelligent illuminance adjustment and environmental parameter Internet of things big data 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 |