CN112596392A - Big data based pigsty environment controller parameter automatic configuration and optimization method - Google Patents
Big data based pigsty environment controller parameter automatic configuration and optimization method Download PDFInfo
- Publication number
- CN112596392A CN112596392A CN202011553396.4A CN202011553396A CN112596392A CN 112596392 A CN112596392 A CN 112596392A CN 202011553396 A CN202011553396 A CN 202011553396A CN 112596392 A CN112596392 A CN 112596392A
- Authority
- CN
- China
- Prior art keywords
- layer
- data
- value
- output
- network
- 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
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a big data-based pigsty environment controller parameter automatic configuration and optimization method, and belongs to the technical field of breeding planning and scheduling. The invention comprises the following steps: s1: forward propagation of working signals: s11: an expert configures a database; s12: preprocessing data; s13: determining parameters of the MLP network; s14: training an MLP network; s15: a parameter configuration model; s2: back propagation of the error signal: s21: the user side related parameters; s22: preprocessing data; s23: recommended parameter configuration; s24: further adjustment by the user; s25: loading equipment to operate; s3: through repeated learning, the weight value is continuously corrected to make the actual output of the neural network approach the expected output; and automatically uploading local environment parameters according to the actual conditions of the site, and automatically downloading and synchronizing the control parameters of the environment controller issued after the optimization of the big data platform, so that the function of one-key parameter setting or intelligent parameter synchronization is realized.
Description
Technical Field
The invention relates to a big data-based pigsty environment controller parameter automatic configuration and optimization method, and belongs to the technical field of breeding planning and scheduling.
Background
Modern environmental parameters of pigsties and poultry houses are developed towards the use of a centralized environmental controller basically, and the environmental controller generally controls actuating mechanisms such as a fan and a wet curtain in the pigsties by acquiring parameters such as temperature, humidity and ammonia concentration in the pigsties to keep the temperature, the humidity and the ammonia concentration in the pigsties constant. The chicken house is actually stabilized in a range, and the precision of the stabilization of the general chicken house relative to the pigsty is higher.
However, the modern ring controller has the following main problems: (1) generally speaking, the setting parameters are more, and dozens of parameters are generally required to be set, so that farmers basically cannot operate the environment-friendly controller and set all the operation parameters; the one-key parameter setting or intelligent remote synchronous parameter setting function is not available; (2) the method has no learning function, and can not store parameter configuration process and historical parameter data set by workers with culture experience on site; (3) a few environment controllers have a networking function, but only perform the functions of parameter transmission and display, and cannot adaptively acquire or optimize the local optimal parameter control algorithm function of the environment controllers by combining a big data analysis platform; (4) the control algorithm is simple, only one parameter variable and two parameter variables are adjusted in a linkage mode, and the problems of seasons (winter and summer), altitude (different in air pressure and air oxygen content), air density and the like are not considered; (5) parameters such as air pressure, oxygen concentration, longitude and latitude, altitude and the like are not collected generally.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a method for automatically configuring and optimizing parameters of a pigsty environment controller based on big data.
The invention relates to a big data-based pigsty environment controller parameter automatic configuration and optimization method, which comprises the following steps:
s1: forward propagation of working signals: the input working signal is transmitted to the output layer from the input layer through the hidden layer, and finally, the output signal is generated, and the connection weight between the layers is fixed and unchanged in the forward transmission process of the working signal, which comprises the following steps:
s11: an expert configures a database;
s12: preprocessing data;
s13: determining parameters of a (Multi-Layer Perceptron, MLP) network;
s14: training an MLP network;
s15: a parameter configuration model;
s2: back propagation of the error signal: the difference exists between the network output signal and the expected signal, and the difference value between the two is an error signal; after the error is calculated by the cost function, the error is reversely propagated from the output layer to the input layer by layer, the connection weight between the adjusting layers is updated through error feedback, and the updated weight is used for forward propagation of the next working signal, which comprises the following steps:
s21: the user side related parameters;
s22: preprocessing data;
s23: recommended parameter configuration;
s24: further adjustment by the user;
s25: loading equipment to operate;
s3: through repeated learning, the weight value is continuously corrected to enable the actual output of the neural network to be close to the expected output; the number of the neurons of the hidden layer is determined by adopting a trial and error method, and the number of the neurons of the hidden layer is determined by establishing a plurality of MLP neural networks with the same conditions except the different number of the neurons of the hidden layer and comparing the network generation times with the error precision.
Preferably, the data preprocessing in 12 is to improve the universality of the network, and a normalization method is adopted to preprocess the original sample data, which is shown in formula (1):
after the training of the MLP neural network is finished, inverse transformation is needed to be carried out, output data are converted into data characteristics of original service area samples, and the inverse transformation is shown in an expression (2):
in the formula: xiConfiguring ith variable original data in a variable set for an expert; ximaxConfiguring ith variable raw data in variable set for expertA large value; ximinFor the ith variable raw data minimum, X, of the sample service areai' configuring ith variable data in the variable set for the transformed expert.
Preferably, in the training MLP network of S14, the model training is performed by using BP algorithm, and the forward propagation of the neural network is performed first, and then
An input value, i.e., an activation value, representing a first layer of neurons, after which the activation value for each layer is implemented by:
in the formula: first layer neuron i-node data input value xiConfiguring environment variables related to the variable set for the normalized expert;
is the output value of the ith node of the l layer;
the activation value of the ith node of the ith layer is obtained;
is a connection weight parameter between the ith node of the l layer and the jth node of the l +1 layer;
is the intercept term of the jth node of the l +1 th layer; f is the function of activation and,the hidden layer and the output layer respectively adopt tansig and logsig activation functions; the output layer configures the set of variables.
Preferably, in the parameter configuration model of S15, in the back propagation process, the error between the loss function output value and the sample value is continuously reduced by a gradient descent method, and the sum of squares of errors of all output layer nodes of the network is taken as the loss function, which is shown in formula (6):
in the formula: y isiFor the normalized sample service area output layer i node reasonable scale expected value, aiAnd outputting the I node reasonable-scale output value of the layer for configuring the variable set.
Preferably, in the further adjustment of the user in S24, the optimization goal is to determine the weight W and the bias b so that the loss function C (W, b) is minimized, i.e. the value of the network output will be closer to the true value;
the iterative formula for W and b is as follows:
in the formula: alpha is learning rate, and (0,1) is taken;
training of the MLP model is achieved by adopting a Python programming, and the prediction effect of the model is evaluated by using an absolute relative error rerr, an average absolute relative error arerr and an equal coefficient EC;
in the formula: y iskConfiguring the actual value of the variable set; a iskPredicting values for configuration variable sets; EC is equal coefficient, the value range is (0,1), the fitting degree between the predicted value and the actual value of the configuration variable set is represented, the more the EC value is greater than 0.9, and the fitting result is ideal.
Preferably, in the relevant parameters of the user side of S21, the parameter configuration set by the staff with the field culture experience is recorded in correspondence with the measured environmental parameter data under the configuration.
Preferably, in the relevant parameters of the user side of S21, measured controlled environment parameter data recorded at a certain time point is used as input of the MLP network, parameter configuration set by a worker at the time point is used as output of the MLP network for training, and the MLP network is continuously trained by using well-defined input and output data; after the MLP network is trained, when the environmental parameters of the user are given, the specific values of the set parameters are given through the trained MLP network.
Preferably, in S3, in the MLP of a single hidden layer, the number of hidden layer neurons is set to 512, Sigmoid and Softmax functions are selected as activation functions of the hidden layer and the output layer, and a Dropout layer is added thereto to prevent the overfitting phenomenon, and the neuron loss rate is set to 0.4; the network takes Tensorflow as a bottom-layer advanced deep learning link library Keras to finish training and prediction, an Adam algorithm is selected as an optimizer, and the learning rate is 0.001.
The invention has the beneficial effects that: (1) the method comprises the steps of automatically uploading local environment parameters according to the actual situation of a user site, automatically downloading and synchronizing the control parameters of the environment controller issued after being optimized by a big data platform, and achieving one-key parameter setting or intelligent parameter synchronization function (without setting parameters by farmers); (2) if a plurality of environment controllers are arranged in one house in the field, after the environment controller of one house is set, the parameters are synchronized to the cloud end, and the environment controllers of other houses can be automatically synchronized; (3) the system has a learning function, and can record all subsequent parameter modification processes on the basis of one-key remote synchronous parameters; the parameter configuration, the historical parameter data and the controlled environment data set by the workers with culture experience on site are stored locally and transmitted to a big data platform for storage, and the controlled parameter control algorithm is optimized locally by means of the parameter setting processes of the historical data, the record and the study, so that the optimal parameter control is achieved. Meanwhile, the big data platform also performs a parameter optimization algorithm of the big data platform, and is in communication synchronization with the environmental controller in real time to correct the local parameter optimization algorithm of the environmental controller; (4) the air pressure sensor, the oxygen concentration sensor and the GPS sensor are added, so that accurate breathing amount control is achieved by combining the air density and the oxygen concentration at different times with common field sensors such as field temperature and humidity according to different seasons, different longitudes, latitudes and altitudes and the like, and the method is particularly important for poultry houses.
Drawings
FIG. 1 is a block flow diagram of the present invention.
Fig. 2 is a schematic diagram of parameter configuration optimization performed by MLP.
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. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example 1:
as shown in fig. 1, the method for automatically configuring and optimizing parameters of a pigsty environment controller based on big data comprises the following steps:
s1: forward propagation of working signals: the input working signal is transmitted to the output layer from the input layer through the hidden layer, and finally, the output signal is generated, and the connection weight between the layers is fixed and unchanged in the forward transmission process of the working signal, which comprises the following steps:
s11: an expert configures a database;
s12: preprocessing data;
s13: determining parameters of the MLP network;
s14: training an MLP network;
s15: a parameter configuration model;
s2: back propagation of the error signal: the difference exists between the network output signal and the expected signal, and the difference value between the two is an error signal; after the error is calculated by the cost function, the error is reversely propagated from the output layer to the input layer by layer, the connection weight between the adjusting layers is updated through error feedback, and the updated weight is used for forward propagation of the next working signal, which comprises the following steps:
s21: the user side related parameters;
s22: preprocessing data;
s23: recommended parameter configuration;
s24: further adjustment by the user;
s25: loading equipment to operate;
s3: through repeated learning, the weight value is continuously corrected to enable the actual output of the neural network to be close to the expected output; the number of the neurons of the hidden layer is determined by adopting a trial and error method, and the number of the neurons of the hidden layer is determined by establishing a plurality of MLP neural networks with the same conditions except the different number of the neurons of the hidden layer and comparing the network generation times with the error precision.
Specifically, in the data preprocessing in 12, in order to improve the universality of the network, the original sample data is preprocessed by using a normalization method, which is shown in formula (1):
after the training of the MLP neural network is finished, inverse transformation is needed to be carried out, output data are converted into data characteristics of original service area samples, and the inverse transformation is shown in an expression (2):
in the formula: xiConfiguring ith variable original data in a variable set for an expert; ximaxConfiguring the maximum value of the ith variable original data in the variable set for the expert; ximinIs the ith variable original data minimum value, X 'of the sample service area'iAnd configuring the ith variable data in the variable set for the transformed expert.
Specifically, in the training MLP network of S14, model training is performed by using BP algorithm, and forward propagation of neural network is performed first, and then
An input value, i.e., an activation value, representing a first layer of neurons, after which the activation value for each layer is implemented by:
in the formula: first layer neuron i-node data input value xiConfiguring environment variables related to the variable set for the normalized expert;
is the output value of the ith node of the l layer;
the activation value of the ith node of the ith layer is obtained;
is a connection weight parameter between the ith node of the l layer and the jth node of the l +1 layer;
is the intercept term of the jth node of the l +1 th layer; f is an activation function, and the hidden layer and the output layer respectively adopt tansig and logsig activation functions; the output layer configures the set of variables.
Specifically, in the parameter configuration model of S15, in the inverse propagation process, the error between the loss function output value and the sample value is continuously reduced by a gradient descent method, and the sum of squares of errors of all output layer nodes of the network is taken as the loss function, which is shown in equation (6):
in the formula: y isiFor the normalized sample service area output layer i node reasonable scale expected value, aiAnd outputting the I node reasonable-scale output value of the layer for configuring the variable set.
Specifically, in the further adjustment of the user in S24, the optimization goal is to determine the weight W and the bias b so that the loss function C (W, b) is minimized, that is, the value of the network output will be closer to the true value;
the iterative formula for W and b is as follows:
in the formula: alpha is learning rate, and (0,1) is taken;
training of the MLP model is achieved by adopting a Python programming, and the prediction effect of the model is evaluated by using an absolute relative error rerr, an average absolute relative error arerr and an equal coefficient EC;
in the formula: y iskConfiguring the actual value of the variable set; a iskPredicting values for configuration variable sets; EC is equal coefficient, the value range is (0,1), the fitting degree between the predicted value and the actual value of the configuration variable set is represented, the more the EC value is greater than 0.9, and the fitting result is ideal.
Specifically, in the relevant parameters of the user side of S21, the parameter configuration set by the staff with the on-site cultivation experience is recorded in correspondence with the measured environmental parameter data under the configuration.
Specifically, in the relevant parameters of the user side of S21, measured controlled environment parameter data recorded at a certain time point is used as input of the MLP network, parameter configuration set by a worker at the time point is used as output of the MLP network for training, and the MLP network is continuously trained by using well-defined input and output data; after the MLP network is trained, when the environmental parameters of the user are given, the specific values of the set parameters are given through the trained MLP network.
Specifically, in S3, in the MLP of a single hidden layer, the number of hidden layer neurons is set to 512, Sigmoid and Softmax functions are selected as activation functions of the hidden layer and the output layer, and a Dropout layer is added thereto to prevent the overfitting phenomenon, and the neuron loss rate is set to 0.4; the network takes Tensorflow as a bottom-layer advanced deep learning link library Keras to finish training and prediction, an Adam algorithm is selected as an optimizer, and the learning rate is 0.001.
As shown in FIG. 2, the parameter configuration optimization is performed by adopting a multi-layer perceptron network MLP, the MLP belongs to a data driving method, a large number of artificial neurons are connected with one another for calculation, data are input for modeling by adjusting weights among the neurons according to training samples, and finally the capability of solving the problem of parameter configuration and the optimization process of a farm is achieved. Different from a model-driven positioning method, the MLP method is used for directly searching features from the environmental parameter data of the farm without environment prior information, so that the problem of mismatch caused by interference of various unknown influence factors is solved.
The invention has the following functions: (1) the learning function is as follows: recording all parameters set by field personnel, recording each modification, and finding out the optimal parameter configuration by combining actual controlled parameter indexes (namely temperature, humidity, ammonia gas and the like) and the number of bred pigs and poultry (namely respiratory index); (2) the automatic configuration function is realized, and after the system is powered on, default parameters optimized according to a big data algorithm are automatically obtained from the cloud server; (3) the configuration data of all on-line controllers are obtained according to the pigsty environment controllers distributed in China, with different longitudes, latitudes, years, months, days (seasons), altitudes and heights.
It should be noted that: the pigsty environment controller is connected with the air pressure sensor, the oxygen concentration sensor and the GPS sensor. . The oxygen content of the air pressure sensor and the oxygen concentration sensor in different seasons and at different temperatures is different, so that the environmental control parameters are different; the GPS sensor can measure altitude, latitude, longitude, and time.
Field operating state data, for example: how many fans are opened, what the opening degree of each fan is, what the actual opening degree is, how the wet curtain is opened, how the small window is opened and the like; the field actual environmental data (temperature, humidity, ammonia gas and the like) and the number of the bred pigs and the bred poultry (namely the index of respiratory capacity).
The invention has the beneficial effects that: (1) the method comprises the steps of automatically uploading local environment parameters according to the actual situation of a user site, automatically downloading and synchronizing the control parameters of the environment controller issued after being optimized by a big data platform, and achieving one-key parameter setting or intelligent parameter synchronization function (without setting parameters by farmers); (2) if a plurality of environment controllers are arranged in one house in the field, after the environment controller of one house is set, the parameters are synchronized to the cloud end, and the environment controllers of other houses can be automatically synchronized; (3) the system has a learning function, and can record all subsequent parameter modification processes on the basis of one-key remote synchronous parameters; the parameter configuration, the historical parameter data and the controlled environment data set by the workers with culture experience on site are stored locally and transmitted to a big data platform for storage, and the controlled parameter control algorithm is optimized locally by means of the parameter setting processes of the historical data, the record and the study, so that the optimal parameter control is achieved. Meanwhile, the big data platform also performs a parameter optimization algorithm of the big data platform, and is in communication synchronization with the environmental controller in real time to correct the local parameter optimization algorithm of the environmental controller; (4) the air pressure sensor, the oxygen concentration sensor and the GPS sensor are added, so that accurate breathing amount control is achieved by combining the air density and the oxygen concentration at different times with common field sensors such as field temperature and humidity according to different seasons, different longitudes, latitudes and altitudes and the like, and the method is particularly important for poultry houses.
The invention can be widely applied to the culture planning and dispatching occasions.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (8)
1. A big data-based pigsty environment controller parameter automatic configuration and optimization method is characterized by comprising the following steps:
s1: forward propagation of working signals: the input working signal is transmitted to the output layer from the input layer through the hidden layer, and finally, the output signal is generated, and the connection weight between the layers is fixed and unchanged in the forward transmission process of the working signal, which comprises the following steps:
s11: an expert configures a database;
s12: preprocessing data;
s13: determining parameters of a multi-layer perceptron MLP network;
s14: training an MLP network;
s15: a parameter configuration model;
s2: back propagation of the error signal: the difference exists between the network output signal and the expected signal, and the difference value between the two is an error signal; after the error is calculated by the cost function, the error is reversely propagated from the output layer to the input layer by layer, the connection weight between the adjusting layers is updated through error feedback, and the updated weight is used for forward propagation of the next working signal, which comprises the following steps:
s21: the user side related parameters;
s22: preprocessing data;
s23: recommended parameter configuration;
s24: further adjustment by the user;
s25: loading equipment to operate;
s3: through repeated learning, the weight value is continuously corrected to enable the actual output of the neural network to be close to the expected output; the number of the neurons of the hidden layer is determined by adopting a trial and error method, and the number of the neurons of the hidden layer is determined by establishing a plurality of MLP neural networks with the same conditions except the different number of the neurons of the hidden layer and comparing the network generation times with the error precision.
2. The big data based pigsty environment controller parameter automatic configuration and optimization method according to claim 1, characterized in that the data preprocessing in 12, in order to improve the versatility of the network, adopts a normalization method to preprocess the original sample data, see formula (1):
after the training of the MLP neural network is finished, inverse transformation is needed to be carried out, output data are converted into data characteristics of original service area samples, and the inverse transformation is shown in an expression (2):
in the formula: xiConfiguring ith variable original data in a variable set for an expert; ximaxConfiguring the maximum value of the ith variable original data in the variable set for the expert; ximinFor the ith variable raw data minimum, X, of the sample service areai' configuring ith variable data in the variable set for the transformed expert.
3. The big data based pigsty environment controller parameter automatic configuration and optimization method according to claim 1, characterized in that in the training MLP network of S14, model training is performed by BP algorithm, first, forward propagation of neural network is performed, and then, the neural network is used for model training
An input value, i.e., an activation value, representing a first layer of neurons, after which the activation value for each layer is implemented by:
in the formula: first layer neuron i-node data input value xiConfiguring environment variables related to the variable set for the normalized expert;
is the output value of the ith node of the l layer;
the activation value of the ith node of the ith layer is obtained;
is a connection weight parameter between the ith node of the l layer and the jth node of the l +1 layer;
is the intercept term of the jth node of the l +1 th layer; f is an activation function, and the hidden layer and the output layer respectively adopt tansig and logsig activation functions; the output layer configures the set of variables.
4. The big-data-based pigsty environment controller parameter automatic configuration and optimization method according to claim 1, characterized in that in the parameter configuration model of S15, the error between the loss function output value and the sample value is continuously reduced by the back propagation process through a gradient descent method, and the sum of the squares of the errors of all output layer nodes of the network is taken as the loss function, see formula (6):
in the formula: y isiFor the normalized sample service area output layer i node reasonable scale expected value, aiAnd outputting the I node reasonable-scale output value of the layer for configuring the variable set.
5. The big-data based pigsty environment controller parameter automatic configuration and optimization method of claim 1, wherein in the further adjustment of the user at S24, the optimization goal is to determine the weight W and the bias b so that the loss function C (W, b) is minimal, i.e. the value of the network output will get closer to the real value;
the iterative formula for W and b is as follows:
in the formula: alpha is learning rate, and (0,1) is taken;
training of the MLP model is achieved by adopting a Python programming, and the prediction effect of the model is evaluated by using an absolute relative error rerr, an average absolute relative error arerr and an equal coefficient EC;
in the formula: y iskConfiguring the actual value of the variable set; a iskPredicting values for configuration variable sets; EC is equal coefficient, the value range is (0,1), the fitting degree between the predicted value and the actual value of the configuration variable set is represented, the more the EC value is greater than 0.9, and the fitting result is ideal.
6. The big data based pigsty environment controller parameter automatic configuration and optimization method according to claim 1, wherein in the user side related parameters of S21, the parameter configuration set by the staff with the on-site breeding experience is recorded corresponding to the measured environment parameter data under the configuration.
7. The big-data-based pigsty environment controller parameter automatic configuration and optimization method according to claim 1, characterized in that in the user-side related parameters of S21, measured controlled environment parameter data recorded at a certain time point is used as the input of the MLP network, the parameter configuration set by the staff at the time point is used as the output of the MLP network for training, and the MLP network is continuously trained by using the defined input and output data; after the MLP network is trained, when the environmental parameters of the user are given, the specific values of the set parameters are given through the trained MLP network.
8. The big-data-based pigsty environment controller parameter automatic configuration and optimization method according to claim 1, wherein in S3, MLP of single hidden layer, the number of hidden layer neurons is set to 512, Sigmoid and Softmax functions are selected as activation functions of hidden layer and output layer, and Dropout layer is added to prevent overfitting phenomenon, and the neuron loss rate is set to 0.4; the network takes Tensorflow as a bottom-layer advanced deep learning link library Keras to finish training and prediction, an Adam algorithm is selected as an optimizer, and the learning rate is 0.001.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011553396.4A CN112596392A (en) | 2020-12-24 | 2020-12-24 | Big data based pigsty environment controller parameter automatic configuration and optimization method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011553396.4A CN112596392A (en) | 2020-12-24 | 2020-12-24 | Big data based pigsty environment controller parameter automatic configuration and optimization method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112596392A true CN112596392A (en) | 2021-04-02 |
Family
ID=75202410
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011553396.4A Pending CN112596392A (en) | 2020-12-24 | 2020-12-24 | Big data based pigsty environment controller parameter automatic configuration and optimization method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112596392A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113254472A (en) * | 2021-06-17 | 2021-08-13 | 浙江大华技术股份有限公司 | Parameter configuration method, device, equipment and readable storage medium |
| CN114124693A (en) * | 2021-11-08 | 2022-03-01 | 中国联合网络通信集团有限公司 | Parameter configuration method, device and storage medium |
| CN116186942A (en) * | 2023-04-23 | 2023-05-30 | 中国航发四川燃气涡轮研究院 | Aeroengine compressor disk temperature prediction method based on multilayer perceptron |
| CN116400600A (en) * | 2023-04-23 | 2023-07-07 | 重庆市畜牧科学院 | Pig farm ventilation dynamic regulation and control system based on intelligent global optimization |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH078128A (en) * | 1993-06-25 | 1995-01-13 | Natl Fedelation Of Agricult Coop Assoc | Breeding control system for livestock and poultry |
| CN104950948A (en) * | 2015-05-21 | 2015-09-30 | 淮阴工学院 | Intelligent cowshed temperature control system |
| US20160120144A1 (en) * | 2014-10-30 | 2016-05-05 | Electronics And Telecommunications Research Institute | Livestock house management system and management method thereof |
| CN107562029A (en) * | 2017-08-29 | 2018-01-09 | 南阳华诚智能有限公司 | Internet of Things intelligently cultivates factory's control method |
| CN108507700A (en) * | 2018-03-20 | 2018-09-07 | 中国农业大学 | A kind of pig house multi-point temperature prediction technique and system |
| CN109631255A (en) * | 2018-12-10 | 2019-04-16 | 珠海格力电器股份有限公司 | A kind of air conditioning control method, device, storage medium and air-conditioning |
| CN110122379A (en) * | 2019-05-15 | 2019-08-16 | 广西壮族自治区水产科学研究院 | A kind of Tilapia mossambica high yielding culture device and method |
| CN111273550A (en) * | 2020-03-12 | 2020-06-12 | 成都英孚克斯科技有限公司 | Pig house environment intelligent control system |
| CN111709652A (en) * | 2020-06-17 | 2020-09-25 | 孙洁 | Cattle raising method based on big data |
-
2020
- 2020-12-24 CN CN202011553396.4A patent/CN112596392A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH078128A (en) * | 1993-06-25 | 1995-01-13 | Natl Fedelation Of Agricult Coop Assoc | Breeding control system for livestock and poultry |
| US20160120144A1 (en) * | 2014-10-30 | 2016-05-05 | Electronics And Telecommunications Research Institute | Livestock house management system and management method thereof |
| CN104950948A (en) * | 2015-05-21 | 2015-09-30 | 淮阴工学院 | Intelligent cowshed temperature control system |
| CN107562029A (en) * | 2017-08-29 | 2018-01-09 | 南阳华诚智能有限公司 | Internet of Things intelligently cultivates factory's control method |
| CN108507700A (en) * | 2018-03-20 | 2018-09-07 | 中国农业大学 | A kind of pig house multi-point temperature prediction technique and system |
| CN109631255A (en) * | 2018-12-10 | 2019-04-16 | 珠海格力电器股份有限公司 | A kind of air conditioning control method, device, storage medium and air-conditioning |
| CN110122379A (en) * | 2019-05-15 | 2019-08-16 | 广西壮族自治区水产科学研究院 | A kind of Tilapia mossambica high yielding culture device and method |
| CN111273550A (en) * | 2020-03-12 | 2020-06-12 | 成都英孚克斯科技有限公司 | Pig house environment intelligent control system |
| CN111709652A (en) * | 2020-06-17 | 2020-09-25 | 孙洁 | Cattle raising method based on big data |
Non-Patent Citations (3)
| Title |
|---|
| 张朝元等: "支持向量机改进的神经网络的函数逼近", 《昆明理工大学学报(理工版)》 * |
| 裴玉龙等: "基于BP神经网络的交通影响预测模型", 《哈尔滨工业大学学报》 * |
| 陈祥光等: "《人工神经网络技术及应用》", 30 September 2003 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113254472A (en) * | 2021-06-17 | 2021-08-13 | 浙江大华技术股份有限公司 | Parameter configuration method, device, equipment and readable storage medium |
| CN114124693A (en) * | 2021-11-08 | 2022-03-01 | 中国联合网络通信集团有限公司 | Parameter configuration method, device and storage medium |
| CN116186942A (en) * | 2023-04-23 | 2023-05-30 | 中国航发四川燃气涡轮研究院 | Aeroengine compressor disk temperature prediction method based on multilayer perceptron |
| CN116400600A (en) * | 2023-04-23 | 2023-07-07 | 重庆市畜牧科学院 | Pig farm ventilation dynamic regulation and control system based on intelligent global optimization |
| CN116400600B (en) * | 2023-04-23 | 2023-11-03 | 重庆市畜牧科学院 | Pig farm ventilation dynamic regulation and control system based on intelligent global optimization |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112596392A (en) | Big data based pigsty environment controller parameter automatic configuration and optimization method | |
| CN110298487B (en) | Indoor temperature prediction method for meeting personalized demands of users | |
| CN111368975A (en) | Method for predicting dynamic change of multi-environment factors in pigsty | |
| CN107085732B (en) | Cowshed environment ammonia intelligent monitoring system based on wireless sensor network | |
| CN112906735B (en) | Domestic fungus environment big data detecting system | |
| CN110109193B (en) | Intelligent eggplant greenhouse temperature detection device based on DRNN neural network | |
| CN108388291A (en) | A kind of greenhouse cluster environment regulation and control method and system | |
| CN111103220B (en) | Method and system for predicting and regulating concentration of atmospheric pollutants | |
| CN109634098A (en) | A kind of fattening house environment conditioning system and method | |
| CN111783987A (en) | Farmland reference crop evapotranspiration prediction method based on improved BP neural network | |
| CN110119767A (en) | A kind of cucumber green house temperature intelligent detection device based on LVQ neural network | |
| CN108958204A (en) | A kind of edible fungus culturing investigating method based on expert system knowledge base | |
| CN107290481B (en) | Cowshed environment ammonia intelligent detection system based on wireless sensor network | |
| CN110069032A (en) | A kind of eggplant greenhouse intelligent checking system based on wavelet neural network | |
| CN116649160B (en) | Edible fungus strain production monitoring system and monitoring method | |
| CN115220506B (en) | Multipoint source cooling and heating control system based on Internet of things terminal | |
| Challa | Crop growth models for greenhouse climate control. | |
| CN112465109A (en) | Green house controlling means based on cloud limit is in coordination | |
| CN116430920A (en) | Intelligent Internet of things adjusting method and system for livestock and poultry raising temperature | |
| CN112947648B (en) | Agricultural greenhouse environment prediction method and system | |
| CN115907366B (en) | Agricultural product growth environment optimal regulation and control method and equipment based on flamingo algorithm | |
| CN112819221A (en) | Grain condition prediction method based on IPSO-GRU network | |
| KR20210114751A (en) | Method and apparatus for estimating crop growth quantity | |
| CN112783228A (en) | Neural network-based large-scale greenhouse PD control system and method | |
| CN110268891A (en) | A kind of shutter intelligent control method based on low temperature stress |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210402 |