CN117114144A - Rice salt and alkali resistance prediction method and system based on artificial intelligence - Google Patents

Rice salt and alkali resistance prediction method and system based on artificial intelligence Download PDF

Info

Publication number
CN117114144A
CN117114144A CN202311382587.2A CN202311382587A CN117114144A CN 117114144 A CN117114144 A CN 117114144A CN 202311382587 A CN202311382587 A CN 202311382587A CN 117114144 A CN117114144 A CN 117114144A
Authority
CN
China
Prior art keywords
neural network
data
parameter
individual
svm
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
Application number
CN202311382587.2A
Other languages
Chinese (zh)
Other versions
CN117114144B (en
Inventor
赵磊
刘振斌
刘霜梅
姜程
李洪霞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Qingdao Agricultural University
Original Assignee
Qingdao Agricultural University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Qingdao Agricultural University filed Critical Qingdao Agricultural University
Priority to CN202311382587.2A priority Critical patent/CN117114144B/en
Publication of CN117114144A publication Critical patent/CN117114144A/en
Application granted granted Critical
Publication of CN117114144B publication Critical patent/CN117114144B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • 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
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/02—Agriculture; Fishing; Forestry; Mining

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Business, Economics & Management (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Marine Sciences & Fisheries (AREA)
  • Economics (AREA)
  • Agronomy & Crop Science (AREA)
  • Animal Husbandry (AREA)
  • Probability & Statistics with Applications (AREA)
  • Mining & Mineral Resources (AREA)
  • Medical Informatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a method and a system for predicting the salt and alkali resistance of rice based on artificial intelligence. The invention belongs to the technical field of crop growth information monitoring, in particular to an artificial intelligence-based rice salt and alkali resistance prediction method and system, wherein the scheme carries out parameter adjustment according to probability and random numbers in the searching process through frequency division and changing steps; introducing a radial basis function and a relaxation variable, and simultaneously considering balance of the relaxation variable and model parameters to realize regularization and optimization of the model; the inertial weight and the acceleration coefficient are adjusted by defining a nonlinear function, and the speed and the exploration capacity are balanced at different stages of the searching process so as to improve the searching precision and the convergence speed.

Description

Rice salt and alkali resistance prediction method and system based on artificial intelligence
Technical Field
The invention relates to the technical field of crop growth information monitoring, in particular to a method and a system for predicting salt and alkali resistance of rice based on artificial intelligence.
Background
The method for predicting the salt and alkali resistance of the rice based on the artificial intelligence predicts the adaptability of the rice to the salt and alkali resistance by applying an artificial intelligence technology so as to improve the yield and the quality of the rice. However, the common neural network has the problems of poor adaptability and weak model performance and accuracy; the traditional SVM model has the problems of over fitting and under fitting; the conventional search method has the problems of low search accuracy and slow convergence speed.
Disclosure of Invention
Aiming at the problems of poor adaptability and poor model performance and accuracy of a general neural network, the scheme is based on the probability of being selected of an individual, and parameter adjustment is carried out according to probability and random numbers in the searching process through frequency division and changing steps, so that the diversity and the adaptability of searching are improved, and the model performance is improved; aiming at the problems of over-fitting and under-fitting of the traditional SVM model, the method introduces a radial basis function and a relaxation variable, considers the balance of the relaxation variable and model parameters, realizes regularization and optimization of the model, and flexibly adjusts regularization parameters in different problems so as to achieve the balance between the over-fitting and the under-fitting of the model; aiming at the problems of low searching precision and low convergence speed of the traditional searching method, the scheme adjusts the inertia weight and the acceleration coefficient by defining a nonlinear function, balances the speed and the exploration capacity at different stages of the searching process, and improves the searching precision and the convergence speed; the experience of the individual and the experience of the global optimal position are comprehensively considered, so that better searching and optimization are realized.
The technical scheme adopted by the invention is as follows: the invention provides an artificial intelligence-based rice salt and alkali resistance prediction method, which comprises the following steps:
step S1: collecting data;
step S2: preprocessing data;
step S3: designing a neural network, and performing frequency division and change based on the calculated selected probability of the individual to realize optimal parameter search of initial weight and threshold of the neural network and complete the construction of the neural network;
step S4: establishing an SVM model based on a relaxation variable, realizing regularization and optimization of the model based on a radial basis function structure and the relaxation variable, and flexibly adjusting regularization parameters;
step S5: SVM parameter searching, wherein parameter searching is carried out based on the definition of nonlinear inertia weight and the definition of nonlinear acceleration coefficient, and local optimum and global optimum are judged based on fitness values;
step S6: and (5) running in real time.
Further, in step S1, the data acquisition is to collect data related to the salt and alkali resistance of the rice, including the soil salt and alkali resistance, environmental conditions and rice variety factors, and the salt and alkali resistance of the rice is used as a corresponding label.
Further, in step S2, the data preprocessing is to process and clean the collected data to eliminate abnormal values and missing values.
Further, in step S3, the designing the neural network specifically includes the following steps:
step S31: initializing a neural network, wherein an input layer of the neural network is used for receiving factor data, an output layer of the neural network is used for predicting performance, a hyperbolic tangent S-shaped function is used as a transmission function of the middle layer, neurons in a hidden layer follow a heuristic formula, and in order to obtain the optimal number of the middle layer neurons, the number of the middle layer neurons is adjusted by iteratively evaluating the difference between an actual value and a predicted value, and the neural network structure is expressed as follows:
；
wherein L is the number of nodes of an input layer, O is the number of nodes of an output layer, H is the number of nodes of an intermediate layer, and αs is the modulation number of nodes of a hidden layer neuron;
step S32: the design learning process is represented as follows:
；
；
；
；
wherein xs h Is the output of the intermediate layer, d eth Is the weighted sum of the input layer and the intermediate layer, xs i Is an input parameter of an input node, W ih Is the weight value from the input layer to the intermediate layer, b h Is the threshold of the middle layer neuron, Y i Is the output of the output layer, sigma s () Is a sigmoid function, W ho Is the node weight of the intermediate layer to the output layer, b o Is the threshold of the output layer, d eto Is the weighted sum of the middle layer and the output layer, N is the input number, i and h are indexes;
step S33: the data propagates forward, the neural network obtains the error between the predicted output and the expected output, the error passes from the output layer to the input layer, the weights and thresholds between the layers are continually corrected to minimize the deviation of the model, the error function E is expressed as follows:
；
wherein omega is nn Is the neural network weight, b nn Is the neural network threshold value and,is the predicted output, y i Is the desired output;
step S34: the neural network parameter searching comprises the following steps:
step S341: taking the initial weight and the threshold value of the neural network as search dimensions, initializing parameter positions, and taking the model accuracy based on the parameter positions as a corresponding fitness value T;
step S342: calculating the probability of the individual being selected:
；
wherein P is j Is the probability that individual j is selected, n is the number of parameter individuals, j is the index of parameter individuals;
step S343: frequency division is performed by the following formula:
；
；
wherein p is k Is the probability that individual k is selected, α k And alpha j Is the individual before frequency division, alpha k * And alpha j * Is the individual after frequency division, b r A random number from 0 to 1;
step S344: the formula used is changed as follows:
；
in the formula e 1 A random number of 0 to 1, e 2 Is a random number for adjusting step length, alpha k ' is post-change individual, alpha k Is the individual before change, alpha max Is the optimal individual, alpha max Is the worst individual, G is the current iteration number, G max Is the maximum number of iterations;
step S345: presetting a neural network threshold, and when the individual fitness value is higher than the neural network threshold, establishing a neural network model based on individual parameters; if the maximum iteration times are reached, the individual positions are reinitialized for searching; otherwise, continuing the fall search.
Further, in step S4, the building of the SVM model based on the relaxation variables specifically includes the following steps:
step S41: defining a regression function, introducing dynamic variables, and adopting a radial basis function structure, wherein the formula is as follows:
；
wherein k is i Andis Lagrangian multiplier, x is input, y is support vector selected by the model, sigma is radial basis function built-in parameter, N is input number, i is input index, b is offset;
step S42: defining an objective function Rs, introducing a relaxation variable, the formula used is as follows:
；
；
in xi i Andis a relaxation variable, C is a regularization parameter, ω s Is a parameter representing the normal vector of the feature space,is a feature vector, y i Is a true label, epsilon is an integer representing the tolerance range that the slack variable allows to exceed the interval limit, s.t is a constraint.
Further, in step S5, the SVM parameter search specifically includes the following steps:
step S51: initializing, namely initializing parameter positions based on an SVM parameter search space, and taking the performance of an SVM model based on parameters as a corresponding fitness value;
step S52: the nonlinear inertia weight ω is defined using the following formula:
；
wherein t is the current iteration number, t max Is the maximum number of iterations omega st Is the inertial weight, omega, set by the iteration end The inertial weight is set at the end of the algorithm;
step S53: the nonlinear acceleration coefficient is defined by the following formula:
；
；
wherein, c 1 And c 2 Is the acceleration coefficient, c st Is the initial value of the acceleration factor c end Is the end value of the acceleration factor;
step S54: updating the parameter position by using the following formula:
；
in the method, in the process of the invention,is the position after the parameter is updated,is the position before the update of the position,is the speed before update, gamma 1 Is an individual experience term, gamma 2 Is a global experience item that is used to determine the experience,is the optimal position of the individual particle history,is the current location of the object in question,is the global optimal position;
step S55: judging, presetting an SVM threshold, and when the individual fitness value is higher than the SVM threshold, establishing an SVM model based on individual parameters; if the maximum iteration times are reached, the individual positions are reinitialized for searching; otherwise, continuing the iterative search.
Further, in step S6, the real-time operation is to use the weighted sum of the outputs of the neural network model and the SVM model as the final prediction result, collect the data of each factor of the rice growth environment in real time, and input the data into the model to output the weighted prediction result.
The invention provides an artificial intelligence-based paddy rice salt and alkali resistance prediction system, which comprises a data acquisition module, a data preprocessing module, a neural network design module, an SVM model establishment module, an SVM parameter search module and a real-time operation module;
the data acquisition module collects data related to the salt and alkali resistance of the rice and sends the data to the data preprocessing module;
the data preprocessing module receives the data sent by the data acquisition module, processes and cleans the collected data, eliminates abnormal values and missing values, and sends the data to the neural network design module and the SVM model building module;
the neural network design module receives the data sent by the data preprocessing module, performs frequency division and change based on the calculated probability of the individual being selected, realizes optimal parameter search of initial weight and threshold of the neural network, completes the construction of the neural network, and sends the data to the real-time operation module;
the SVM model building module receives data sent by the data preprocessing module and the SVM parameter searching module, realizes regularization and optimization of the model based on a radial basis function structure and a relaxation variable, flexibly adjusts regularization parameters and sends the data to the real-time operation module;
the SVM parameter searching module performs parameter searching based on the defined nonlinear inertia weight and the defined nonlinear acceleration coefficient, judges local optimum and global optimum based on the fitness value, and sends data to the SVM model building module;
the real-time operation module receives data sent by the neural network design module and the SVM model building module, takes the output weighted summation of the neural network model and the SVM model as a final prediction result, collects the data of each factor of the rice growth environment in real time, and outputs the weighted prediction result in the input model.
By adopting the scheme, the beneficial effects obtained by the invention are as follows:
(1) Aiming at the problems of poor adaptability and poor model performance and accuracy of a general neural network, the scheme is based on the probability of being selected by an individual, and parameter adjustment is carried out according to the probability and the random number in the searching process through the steps of frequency division and change, so that the searching diversity and the adaptability are improved, and the model performance is improved.
(2) Aiming at the problems of over-fitting and under-fitting of the traditional SVM model, the method introduces a radial basis function and a relaxation variable, considers the balance of the relaxation variable and model parameters, realizes regularization and optimization of the model, and flexibly adjusts regularization parameters in different problems so as to balance the over-fitting and under-fitting of the model.
(3) Aiming at the problems of low searching precision and low convergence speed of the traditional searching method, the scheme adjusts the inertia weight and the acceleration coefficient by defining a nonlinear function, balances the speed and the exploration capacity at different stages of the searching process, and improves the searching precision and the convergence speed; the experience of the individual and the experience of the global optimal position are comprehensively considered, so that better searching and optimization are realized.
Drawings
FIG. 1 is a flow chart of an artificial intelligence-based rice salt and alkali resistance prediction method;
FIG. 2 is a schematic diagram of an artificial intelligence-based system for predicting salt and alkali resistance of rice;
FIG. 3 is a flow chart of step S3;
fig. 4 is a flow chart of step S4;
fig. 5 is a flow chart of step S5.
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention; all other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "top," "bottom," "inner," "outer," and the like indicate orientation or positional relationships based on those shown in the drawings, merely to facilitate description of the invention and simplify the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the invention.
Embodiment one, referring to fig. 1, the method for predicting salt and alkali resistance of rice based on artificial intelligence provided by the invention comprises the following steps:
step S1: collecting data;
step S2: preprocessing data;
step S3: designing a neural network, and performing frequency division and change based on the calculated selected probability of the individual to realize optimal parameter search of initial weight and threshold of the neural network and complete the construction of the neural network;
step S4: establishing an SVM model based on a relaxation variable, realizing regularization and optimization of the model based on a radial basis function structure and the relaxation variable, and flexibly adjusting regularization parameters;
step S5: SVM parameter searching, wherein parameter searching is carried out based on the definition of nonlinear inertia weight and the definition of nonlinear acceleration coefficient, and local optimum and global optimum are judged based on fitness values;
step S6: and (5) running in real time.
In the second embodiment, referring to fig. 1, the embodiment is based on the above embodiment, and in step S1, data is collected on the salt and alkali tolerance of rice, including the soil salt alkalinity, the environmental condition, and the rice variety factor, and the salt and alkali tolerance of rice is used as a corresponding label.
Embodiment III referring to FIG. 1, the embodiment is based on the above embodiment, and in step S2, the data preprocessing is to process and clean the collected data to eliminate outliers and missing values.
In a fourth embodiment, referring to fig. 1 and 3, the neural network is designed based on the above embodiment, and in step S3, the neural network specifically includes the following steps:
step S31: initializing a neural network, wherein an input layer of the neural network is used for receiving factor data, an output layer of the neural network is used for predicting performance, a hyperbolic tangent S-shaped function is used as a transmission function of the middle layer, neurons in a hidden layer follow a heuristic formula, and in order to obtain the optimal number of the middle layer neurons, the number of the middle layer neurons is adjusted by iteratively evaluating the difference between an actual value and a predicted value, and the neural network structure is expressed as follows:
；
wherein L is the number of nodes of an input layer, O is the number of nodes of an output layer, H is the number of nodes of an intermediate layer, and αs is the modulation number of nodes of a hidden layer neuron;
step S32: the design learning process is represented as follows:
；
；
；
；
wherein xs h Is the output of the intermediate layer, d eth Is the weighted sum of the input layer and the intermediate layer, xs i Is an input parameter of an input node, W ih Is the weight value from the input layer to the intermediate layer, b h Is the threshold of the middle layer neuron, Y i Is the output of the output layer, sigma s () Is a sigmoid function, W ho Is the node weight of the intermediate layer to the output layer, b o Is the threshold of the output layer, d eto Is the weighted sum of the middle layer and the output layer, N is the input number, i and h are indexes;
step S33: the data propagates forward, the neural network obtains the error between the predicted output and the expected output, the error passes from the output layer to the input layer, the weights and thresholds between the layers are continually corrected to minimize the deviation of the model, the error function E is expressed as follows:
；
wherein omega is nn Is the neural network weight, b nn Is the neural network threshold value and,is the predicted output, y i Is the desired output;
step S34: the neural network parameter searching comprises the following steps:
step S341: taking the initial weight and the threshold value of the neural network as search dimensions, initializing parameter positions, and taking the model accuracy based on the parameter positions as a corresponding fitness value T;
step S342: calculating the probability of the individual being selected:
；
wherein P is j Is the probability that individual j is selected, n is the number of parameter individuals, j is the index of parameter individuals;
step S343: frequency division is performed by the following formula:
；
；
wherein p is k Is the probability that individual k is selected, α k And alpha j Is the individual before frequency division, alpha k * And alpha j * Is the individual after frequency division, b r A random number from 0 to 1;
step S344: the formula used is changed as follows:
；
in the formula e 1 A random number of 0 to 1, e 2 Is a random number for adjusting step length, alpha k ' is post-change individual, alpha k Is the individual before change, alpha max Is the optimal individual, alpha max Is the worst individual, G is the current iteration number, G max Is the maximum number of iterations;
step S345: presetting a neural network threshold, and when the individual fitness value is higher than the neural network threshold, establishing a neural network model based on individual parameters; if the maximum iteration times are reached, the individual positions are reinitialized for searching; otherwise, continuing the fall search.
By executing the operation, the problems of poor adaptability and weak model performance and accuracy of a general neural network are solved, and the scheme is based on the probability of being selected by an individual, and parameter adjustment is performed according to the probability and the random number in the searching process through the steps of frequency division and change, so that the searching diversity and the adaptability are improved, and the model performance is improved.
Embodiment five, referring to fig. 1 and 4, based on the above embodiment, in step S4, building an SVM model based on a relaxation variable specifically includes the following steps:
step S41: defining a regression function, introducing dynamic variables, and adopting a radial basis function structure, wherein the formula is as follows:
；
wherein k is i Andis Lagrangian multiplier, x is input, y is support vector selected by the model, sigma is radial basis function built-in parameter, N is input number, i is input index, b is offset;
step S42: defining an objective function Rs, introducing a relaxation variable, the formula used is as follows:
；
；
in xi i Andis a relaxation variable, C is a regularization parameter, ω s Is a parameter representing the normal vector of the feature space,is a feature vector, y i Is a true label, epsilon is an integer representing the tolerance range that the slack variable allows to exceed the interval limit, s.t is a constraint.
By executing the operation, the problems of over-fitting and under-fitting exist for the traditional SVM model, the radial basis function and the relaxation variable are introduced, meanwhile, the balance of the relaxation variable and the model parameter is considered, regularization and optimization of the model are realized, and regularization parameters are flexibly adjusted in different problems so as to achieve the balance between the over-fitting and the under-fitting of the model.
Embodiment six, referring to fig. 1 and 5, based on the above embodiment, in step S5, the SVM parameter search specifically includes the following steps:
step S51: initializing, namely initializing parameter positions based on an SVM parameter search space, and taking the performance of an SVM model based on parameters as a corresponding fitness value;
step S52: the nonlinear inertia weight ω is defined using the following formula:
；
wherein t is the current iteration number, t max Is the maximum number of iterations omega st Is the inertial weight, omega, set by the iteration end The inertial weight is set at the end of the algorithm;
step S53: the nonlinear acceleration coefficient is defined by the following formula:
；
；
wherein, c 1 And c 2 Is the acceleration coefficient, c st Is the initial value of the acceleration factor c end Is the end value of the acceleration factor;
step S54: updating the parameter position by using the following formula:
；
in the method, in the process of the invention,is the position after the parameter is updated,is the position before the update of the position,is the speed before update, gamma 1 Is an individual experience term, gamma 2 Is a global experience item that is used to determine the experience,is the optimal position of the individual particle history,is the current location of the object in question,is the global optimal position;
step S55: judging, presetting an SVM threshold, and when the individual fitness value is higher than the SVM threshold, establishing an SVM model based on individual parameters; if the maximum iteration times are reached, the individual positions are reinitialized for searching; otherwise, continuing the iterative search.
By executing the operation, aiming at the problems of low searching precision and low convergence speed of the traditional searching method, the scheme adjusts the inertia weight and the acceleration coefficient by defining a nonlinear function, balances the speed and the exploration capacity at different stages of the searching process, and improves the searching precision and the convergence speed; the experience of the individual and the experience of the global optimal position are comprehensively considered, so that better searching and optimization are realized.
In step S6, the real-time operation is to use the weighted sum of the outputs of the neural network model and the SVM model as the final prediction result, collect the data of each factor of the rice growth environment in real time, and input the data into the model to output the weighted prediction result.
An embodiment eight, referring to fig. 2, based on the above embodiment, the system for predicting salt and alkali resistance of rice based on artificial intelligence provided by the invention includes a data acquisition module, a data preprocessing module, a neural network design module, an SVM model building module, an SVM parameter searching module and a real-time operation module;
the data acquisition module collects data related to the salt and alkali resistance of the rice and sends the data to the data preprocessing module;
the data preprocessing module receives the data sent by the data acquisition module, processes and cleans the collected data, eliminates abnormal values and missing values, and sends the data to the neural network design module and the SVM model building module;
the neural network design module receives the data sent by the data preprocessing module, performs frequency division and change based on the calculated probability of the individual being selected, realizes optimal parameter search of initial weight and threshold of the neural network, completes the construction of the neural network, and sends the data to the real-time operation module;
the SVM model building module receives data sent by the data preprocessing module and the SVM parameter searching module, realizes regularization and optimization of the model based on a radial basis function structure and a relaxation variable, flexibly adjusts regularization parameters and sends the data to the real-time operation module;
the SVM parameter searching module performs parameter searching based on the defined nonlinear inertia weight and the defined nonlinear acceleration coefficient, judges local optimum and global optimum based on the fitness value, and sends data to the SVM model building module;
the real-time operation module receives data sent by the neural network design module and the SVM model building module, takes the output weighted summation of the neural network model and the SVM model as a final prediction result, collects the data of each factor of the rice growth environment in real time, and outputs the weighted prediction result in the input model.
It is noted that relational terms such as first and second, and the like are 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. Moreover, 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 understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
The invention and its embodiments have been described above with no limitation, and the actual construction is not limited to the embodiments of the invention as shown in the drawings. In summary, if one of ordinary skill in the art is informed by this disclosure, a structural manner and an embodiment similar to the technical solution should not be creatively devised without departing from the gist of the present invention.

Claims (8)

1. The method for predicting the salt and alkali resistance of the rice based on artificial intelligence is characterized by comprising the following steps of: the method comprises the following steps:
step S1: collecting data;
step S2: preprocessing data;
step S3: designing a neural network, and performing frequency division and change based on the calculated selected probability of the individual to realize optimal parameter search of initial weight and threshold of the neural network and complete the construction of the neural network;
step S4: establishing an SVM model based on a relaxation variable, realizing regularization and optimization of the model based on a radial basis function structure and the relaxation variable, and flexibly adjusting regularization parameters;
step S5: SVM parameter searching, wherein parameter searching is carried out based on the definition of nonlinear inertia weight and the definition of nonlinear acceleration coefficient, and local optimum and global optimum are judged based on fitness values;
step S6: and (5) running in real time.
2. The artificial intelligence-based rice salt and alkali resistance prediction method according to claim 1, wherein the method comprises the following steps: in step S5, the SVM parameter search specifically includes the following steps:
step S51: initializing, namely initializing parameter positions based on an SVM parameter search space, and taking the performance of an SVM model based on parameters as a corresponding fitness value;
step S52: the nonlinear inertia weight ω is defined using the following formula:
；
wherein t is the current iteration number, t max Is the maximum number of iterations omega st Is the inertial weight, omega, set by the iteration end The inertial weight is set at the end of the algorithm;
step S53: the nonlinear acceleration coefficient is defined by the following formula:
；
；
wherein, c 1 And c 2 Is the acceleration coefficient, c st Is the initial value of the acceleration factor c end Is the end value of the acceleration factor;
step S54: updating the parameter position by using the following formula:
；
in the method, in the process of the invention,is the position after parameter update,/->Is the pre-update location,/->Is the speed before update, gamma 1 Is an individual experience term, gamma 2 Is a global experience item,/->Is the optimal position of individual particle history,/->Is the current location, +.>Is the global optimal position;
step S55: judging, presetting an SVM threshold, and when the individual fitness value is higher than the SVM threshold, establishing an SVM model based on individual parameters; if the maximum iteration times are reached, the individual positions are reinitialized for searching; otherwise, continuing the iterative search.
3. The artificial intelligence-based rice salt and alkali resistance prediction method according to claim 1, wherein the method comprises the following steps: in step S4, the building of the SVM model based on the relaxation variables specifically includes the following steps:
step S41: defining a regression function, introducing dynamic variables, and adopting a radial basis function structure, wherein the formula is as follows:
；
wherein k is i Andis Lagrangian multiplier, x is input, y is support vector selected by the model, sigma is radial basis function built-in parameter, N is input number, i is input index, b is offset;
step S42: defining an objective function Rs, introducing a relaxation variable, the formula used is as follows:
；
；
in xi i Andis a relaxation variable, C is a regularization parameter, ω s Is a parameter representing the normal vector of the feature space, < ->Is a feature vector, y i Is a true label, epsilon is an integer representing the tolerance range that the slack variable allows to exceed the interval limit, s.t is a constraint.
4. The artificial intelligence-based rice salt and alkali resistance prediction method according to claim 1, wherein the method comprises the following steps: in step S3, the designing a neural network specifically includes the following steps:
step S31: initializing a neural network, wherein an input layer of the neural network is used for receiving factor data, an output layer of the neural network is used for predicting performance, a hyperbolic tangent S-shaped function is used as a transmission function of the middle layer, neurons in a hidden layer follow a heuristic formula, and in order to obtain the optimal number of the middle layer neurons, the number of the middle layer neurons is adjusted by iteratively evaluating the difference between an actual value and a predicted value, and the neural network structure is expressed as follows:
；
wherein L is the number of nodes of an input layer, O is the number of nodes of an output layer, H is the number of nodes of an intermediate layer, and αs is the modulation number of nodes of a hidden layer neuron;
step S32: the design learning process is represented as follows:
；
；
；
；
wherein xs h Is the output of the intermediate layer, d eth Is the weighted sum of the input layer and the intermediate layer, xs i Is an input parameter of an input node, W ih Is the weight value from the input layer to the intermediate layer, b h Is the threshold of the middle layer neuron, Y i Is the output of the output layer, sigma s () Is a sigmoid function, W ho Is the node weight of the intermediate layer to the output layer, b o Is the threshold of the output layer, d eto Is the weighted sum of the middle layer and the output layer, N is the input number, i and h are indexes;
step S33: the data propagates forward, the neural network obtains the error between the predicted output and the expected output, the error passes from the output layer to the input layer, the weights and thresholds between the layers are continually corrected to minimize the deviation of the model, the error function E is expressed as follows:
；
wherein omega is nn Is the neural network weight, b nn Is the neural network threshold value and,is the predicted output, y i Is the desired output;
step S34: the neural network parameter searching comprises the following steps:
step S341: taking the initial weight and the threshold value of the neural network as search dimensions, initializing parameter positions, and taking the model accuracy based on the parameter positions as a corresponding fitness value T;
step S342: calculating the probability of the individual being selected:
；
wherein P is j Is the probability that individual j is selected, n is the number of parameter individuals, j is the index of parameter individuals;
step S343: frequency division is performed by the following formula:
；
；
wherein p is k Is the probability that individual k is selected, α k And alpha j Is the individual before frequency division, alpha k * And alpha j * Is the individual after frequency division, b r A random number from 0 to 1;
step S344: the formula used is changed as follows:
；
in the formula e 1 A random number of 0 to 1, e 2 Is a random number for adjusting step length, alpha k ' is post-change individual, alpha k Is the individual before change, alpha max Is the optimal individual, alpha max Is the worst individual, G is the current iteration number, G max Is the maximum number of iterations;
step S345: presetting a neural network threshold, and when the individual fitness value is higher than the neural network threshold, establishing a neural network model based on individual parameters; if the maximum iteration times are reached, the individual positions are reinitialized for searching; otherwise, continuing the fall search.
5. The artificial intelligence-based rice salt and alkali resistance prediction method according to claim 1, wherein the method comprises the following steps: in step S1, the data acquisition is to collect data related to the salt and alkali resistance of the rice, including soil salt and alkali resistance, environmental conditions and rice variety factors, and take the salt and alkali resistance of the rice as a corresponding label.
6. The artificial intelligence-based rice salt and alkali resistance prediction method according to claim 1, wherein the method comprises the following steps: in step S2, the data preprocessing is to process and clean the collected data to eliminate abnormal values and missing values.
7. The artificial intelligence-based rice salt and alkali resistance prediction method according to claim 1, wherein the method comprises the following steps: in step S6, the real-time operation is to output weighted summation of the neural network model and the SVM model as a final prediction result, collect the data of each factor of the rice growth environment in real time, and output the weighted prediction result in the input model.
8. An artificial intelligence-based rice salt and alkali resistance prediction system for realizing the artificial intelligence-based rice salt and alkali resistance prediction method according to any one of claims 1 to 7, which is characterized in that: the system comprises a data acquisition module, a data preprocessing module, a neural network design module, an SVM model building module, an SVM parameter searching module and a real-time operation module;
the data acquisition module collects data related to the salt and alkali resistance of the rice and sends the data to the data preprocessing module;
the data preprocessing module receives the data sent by the data acquisition module, processes and cleans the collected data, eliminates abnormal values and missing values, and sends the data to the neural network design module and the SVM model building module;
the neural network design module receives the data sent by the data preprocessing module, performs frequency division and change based on the calculated probability of the individual being selected, realizes optimal parameter search of initial weight and threshold of the neural network, completes the construction of the neural network, and sends the data to the real-time operation module;
the SVM model building module receives data sent by the data preprocessing module and the SVM parameter searching module, realizes regularization and optimization of the model based on a radial basis function structure and a relaxation variable, flexibly adjusts regularization parameters and sends the data to the real-time operation module;
the SVM parameter searching module performs parameter searching based on the defined nonlinear inertia weight and the defined nonlinear acceleration coefficient, judges local optimum and global optimum based on the fitness value, and sends data to the SVM model building module;
the real-time operation module receives data sent by the neural network design module and the SVM model building module, takes the output weighted summation of the neural network model and the SVM model as a final prediction result, collects the data of each factor of the rice growth environment in real time, and outputs the weighted prediction result in the input model.
CN202311382587.2A 2023-10-24 2023-10-24 Rice salt and alkali resistance prediction method and system based on artificial intelligence Active CN117114144B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311382587.2A CN117114144B (en) 2023-10-24 2023-10-24 Rice salt and alkali resistance prediction method and system based on artificial intelligence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311382587.2A CN117114144B (en) 2023-10-24 2023-10-24 Rice salt and alkali resistance prediction method and system based on artificial intelligence

Publications (2)

Publication Number Publication Date
CN117114144A true CN117114144A (en) 2023-11-24
CN117114144B CN117114144B (en) 2024-01-26

Family

ID=88805997

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311382587.2A Active CN117114144B (en) 2023-10-24 2023-10-24 Rice salt and alkali resistance prediction method and system based on artificial intelligence

Country Status (1)

Country Link
CN (1) CN117114144B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117421691A (en) * 2023-12-18 2024-01-19 青岛奥克生物开发有限公司 Cell culture monitoring method and system based on artificial intelligence
CN117788165A (en) * 2024-02-27 2024-03-29 薪海数字科技(烟台)有限公司 Enterprise supply chain transaction risk prediction method and system based on artificial intelligence
CN117851922A (en) * 2024-03-08 2024-04-09 山东新泽仪器有限公司 Operation monitoring method of air concentration monitoring instrument
CN117851922B (en) * 2024-03-08 2024-06-07 山东新泽仪器有限公司 Operation monitoring method of air concentration monitoring instrument

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018072351A1 (en) * 2016-10-20 2018-04-26 北京工业大学 Method for optimizing support vector machine on basis of particle swarm optimization algorithm
CN109151332A (en) * 2018-06-26 2019-01-04 杭州电子科技大学 Camera coding based on fitness function exposes optimal code word sequence search method
CN110708318A (en) * 2019-10-10 2020-01-17 国网湖北省电力有限公司电力科学研究院 Network abnormal flow prediction method based on improved radial basis function neural network algorithm
CN111262858A (en) * 2020-01-16 2020-06-09 郑州轻工业大学 Network security situation prediction method based on SA _ SOA _ BP neural network
CN112131907A (en) * 2019-06-24 2020-12-25 华为技术有限公司 Method and device for training classification model
CN113627606A (en) * 2020-05-06 2021-11-09 江苏科技大学 RBF neural network optimization method based on improved particle swarm optimization
CN114037163A (en) * 2021-11-10 2022-02-11 南京工业大学 Sewage treatment effluent quality early warning method based on dynamic weight PSO (particle swarm optimization) optimization BP (Back propagation) neural network
CN114492198A (en) * 2022-02-15 2022-05-13 重庆大学 Cutting force prediction method based on improved PSO algorithm assisted SVM algorithm
US20230035108A1 (en) * 2020-11-09 2023-02-02 Guizhou Power Grid Company Limited Photovoltaic cell parameter identification method based on improved equilibrium optimizer algorithm
CN116359858A (en) * 2022-09-13 2023-06-30 中国人民解放军战略支援部队信息工程大学 Collaborative interference resource scheduling method based on improved sparrow search algorithm
CN116740650A (en) * 2023-08-10 2023-09-12 青岛农业大学 Crop breeding monitoring method and system based on deep learning

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018072351A1 (en) * 2016-10-20 2018-04-26 北京工业大学 Method for optimizing support vector machine on basis of particle swarm optimization algorithm
CN109151332A (en) * 2018-06-26 2019-01-04 杭州电子科技大学 Camera coding based on fitness function exposes optimal code word sequence search method
CN112131907A (en) * 2019-06-24 2020-12-25 华为技术有限公司 Method and device for training classification model
CN110708318A (en) * 2019-10-10 2020-01-17 国网湖北省电力有限公司电力科学研究院 Network abnormal flow prediction method based on improved radial basis function neural network algorithm
CN111262858A (en) * 2020-01-16 2020-06-09 郑州轻工业大学 Network security situation prediction method based on SA _ SOA _ BP neural network
CN113627606A (en) * 2020-05-06 2021-11-09 江苏科技大学 RBF neural network optimization method based on improved particle swarm optimization
US20230035108A1 (en) * 2020-11-09 2023-02-02 Guizhou Power Grid Company Limited Photovoltaic cell parameter identification method based on improved equilibrium optimizer algorithm
CN114037163A (en) * 2021-11-10 2022-02-11 南京工业大学 Sewage treatment effluent quality early warning method based on dynamic weight PSO (particle swarm optimization) optimization BP (Back propagation) neural network
CN114492198A (en) * 2022-02-15 2022-05-13 重庆大学 Cutting force prediction method based on improved PSO algorithm assisted SVM algorithm
CN116359858A (en) * 2022-09-13 2023-06-30 中国人民解放军战略支援部队信息工程大学 Collaborative interference resource scheduling method based on improved sparrow search algorithm
CN116740650A (en) * 2023-08-10 2023-09-12 青岛农业大学 Crop breeding monitoring method and system based on deep learning

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
DHRUBA JYOTI KALITA等: "A dynamic framework for tuning SVM hyper parameters based on Moth-Flame Optimization and knowledge-based-search", EXPERT SYSTEMS WITH APPLICATIONS *
于湘涛等: "基于微粒群算法优化支持向量机的加速度计静态模型辨识", 中国惯性技术学报, vol. 18, no. 3 *
王云锋;刘丹;裴作飞;姚丽霜;: "基于改进引力搜索算法的SVM的参数优化及应用", 计算机应用研究, no. 1 *
胡泽民;刘致宏;: "基于BSO-BP神经网络的赤潮预测模型", 桂林航天工业学院学报, no. 01 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117421691A (en) * 2023-12-18 2024-01-19 青岛奥克生物开发有限公司 Cell culture monitoring method and system based on artificial intelligence
CN117421691B (en) * 2023-12-18 2024-03-15 青岛奥克生物开发有限公司 Cell culture monitoring method and system based on artificial intelligence
CN117788165A (en) * 2024-02-27 2024-03-29 薪海数字科技(烟台)有限公司 Enterprise supply chain transaction risk prediction method and system based on artificial intelligence
CN117788165B (en) * 2024-02-27 2024-05-14 薪海数字科技(烟台)有限公司 Enterprise supply chain transaction risk prediction method and system based on artificial intelligence
CN117851922A (en) * 2024-03-08 2024-04-09 山东新泽仪器有限公司 Operation monitoring method of air concentration monitoring instrument
CN117851922B (en) * 2024-03-08 2024-06-07 山东新泽仪器有限公司 Operation monitoring method of air concentration monitoring instrument

Also Published As

Publication number Publication date
CN117114144B (en) 2024-01-26

Similar Documents

Publication Publication Date Title
CN117114144B (en) Rice salt and alkali resistance prediction method and system based on artificial intelligence
Bai et al. Double-layer staged training echo-state networks for wind speed prediction using variational mode decomposition
CN108776683A (en) A kind of electric power operation/maintenance data cleaning method based on isolated forest algorithm and neural network
CN111563706A (en) Multivariable logistics freight volume prediction method based on LSTM network
CN111079931A (en) State space probabilistic multi-time-series prediction method based on graph neural network
EP3948692A1 (en) Process and system including an optimization engine with evolutionary surrogate-assisted prescriptions
CN113850414B (en) Logistics scheduling planning method based on graph neural network and reinforcement learning
CN114167898B (en) Global path planning method and system for collecting data of unmanned aerial vehicle
EP3502978A1 (en) Meta-learning system
CN116562514B (en) Method and system for immediately analyzing production conditions of enterprises based on neural network
CN116992779A (en) Simulation method and system of photovoltaic energy storage system based on digital twin model
CN117077041B (en) Intelligent community management method and system based on Internet of things
CN117273237B (en) Method and system for predicting capacity of automobile parts based on artificial intelligence
CN117077671B (en) Interactive data generation method and system
CN117196418B (en) Reading teaching quality assessment method and system based on artificial intelligence
CN110852435A (en) Neural evolution calculation model
CN116862021B (en) Anti-Bayesian-busy attack decentralization learning method and system based on reputation evaluation
CN113033878A (en) Landslide displacement prediction method based on multi-topology hierarchical cooperative particle swarm LSTM
CN116781139A (en) Flow prediction satellite path selection method and system based on reinforcement learning
CN110070177A (en) Community structure detection method in a kind of nonoverlapping network and overlapping network
Alkhliwi Energy efficient cluster based routing protocol with secure IDS for IoT assisted heterogeneous WSN
CN115619028A (en) Clustering algorithm fusion-based power load accurate prediction method
CN115081323A (en) Method for solving multi-objective constrained optimization problem and storage medium thereof
CN117236809B (en) Automobile part production management method and system based on artificial intelligence
CN114202063A (en) Fuzzy neural network greenhouse temperature prediction method based on genetic algorithm optimization

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