CN111723909A - Optimization method and system of fuzzy neural network model - Google Patents

Abstract

The invention relates to the field of computer technology, and discloses a method and system for optimizing a fuzzy neural network model. The method includes: acquiring parameters to be optimized of a fuzzy neural network model, and generating each population in a preset gray wolf population based on the parameters to be optimized The initial position information of the individual, calculate the first fitness value corresponding to each population individual according to the initial position information, update the initial position information, obtain the updated initial position information, and use the updated initial position information As the target location information, the second fitness value corresponding to each individual population is calculated according to the target location information, the target location information is iteratively processed based on the first fitness value and the second fitness value, and the iteratively processed target location information is determined Whether the preset iteration stop condition is met, if so, output the optimized fuzzy neural network model according to the iteratively processed target position information to reduce the model training time and improve the model convergence accuracy.

Description

Optimization method and system of fuzzy neural network model

technical field

The invention relates to the field of computer technology, in particular to a method and system for optimizing a fuzzy neural network model.

Background technique

Fuzzy theory and neural network technology are two active areas of artificial intelligence research in recent years, and fuzzy neural network is the product of combining fuzzy theory and neural network. Self-learning ability and mapping ability. The network mainly has five layers, input layer, fuzzy layer, fuzzy rule layer, fuzzy decision layer and output layer. The center point, width vector and weight of output layer of membership function in fuzzy layer can use swarm intelligence The optimization algorithm is optimized. In specific applications, according to different actual needs, different fuzzy neural network models will be established, but the center point of the membership function in the fuzzy layer, the width vector and the weight of the output layer can still be trained through the swarm intelligence optimization algorithm. In order to improve the model accuracy of the fuzzy neural network model, the existing technologies mostly optimize the fuzzy neural network model through swarm intelligence optimization algorithms such as ant colony algorithm, particle swarm algorithm, artificial bee colony algorithm, chicken swarm algorithm, etc. The Grey Wolf Optimizer optimizes the fuzzy neural network model. However, compared with the genetic algorithm and particle swarm optimization proposed earlier, the Grey Wolf algorithm has been proved to have higher performance in solving the target optimization problem. Convergence accuracy and faster convergence speed. Therefore, how to optimize the fuzzy neural network model based on the gray wolf algorithm to reduce the model training time and improve the model convergence accuracy has become an urgent problem to be solved.

The above content is only used to assist the understanding of the technical solutions of the present invention, and does not mean that the above content is the prior art.

SUMMARY OF THE INVENTION

The main purpose of the present invention is to provide an optimization method and system for a fuzzy neural network model, aiming at solving the technology of how to optimize the fuzzy neural network model based on the gray wolf algorithm to reduce the model training time and improve the model convergence accuracy question.

To achieve the above object, the present invention provides a method for optimizing a fuzzy neural network model, the method comprising the following steps:

Obtaining parameters to be optimized of the fuzzy neural network model, and generating initial position information of each individual in the preset gray wolf population based on the parameters to be optimized;

Calculate the first fitness value corresponding to each population individual according to the initial position information;

The initial position information is updated, the updated initial position information is obtained, and the updated initial position information is used as the target position information;

Calculate the second fitness value corresponding to each population individual according to the target location information, and perform iterative processing on the target location information based on the first fitness value and the second fitness value;

It is judged whether the iteratively processed target position information satisfies a preset iterative stop condition, and if so, an optimized fuzzy neural network model is output according to the iteratively processed target position information.

Preferably, before the step of obtaining the parameters to be optimized of the fuzzy neural network model, based on the parameters to be optimized, the step of generating the initial position information of each individual in the preset gray wolf population further includes:

Obtain attribute information and category information of the indicator to be measured, and use the attribute information and the category information as output indicators of the fuzzy neural network model;

performing a correlation analysis on the output index to obtain a correlation index of the output index and a correlation degree corresponding to the correlation index;

Select the correlation index whose correlation degree is greater than the preset correlation degree as the input index of the fuzzy neural network model;

Normalize the index information corresponding to the input index to obtain the index information to be measured;

After the step of judging whether the iteratively processed target position information satisfies a preset iterative stop condition, and if so, outputting an optimized fuzzy neural network model according to the iteratively processed target position information, the method further includes:

Inputting the index information to be measured into the optimized fuzzy neural network model to obtain a predicted value corresponding to the index information to be measured.

Preferably, the step of updating the initial location information, obtaining the updated initial location information, and using the updated initial location information as the target location information specifically includes:

Sorting the first fitness values in descending order to obtain a sorted set;

Selecting a target fitness value set from the sorting set according to a preset screening rule;

Update the initial position information according to the first fitness value included in the target fitness value set to obtain the updated initial position information;

The updated initial position information is used as target position information.

Preferably, the second fitness value corresponding to each population individual is calculated according to the target position information, and the target position information is iteratively processed based on the first fitness value and the second fitness value steps, including:

Calculate the second fitness value corresponding to each population individual according to the target location information, and use the population individual whose first fitness value is greater than the second fitness value as a disturbed population individual;

Calculate the center disturbance probability corresponding to the disturbance population individual according to the preset probability formula;

According to the center disturbance probability, it is judged whether it is necessary to perform center location disturbance processing on the target location information corresponding to the disturbance population individual;

If yes, iteratively process the target position information of the disturbed population individuals according to the preset disturbance formula.

Preferably, the step of judging whether the iteratively processed target position information satisfies a preset iterative stop condition, and if so, outputting an optimized fuzzy neural network model according to the iteratively processed target position information, specifically includes:

Calculate the third fitness value corresponding to each population individual according to the iteratively processed target position information;

Selecting the population individual with the largest third fitness value in the population as the optimal population individual and obtaining the current iteration processing times;

Judging whether a preset iteration stop condition is met according to the fitness value of the optimal population individual and the current iteration processing times;

When the fitness value of the optimal population individual is greater than or equal to a preset fitness value or the current number of iterations reaches a preset number of iterations, it is determined that the target location information after iterative processing satisfies a preset iteration condition, and Output the optimized fuzzy neural network model.

In addition, in order to achieve the above purpose, the present invention also proposes an optimization system for a fuzzy neural network model, the system comprising:

an information acquisition module, used for acquiring parameters to be optimized of the fuzzy neural network model, and generating initial position information of each individual in the preset gray wolf population based on the parameters to be optimized;

a fitness calculation module, configured to calculate the first fitness value corresponding to each population individual according to the initial position information;

a location update module, configured to update the initial location information, obtain updated initial location information, and use the updated initial location information as target location information;

an iterative processing module, configured to calculate a second fitness value corresponding to each population individual according to the target position information, and to iterate the target position information based on the first fitness value and the second fitness value deal with;

The model output module is used for judging whether the iteratively processed target position information satisfies a preset iterative stop condition, and when the preset iterative stop condition is met, outputs an optimized fuzzy neural network according to the iteratively processed target position information Model.

Preferably, the system further includes:

an information prediction module, used to obtain attribute information and category information of the indicator to be measured, and use the attribute information and the category information as output indicators of the fuzzy neural network model;

The information prediction module is further configured to perform a correlation analysis on the output index, and obtain a correlation index of the output index and a correlation degree corresponding to the correlation index;

The information prediction module is further configured to select the correlation index whose correlation degree is greater than the preset correlation degree as the input index of the fuzzy neural network model;

The information prediction module is further configured to perform normalization processing on the index information corresponding to the input index to obtain the index information to be measured;

The information prediction module is further configured to input the index information to be measured into the optimized fuzzy neural network model, so as to obtain the predicted value corresponding to the index information to be measured.

Preferably, the position update module is further configured to sort the first fitness values in descending order to obtain a sorted set;

The location update module is further configured to select a target fitness value set from the sorting set according to a preset screening rule;

The position update module is further configured to update the initial position information according to the first fitness value included in the target fitness value set, to obtain the updated initial position information;

The location update module is further configured to use the updated initial location information as target location information.

Preferably, the iterative processing module is further configured to calculate the second fitness value corresponding to each population individual according to the target location information;

The iterative processing module is further configured to use a population individual whose first fitness value is greater than the second fitness value as a disturbed population individual;

The iterative processing module is further configured to calculate the center disturbance probability corresponding to the disturbance population individual according to a preset probability formula;

The iterative processing module is further configured to determine, according to the center disturbance probability, whether it is necessary to perform center location disturbance processing on the target location information corresponding to the disturbance population individuals;

The iterative processing module is further configured to perform iterative processing on the target position information of the disturbed population individuals according to a preset disturbance formula when the center position disturbance processing needs to be performed on the target position information corresponding to the disturbed population individuals.

Preferably, the model output module is further configured to calculate the third fitness value corresponding to each population individual according to the iteratively processed target position information;

The model output module is also used to select the population individual with the largest third fitness value in the population as the optimal population individual and obtain the current iteration processing times;

The model output module is further configured to judge whether a preset iteration stop condition is met according to the fitness value of the optimal population individual and the current iteration processing times;

The model output module is further configured to determine the target after iterative processing when the fitness value of the optimal population individual is greater than or equal to a preset fitness value or the current number of iterations reaches a preset number of iterations The location information satisfies the preset iterative conditions, and the optimized fuzzy neural network model is output.

In the present invention, the parameters to be optimized of the fuzzy neural network model are obtained, the initial position information of each population individual in the preset gray wolf population is generated based on the parameters to be optimized, and the first position corresponding to each population individual is calculated according to the initial position information. fitness value, update the initial position information, obtain the updated initial position information, and use the updated initial position information as the target position information, and calculate the first position corresponding to each population individual according to the target position information. Two fitness values, and based on the first fitness value and the second fitness value, the target location information is iteratively processed to determine whether the iteratively processed target location information satisfies the preset iteration stop condition, if so , then output the optimized fuzzy neural network model according to the iteratively processed target position information to complete the training of the fuzzy neural network model, and train the fuzzy neural network model by using the improved gray wolf algorithm, The training time of the fuzzy neural network model is reduced, and the convergence accuracy of the fuzzy neural network model is improved.

Description of drawings

1 is a schematic flowchart of a first embodiment of an optimization method for a fuzzy neural network model of the present invention;

2 is a schematic flowchart of a second embodiment of an optimization method for a fuzzy neural network model of the present invention;

FIG. 3 is a structural block diagram of the first embodiment of the optimization system of the fuzzy neural network model of the present invention.

The realization, functional characteristics and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments.

Detailed ways

It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

An embodiment of the present invention provides a method for optimizing a fuzzy neural network model. Referring to FIG. 1 , FIG. 1 is a schematic flowchart of the first embodiment of the method for optimizing a fuzzy neural network model of the present invention.

In this embodiment, the optimization method of the fuzzy neural network model includes the following steps:

Step S10: obtaining the parameters to be optimized of the fuzzy neural network model, and generating initial position information of each population individual in the preset gray wolf population based on the parameters to be optimized;

It is easy to understand that the fuzzy neural network model described in this embodiment is constructed based on a fuzzy neural network, and the fuzzy neural network model can be further refined according to different actual needs. This embodiment mainly focuses on the fuzzy neural network model. The center point, the width vector and the weight of the output layer of the membership function in the fuzzy layer are trained, that is, the center point, the width vector and the weight of the output layer of the membership function in the fuzzy layer are used as the weights of the fuzzy neural network model. The parameters to be optimized are then iteratively processed by the improved gray wolf algorithm to the parameters to be optimized of the fuzzy neural network model to complete the training of the fuzzy neural network model. In specific implementation, the parameters to be optimized can be based on the Generate the initial position information of each population individual in the gray wolf population. The gray wolf population is the population corresponding to the improved gray wolf algorithm, and its size is determined by the structure of the fuzzy neural network. In specific applications, it can be It is further defined according to the input data of the fuzzy neural network model. The input data may be specific index information of the index to be measured. In this embodiment, the specific index information of the index to be measured is used as the Metric information can be obtained in the following ways:

Obtain attribute information and category information of the indicator to be measured, and use the attribute information and the category information as output indicators of the fuzzy neural network model, perform correlation analysis on the output indicators, and obtain the correlation indicators of the output indicators and The correlation degree corresponding to the correlation index, select the correlation index whose correlation degree is greater than the preset correlation degree as the input index of the fuzzy neural network model, perform normalization processing on the index information corresponding to the input index, and obtain the desired correlation index. For example, when measuring soil heavy metal content, the to-be-measured index is soil heavy metal content, its attribute information can be heavy metal content, and its category information can be soil type, and then the heavy metal content and the soil type are determined. Correlation analysis is performed to obtain the correlation index between the heavy metal content and the soil type, such as the coordinates of the soil sampling point, the altitude of the soil sampling point, the functional type of the soil sampling point, the organic matter content of the soil sampling point, and the soil sampling point. microbial content, moisture content of soil sampling points, etc., and then obtain the correlation degree corresponding to the above correlation index, select the correlation index with the correlation degree greater than the preset correlation degree as the input index of the fuzzy neural network model, and analyze the input The index information corresponding to the index is normalized to obtain the index information to be measured.

Step S20: Calculate the first fitness value corresponding to each population individual according to the initial position information;

It should be noted that, when the initial position information is obtained, the first fitness value corresponding to each population individual can be calculated according to the initial position information. Specifically, the initial position information can be used as the fuzzy neural network model. The center point of the membership function of the fuzzy layer, the width vector and the weight of the output layer, and then the indicator information to be measured is input to the input layer, fuzzy layer, fuzzy rule layer and fuzzy decision layer of the fuzzy neural network model in turn. , the output layer performs calculation to obtain the index output value, and then uses the mean square error value between the index output value and the preset target value as the first fitness value corresponding to each population individual.

Step S30: Update the initial position information, obtain the updated initial position information, and use the updated initial position information as target position information;

In a specific implementation, when the initial position information is updated, the first fitness values may be sorted in descending order to obtain a sorting set, and then the sorting is performed from the sorting according to a preset filtering rule. Select a target fitness value set from the set, then update the initial position information according to the first fitness value included in the target fitness value set, obtain the updated initial position information, and then update the updated initial position information. The initial position information is used as target position information. The preset screening rule may be to screen out the top three population individuals with the first fitness value in the sorted set, and then select the first fitness value corresponding to the top three population individuals in the sorting order. The initial position information is updated, the updated initial position information is obtained, and then the updated initial position information is used as the target position information. The following formula updates the initial position information,

x _inew =(w ₁ *x ₁ +w ₂ *x ₂ +w ₃ *x ₃ )/(w ₁ +w ₂ +w ₃ )

In the formula, x _alpha , x _beta , x _delta are the initial position information of the top three population individuals in the first fitness value ranking respectively, x _i is the initial position information of the population individual to be updated, f _xalpha , f _xbeta , f _xdelta are the first fitness values of the top three population individuals in the order of the first fitness value respectively, f _xi is the first fitness value of the population individuals to be updated, i is the current number of iterations, and i _max is the total number of iterations times, a _max is the maximum linear factor, c is a real number between 0 and 1 randomly generated in each calculation, x ₁ , x ₂ , and x ₃ are the top three population individuals in the first fitness value ranking respectively The location attractiveness of the population individuals to be updated, w ₁ , w ₂ , and w ₃ are respectively the attractiveness ratios of the population individuals with the top three rankings of the first fitness value to the population individuals to be updated, and x _inew is the target location information.

Step S40: Calculate the second fitness value corresponding to each population individual according to the target location information, and perform iterative processing on the target location information based on the first fitness value and the second fitness value;

It is easy to understand that, after obtaining the target position information, the second fitness value corresponding to each individual population can be calculated according to the target position information. Specifically, the target position information can be used as the fuzzy neural network model. The center point of the membership function of the fuzzy layer, the width vector and the weight of the output layer, and then the indicator information to be measured is input to the input layer, fuzzy layer, fuzzy rule layer and fuzzy decision layer of the fuzzy neural network model in turn. , the output layer performs calculations to obtain the index output value, and then the mean square error value between the index output value and the preset target value is used as the second fitness value corresponding to each population individual, and then based on the first fitness value The target location information is iteratively processed by the degree value and the second fitness value.

Step S50: Determine whether the iteratively processed target position information satisfies a preset iteration stop condition, and if so, output an optimized fuzzy neural network model according to the iteratively processed target position information.

It should be noted that, after the iterative processing is performed on the target location information, the third fitness value corresponding to each population individual can be calculated according to the iteratively processed target location information. Specifically, the iteratively processed target location information can be calculated first. They are respectively used as the center point, the width vector and the weight of the output layer of the membership function of the fuzzy layer in the fuzzy neural network model, and then the indicator information to be measured is input into the input layer, the fuzzy layer, the fuzzy neural network model in turn. The fuzzy rule layer, the fuzzy decision layer, and the output layer perform calculations to obtain the index output value, and then use the mean square error value between the index output value and the preset target value as the third fitness value corresponding to each population individual, Then select the population individual with the third largest fitness value in the population as the optimal population individual and obtain the current iteration processing times, and judge whether the preset number is satisfied according to the fitness value of the optimal population individual and the current iteration processing times. Iterative stop condition, when the fitness value of the optimal population individual is greater than or equal to the preset fitness value or the current iteration number reaches the preset iteration number, it is determined that the target location information after iterative processing satisfies the preset number of iterations Iterative conditions, and output the optimized fuzzy neural network model, and then input the indicator information to be measured into the optimized fuzzy neural network model to obtain the predicted value corresponding to the indicator information to be measured. When the soil heavy metal content is predicted, the predicted value corresponding to the indicator information to be measured is the predicted value of the soil heavy metal content generated based on the attribute information and the category information.

In this embodiment, the parameters to be optimized of the fuzzy neural network model are obtained, the initial position information of each population individual in the preset gray wolf population is generated based on the parameters to be optimized, and the first position corresponding to each population individual is calculated according to the initial position information. a fitness value, update the initial position information, obtain the updated initial position information, use the updated initial position information as the target position information, and calculate the corresponding second fitness value, and iteratively process the target position information based on the first fitness value and the second fitness value, and determine whether the iteratively processed target position information satisfies the preset iteration stop condition, if Satisfaction, then output the optimized fuzzy neural network model according to the iteratively processed target position information to complete the training of the fuzzy neural network model, and train the fuzzy neural network model by using the improved gray wolf algorithm , reducing the training time of the fuzzy neural network model and improving the convergence accuracy of the fuzzy neural network model.

Referring to FIG. 2 , FIG. 2 is a schematic flowchart of a second embodiment of a method for optimizing a fuzzy neural network model according to the present invention.

Based on the above-mentioned first embodiment, in this embodiment, the step S40 includes:

Step S401: Calculate the second fitness value corresponding to each population individual according to the target location information, and use the population individual whose first fitness value is greater than the second fitness value as a disturbed population individual;

It is easy to understand that, after obtaining the target position information, the second fitness value corresponding to each individual population can be calculated according to the target position information. Specifically, the target position information can be used as the fuzzy neural network model. The center point of the membership function of the fuzzy layer, the width vector and the weight of the output layer, and then the indicator information to be measured is input to the input layer, fuzzy layer, fuzzy rule layer and fuzzy decision layer of the fuzzy neural network model in turn. , the output layer is calculated to obtain the index output value, and then the mean square error value between the index output value and the preset target value is used as the second fitness value corresponding to each population individual, and then the The size of the first fitness value and the second fitness value, and a population individual whose first fitness value is greater than the second fitness value is used as a disturbance population individual.

Step S402: Calculate the center disturbance probability corresponding to the disturbance population individual according to a preset probability formula;

In a specific implementation, if the corresponding fitness value of the disturbed population individuals decreases after the location information is updated, the number of times the fitness value decreases of the disturbed population individual can be counted, and the number of times the fitness value decreases is initially set to 0, When the fitness value decreases continuously, the number of times of reducing the fitness value is increased in real time. When the fitness value does not continuously decrease or the center position perturbation processing is performed on the population individuals, the number of times of reducing the fitness value is repeated. is set to 0, and the center disturbance probability is calculated according to the reduction times of the fitness value, wherein the preset probability formula is,

In the formula, _pi is the center disturbance probability corresponding to the disturbance population individual, and T is the number of times the fitness value of the disturbance population individual decreases.

Step S403: Judging, according to the center disturbance probability, whether it is necessary to perform center location disturbance processing on the target location information corresponding to the disturbance population individual;

It should be noted that when judging whether the target position information corresponding to the disturbed population individual needs to be subjected to center position disturbance processing according to the center disturbance probability, a real number between 0 and 1 can be randomly generated, if the number is smaller than the center If the perturbation probability is determined, the center position perturbation processing is performed on the target position information of the perturbed population individuals.

Step S404: If yes, perform iterative processing on the target position information of the disturbed population individuals according to a preset disturbance formula.

In a specific implementation, when it is necessary to perform center position perturbation processing on the target position information of the perturbed population individuals, the position information of the center point of the gray wolf population can be obtained, and the location information of the center point of the gray wolf population is the gray wolf The mean value of the position information of all individuals in the population, and the current iteration number and the total number of iterations of the iterative processing of the target position information of the disturbed population individuals, and the target position information of the disturbed population individuals is iterated according to the preset disturbance formula processing, wherein the preset disturbance formula is,

In the formula, x _inew is the target position information of the disturbed population individual after iterative processing, x _mean is the position information of the center point of the gray wolf population, x _i is the target position information of the disturbed population individual, i is the current iteration number, and i _max is The total number of iterations, a _max is the maximum linear factor, and c is a real number between 0 and 1 randomly generated at each calculation.

In this embodiment, the second fitness value corresponding to each population individual is calculated according to the target location information, and the population individual whose first fitness value is greater than the second fitness value is regarded as a disturbed population individual. Set the probability formula to calculate the center disturbance probability value corresponding to the disturbance population individual, and judge whether it is necessary to perform center location disturbance processing on the target position information corresponding to the disturbance population individual according to the center disturbance probability value, and if so, according to the preset disturbance The formula iteratively processes the target position information of the disturbed population individuals, so that the fuzzy neural network can easily and quickly obtain the global optimal solution, thereby further shortening the training time of the fuzzy neural network model and improving the model neural network. The decision accuracy of the model.

Referring to FIG. 3 , FIG. 3 is a structural block diagram of a first embodiment of an optimization system for a fuzzy neural network model of the present invention.

As shown in FIG. 3 , the optimization system of the fuzzy neural network model proposed by the embodiment of the present invention includes:

An information acquisition module 10, configured to acquire parameters to be optimized of the fuzzy neural network model, and to generate initial position information of each individual in the preset gray wolf population based on the parameters to be optimized;

a fitness calculation module 20, configured to calculate the first fitness value corresponding to each population individual according to the initial position information;

a location update module 30, configured to update the initial location information, obtain updated initial location information, and use the updated initial location information as target location information;

The iterative processing module 40 is configured to calculate the second fitness value corresponding to each population individual according to the target position information, and perform the target position information based on the first fitness value and the second fitness value. iterative processing;

The model output module 50 is used for judging whether the iteratively processed target position information satisfies a preset iterative stop condition, and when the preset iterative stop condition is met, outputs the optimized fuzzy neural network according to the iteratively processed target position information network model.

In this embodiment, the parameters to be optimized of the fuzzy neural network model are obtained, the initial position information of each population individual in the preset gray wolf population is generated based on the parameters to be optimized, and the first position corresponding to each population individual is calculated according to the initial position information. a fitness value, update the initial position information, obtain the updated initial position information, use the updated initial position information as the target position information, and calculate the corresponding second fitness value, and iteratively process the target position information based on the first fitness value and the second fitness value, and determine whether the iteratively processed target position information satisfies the preset iteration stop condition, if Satisfaction, then output the optimized fuzzy neural network model according to the iteratively processed target position information to complete the training of the fuzzy neural network model, and train the fuzzy neural network model by using the improved gray wolf algorithm , reducing the training time of the fuzzy neural network model and improving the convergence accuracy of the fuzzy neural network model.

Based on the above-mentioned first embodiment of the optimization system of the fuzzy neural network model of the present invention, a second embodiment of the optimization system of the fuzzy neural network model of the present invention is proposed.

In this embodiment, the system further includes:

an information prediction module 60, configured to obtain attribute information and category information of the indicator to be measured, and use the attribute information and the category information as output indicators of the fuzzy neural network model;

The information prediction module 60 is further configured to perform a correlation analysis on the output index, and obtain a correlation index of the output index and a correlation degree corresponding to the correlation index;

The information prediction module 60 is further configured to select the correlation index whose correlation degree is greater than the preset correlation degree as the input index of the fuzzy neural network model;

The information prediction module 60 is further configured to perform normalization processing on the index information corresponding to the input index to obtain the index information to be measured;

The information prediction module 60 is further configured to input the indicator information to be measured into the optimized fuzzy neural network model to obtain the predicted value corresponding to the indicator information to be measured.

The position update module 30 is further configured to sort the first fitness values in descending order to obtain a sorted set;

The position update module 30 is further configured to select a target fitness value set from the sorting set according to a preset screening rule;

The position update module 30 is further configured to update the initial position information according to the first fitness value included in the target fitness value set, to obtain the updated initial position information;

The location update module 30 is further configured to use the updated initial location information as target location information.

The iterative processing module 40 is further configured to calculate the second fitness value corresponding to each population individual according to the target location information;

The iterative processing module 40 is further configured to use a population individual whose first fitness value is greater than the second fitness value as a disturbed population individual;

The iterative processing module 40 is further configured to calculate the center disturbance probability value corresponding to the disturbance population individual according to a preset probability formula;

The iterative processing module 40 is further configured to determine, according to the central perturbation probability value, whether it is necessary to perform central position perturbation processing on the target position information corresponding to the perturbed population individuals;

The iterative processing module 40 is further configured to perform iterative processing on the target position information of the disturbed population individual according to a preset disturbance formula when the center position disturbance processing needs to be performed on the target position information corresponding to the disturbed population individual.

The model output module is further configured to calculate the third fitness value corresponding to each population individual according to the iteratively processed target position information;

The model output module 50 is further configured to select the population individual with the largest third fitness value in the population as the optimal population individual and obtain the current iteration processing times;

The model output module 50 is further configured to judge whether a preset iteration stop condition is satisfied according to the fitness value of the optimal population individual and the current iteration processing times;

The model output module 50 is further configured to determine the said fitness value after iterative processing when the fitness value of the optimal population individual is greater than or equal to a preset fitness value or the current iteration number reaches a preset iteration number. The target position information meets the preset iterative conditions, and the optimized fuzzy neural network model is output.

For other embodiments or specific implementations of the system for optimizing the fuzzy neural network model of the present invention, reference may be made to the above method embodiments, which will not be repeated here.

It should be noted that, herein, the terms "comprising", "comprising" or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article or system comprising a series of elements includes not only those elements, It also includes other elements not expressly listed or inherent to such a process, method, article or system. Without further limitation, an element qualified by the phrase "comprising a..." does not preclude the presence of additional identical elements in the process, method, article or system that includes the element.

The above-mentioned serial numbers of the embodiments of the present invention are only for description, and do not represent the advantages or disadvantages of the embodiments.

From the description of the above embodiments, those skilled in the art can clearly understand that the method of the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is better implementation. Based on this understanding, the technical solutions of the present invention can be embodied in the form of software products that are essentially or contribute to the prior art, and the computer software products are stored in a storage medium (such as read-only memory/random access). memory, magnetic disk, optical disc), including several instructions to make a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) execute the methods described in the various embodiments of the present invention.

The above are only preferred embodiments of the present invention, and are not intended to limit the scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the description and drawings of the present invention, or directly or indirectly applied in other related technical fields , are similarly included in the scope of patent protection of the present invention.

Claims (10)

1. an optimization method of fuzzy neural network model, is characterized in that, described method comprises the following steps: Obtaining parameters to be optimized of the fuzzy neural network model, and generating initial position information of each individual in the preset gray wolf population based on the parameters to be optimized; Calculate the first fitness value corresponding to each population individual according to the initial position information; The initial position information is updated, the updated initial position information is obtained, and the updated initial position information is used as the target position information; Calculate the second fitness value corresponding to each population individual according to the target location information, and perform iterative processing on the target location information based on the first fitness value and the second fitness value; It is judged whether the iteratively processed target position information satisfies a preset iterative stop condition, and if so, an optimized fuzzy neural network model is output according to the iteratively processed target position information. 2. The method according to claim 1, characterized in that, the step of obtaining the parameters to be optimized of the fuzzy neural network model, based on the parameters to be optimized, generates the initial position information of each population individual in the preset gray wolf population Before, also included: Obtain attribute information and category information of the indicator to be measured, and use the attribute information and the category information as output indicators of the fuzzy neural network model; performing a correlation analysis on the output index to obtain a correlation index of the output index and a correlation degree corresponding to the correlation index; Select the correlation index whose correlation degree is greater than the preset correlation degree as the input index of the fuzzy neural network model; Normalize the index information corresponding to the input index to obtain the index information to be measured; After the step of judging whether the iteratively processed target position information satisfies a preset iterative stop condition, and if so, outputting an optimized fuzzy neural network model according to the iteratively processed target position information, the method further includes: Inputting the index information to be measured into the optimized fuzzy neural network model to obtain a predicted value corresponding to the index information to be measured. 3. The method according to claim 1, wherein the step of updating the initial position information, obtaining the updated initial position information, and using the updated initial position information as the target position information , including: Sorting the first fitness values in descending order to obtain a sorted set; Selecting a target fitness value set from the sorting set according to a preset screening rule; Update the initial position information according to the first fitness value included in the target fitness value set to obtain the updated initial position information; The updated initial position information is used as target position information. 4. The method according to claim 1, wherein the second fitness value corresponding to each population individual is calculated according to the target location information, and based on the first fitness value and the second fitness value The step of iteratively processing the target location information by the fitness value specifically includes: Calculate the second fitness value corresponding to each population individual according to the target location information, and use the population individual whose first fitness value is greater than the second fitness value as a disturbed population individual; Calculate the center disturbance probability corresponding to the disturbance population individual according to the preset probability formula; According to the center disturbance probability, it is judged whether it is necessary to perform center location disturbance processing on the target location information corresponding to the disturbance population individual; If yes, iteratively process the target position information of the disturbed population individuals according to the preset disturbance formula. 5. The method according to claim 1, wherein the judging whether the iteratively processed target position information satisfies a preset iterative stop condition, and if so, outputting the optimized output according to the iteratively processed target position information. The steps of the fuzzy neural network model include: Calculate the third fitness value corresponding to each population individual according to the iteratively processed target position information; Selecting the population individual with the largest third fitness value in the population as the optimal population individual and obtaining the current iteration processing times; Judging whether a preset iteration stop condition is met according to the fitness value of the optimal population individual and the current iteration processing times; When the fitness value of the optimal population individual is greater than or equal to a preset fitness value or the current number of iterations reaches a preset number of iterations, it is determined that the target location information after iterative processing satisfies a preset iteration condition, and Output the optimized fuzzy neural network model. 6. An optimization system of a fuzzy neural network model, wherein the system comprises: an information acquisition module, used for acquiring parameters to be optimized of the fuzzy neural network model, and generating initial position information of each individual in the preset gray wolf population based on the parameters to be optimized; a fitness calculation module, configured to calculate the first fitness value corresponding to each population individual according to the initial position information; a location update module, configured to update the initial location information, obtain updated initial location information, and use the updated initial location information as target location information; an iterative processing module, configured to calculate a second fitness value corresponding to each population individual according to the target position information, and to iterate the target position information based on the first fitness value and the second fitness value deal with; The model output module is used for judging whether the iteratively processed target position information satisfies a preset iterative stop condition, and when the preset iterative stop condition is met, outputs an optimized fuzzy neural network according to the iteratively processed target position information Model. 7. The system of claim 6, further comprising: an information prediction module, used to obtain attribute information and category information of the indicator to be measured, and use the attribute information and the category information as output indicators of the fuzzy neural network model; The information prediction module is further configured to perform a correlation analysis on the output index, and obtain a correlation index of the output index and a correlation degree corresponding to the correlation index; The information prediction module is further configured to select the correlation index whose correlation degree is greater than the preset correlation degree as the input index of the fuzzy neural network model; The information prediction module is further configured to perform normalization processing on the index information corresponding to the input index to obtain the index information to be measured; The information prediction module is further configured to input the index information to be measured into the optimized fuzzy neural network model, so as to obtain the predicted value corresponding to the index information to be measured. 8. The system of claim 6, wherein the location update module is further configured to sort the first fitness values in descending order to obtain a sorted set; The location update module is further configured to select a target fitness value set from the sorting set according to a preset screening rule; The position update module is further configured to update the initial position information according to the first fitness value included in the target fitness value set, to obtain the updated initial position information; The location update module is further configured to use the updated initial location information as target location information. 9. The system according to claim 6, wherein the iterative processing module is further configured to calculate the second fitness value corresponding to each population individual according to the target position information; The iterative processing module is further configured to use a population individual whose first fitness value is greater than the second fitness value as a disturbed population individual; The iterative processing module is further configured to calculate the center disturbance probability corresponding to the disturbance population individual according to a preset probability formula; The iterative processing module is further configured to determine, according to the center disturbance probability, whether it is necessary to perform center location disturbance processing on the target location information corresponding to the disturbance population individuals; The iterative processing module is further configured to perform iterative processing on the target position information of the disturbed population individuals according to a preset disturbance formula when the center position disturbance processing needs to be performed on the target position information corresponding to the disturbed population individuals. 10. The system according to claim 6, wherein the model output module is further configured to calculate the third fitness value corresponding to each population individual according to the iteratively processed target position information; The model output module is also used to select the population individual with the largest third fitness value in the population as the optimal population individual and obtain the current iteration processing times; The model output module is further configured to judge whether a preset iteration stop condition is met according to the fitness value of the optimal population individual and the current iteration processing times; The model output module is further configured to determine the target after iterative processing when the fitness value of the optimal population individual is greater than or equal to a preset fitness value or the current number of iterations reaches a preset number of iterations The location information satisfies the preset iterative conditions, and the optimized fuzzy neural network model is output.

Images