This article is devoted to the solution of the problem of constructing a neuro-fuzzy model for classifying the type of glass. The solution of this problem is relevant in criminological investigations, in medicine, the window industry, the automotive industry and other subject areas. To solve it, the feasibility of constructing and using a neuro-fuzzy model is substantiated. To build the model, it was necessary to select and prepare a data set characterizing various types of glass, as well as select and use the tool environment of neuro-fuzzy modeling. To solve the problem, a search for data sets was performed in the UCI repository and the "Glass identification" set was found, designed to recognize the following types of glass: heat-polished glass of buildings, ordinary glass of buildings, heat-polished glass of cars, glass containers, glass for dishes, glass of car headlights. The data set for analysis was prepared on the basis of the Deductor analytical platform. For this, a scenario was developed that provides for the following steps: loading data, setting up a data set, performing correlation analysis, assessing the quality of data, editing outliers and extreme values, exporting data. At the data setup stage, the data types (integer or real) and types (discrete or continuous) were set for the corresponding input and output columns. Correlation analysis allowed us to identify the degrees of influence of each input column on the output column and select informative features for analysis. The data quality assessment stages, editing of outliers and extreme values were also implemented. After completing the specified procedures, the data were exported to a text file for further analysis and building a neuro-fuzzy model. The final data sample for analysis included 214 rows, 4 input columns (Na, Mg, Al, Ba) and 1 output (Type of glass) with 6 classes of glass type. Based on the prepared data, the fuzzy neural network was trained in the Neuro-Fuzzy System for Forming Fuzzy Models for Assessing the Discrete State of Objects software package. The model construction time was 6 minutes and 42 seconds. During this time, 6 full cycles of fuzzy neural network training were implemented. In each cycle, the genetic algorithm adjusted the values of 20 parameters of the membership functions. During training, it was possible to achieve classification accuracy of 93.62% on the training data sample and 92.21% on the test sample. This indicates the adequacy of the constructed model and the possibility of its effective practical use.

The paper substantiates a method for creating training datasets for fuzzy neural networks, which can be used to promptly obtain probabilistic estimates for the causes of abnormal critical events or incidents in diagnostic systems. The rules for converting the hypotheses on potential incident causes into intervals of defect probability in a process chain at a certain stage of continuous production are considered using belief functions. We propose a procedure for converting these hypotheses into a database of fuzzy production rules automatically, which provides training an adaptive neural network based on the Takagi–Sugeno–Kang fuzzy inference system. This makes it possible to quickly calculate a relatively accurate probabilistic estimate of a malfunction in the process chain without using expensive computing resources.

In this paper, an evolutionary-neuro-fuzzy-based task scheduling approach (ENF-S) to jointly optimize the main critical parameters of heterogeneous multi-core systems is proposed. This approach has two phases: first, the fuzzy neural network (FNN) is trained using a non-dominated sorting genetic algorithm (NSGA-II), considering the critical parameters of heterogeneous multi-core systems on a training data set consisting of different application graphs. These critical parameters are execution time, temperature, failure rate, and power consumption. The output of the trained FNN determines the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">criticality degree</i> for various processing cores based on the system's current state. Next, the trained FNN is employed as an online scheduler to jointly optimize the critical objectives of multi-core systems at runtime. Due to the uncertainty in sensor measurements and the difference between computational models and reality, applying the fuzzy neural network is advantageous. The efficiency of ENF-S is investigated in various aspects including its joint optimization capability, appropriateness of generated fuzzy rules, comparison with related research, and its overhead analysis through several experiments on real-world and synthetic application graphs. Based on these experiments, our ENF-S outperforms the related studies in optimizing all design criteria. Its improvements over related methods are estimated <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$19.21\%$</tex-math></inline-formula> in execution time, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$13.07\%$</tex-math></inline-formula> in temperature, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$25.09\%$</tex-math></inline-formula> in failure rate, and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$13.16\%$</tex-math></inline-formula> in power consumption, averagely.

Abstract The survival soil on which the traditional martial arts culture depends is becoming thinner and thinner. A large number of superb martial arts skills and martial arts experience disappear with the passing of the older generation of martial artists. In order to realize the information preservation of traditional Wushu, this paper puts forward the establishment and application of traditional Wushu intelligent learning resource database under the background of big data. In the context of big data, data mining technology is used to realize personalized recommended learning services, obtain the feature structure of intelligent learning system through operation, extract the average dynamic features of specific data in the resource database, obtain fuzzy constraints according to the value, determine the membership function, and adjust the membership function parameters through training fuzzy neural network, so as to gradually improve the reasoning accuracy. Finally, the matching rule set is used to filter the resource data packets to achieve the purpose of communication. In order to verify the effect of the model, compared with the traditional model, the results show that when the number of nodes is 500, the average transmission rate is as high as 90%, the average delay is 12 seconds, and the throughput performance is 91%. It can be verified that the model designed in this paper can effectively improve the propagation rate, reduce the average delay and strengthen the throughput performance.

Training fuzzy deep neural network with honey badger algorithm for intrusion detection in cloud environment

Under the influence of COVID-19, online office and online education has ushered in a golden period of development. The teaching quality of online education has been a controversial issue. Our study takes online course teaching quality assessment as the starting point, explores the influencing factors of online course quality assessment with online courses as the research object, and analyzes the latest research proposal for an online course quality index. To make the online course quality assessment more intelligent, we propose an online course quality assessment method based on a fuzzy neural network. The method uses fuzzy rules as the baseline and adds a TSK perception mechanism to expand the perception domain of the fuzzy neural network and improve the course quality index prediction accuracy. At the input side of the fuzzy neural network, we preclassify the online course data into four parts, and each part of the data represents a different assessment domain. Due to the large data cost, we expanded the collective amount of data using data augmentation methods. In addition, we parse the structure of the fuzzy neural network hierarchy and introduce the construction and role of the TSK perception mechanism in the fuzzy rules. An optimal learning strategy is proposed in the fuzzy neural network training. Finally, in the experimental session, we verify the effectiveness of data augmentation and explore the distribution of course quality assessment weights. In the comparison of the model prediction results with the actual assessment results, our method achieves an excellent matching rate, which proves the high efficiency of our method in the online course quality assessment system.

The article examines some algorithms for joint processing of raw data on the state of a complex multistage continuous production process to obtain probabilistic characteristics of abnormal critical events that can potentially lead to single failures or even emergencies. The article, thus, proposes and substantiates an approach to developing a technology to detect and predict malfunctions and determine their causes. The sequence of operations to process and convert diagnostic process data is considered essential. As a result, the article presents a general diagnostic model of a multistage production process. The model can formalize the main objects and processes in terms of the problem being solved. An incident is defined as an abnormal critical event described by non-normative values of diagnostic variables. Incidents are shown to be indicated by the corresponding membership functions. The hypotheses on potential incident causes are discussed to be built with belief functions being the basis of evidence theory or Dempster–Shafer theory. The hypotheses are characterized by an interval of malfunction probability in some process chain. The authors propose a procedure of converting these hypotheses into fuzzy production rules automatically. The automatical procedure is a prerequisite to using fuzzy neural networks to obtain a reliable estimate of the degree of belief in the incident cause. As a summary, the generated database of the production rules to train a neural network is substantiated to be used with the TSK architecture that makes possible to estimate a malfunction probability in the process chain quickly without resource-intensive computations.

The design of a fuzzy neural network (FNN) has long been a challenging problem since most methods rely on approximation error to train an FNN, which may easily result in overfitting phenomenon to degrade the generalization performance. To improve the generalization performance, an FNN with a multiobjective optimization algorithm (MOO-FNN) is proposed in this article. First, the multilevel learning objectives are designed around the generalization performance to guide the training process of an FNN. Then, the method utilizes the approximation error, the structure complexity, and the output smoothness indicators instead of a single indicator to improve the evaluation accuracy of generalization performance. Second, an MOO algorithm with continuous–discrete variables is developed to optimize the FNN. Then, MOO is able to use a novel particle update method to adjust both the structure and parameters rather than adjusting them separately, thereby achieving suitable generalization performance of the FNN. Third, the convergence of MOO-FNN is analyzed in detail to guarantee its successful applications. Finally, the experimental studies of MOO-FNN have been performed on model identification of nonlinear systems to verify the effectiveness. The results illustrate that MOO-FNN has a significant improvement over some state-of-the-art algorithms.

Transfer learning algorithm can provide a framework to utilize the previous knowledge to train fuzzy neural network (FNN). However, the performance of TL-based FNN will be destroyed by the knowledge over-fitting problem in the learning process. To solve this problem, an interactive transfer learning (ITL) algorithm, which can alleviate the negative transfer among different domains to improve the learning performance of FNN, is designed and analyzed in this article. This ITL-assisted FNN (ITL-FNN) contains the following advantages. First, a knowledge filter algorithm is developed to reconstruct the knowledge in source scene by balancing the matching accuracy and diversity. Then, the knowledge from source scene can fit the instance of target scene with suitable accuracy. Second, a self-balancing mechanism is designed to balance the driven information between the source and target scenes. Then, the knowledge can be refitted to reduce the useless information. Third, a structural competition algorithm is proposed to adjust the knowledge of FNN. Then, the proposed ITL-FNN can achieve compact structure to improve the generalization performance. Finally, some benchmark problems and industrial applications are provided to demonstrate the merits of ITL-FNN.

A novel learning algorithm based on computing the rules’ desired outputs of a TSK fuzzy neural network with non-separable fuzzy rules

The basic fuzzy neural network algorithm has slow convergence and large amount of calculation, so this paper designed a particle swarm optimization trained fuzzy neural network algorithm to solve this problem. Traditional particle swarm optimization is easy to fall into local extremes and has low efficiency, this paper designed new update rules for inertia weight and learning factors to overcome these problems. We also designed training rules for the improved particle swarm optimization to train fuzzy neural network, and the hybrid algorithm is applied to solve the path planning problem of intelligent driving vehicles. The efficiency and practicability of the algorithm are proved by experiments.

In rainy weather, the accurate prediction of traffic status not only helps road traffic managers to formulate traffic management methods but also helps travelers design travel routes and even adjust travel time. In this paper, based on six-dimensional data (e.g., past and present spatiotemporal traffic status, road network structure, pavement type, water accumulation, and rainfall level), a fuzzy neural network (FNN) prediction system is proposed to predict traffic status. The traffic status evolution trend is related not only to the existing traffic but also to the new traffic demand. Therefore, the FNN prediction system designed includes offline and online parts using the data of the past and the day separately and avoids the forecast of new traffic demand. The fuzzy C-means clustering algorithm is applied to cluster traffic status data under similar rainy weather in the past to form an offline initial dataset, which is used to train FNN weight parameters. The online part uses real-time detection data and the parameters trained by the offline part to further predict the traffic status and returns the prediction errors to the offline part to correct the weight parameters to further improve prediction accuracy. Finally, the FNN prediction system is verified using real Beijing expressway network data. The verification results show that the prediction system can guarantee prediction accuracy and can be used to effectively identify traffic status.

A gradient aggregate asymptotical smoothing algorithm for training max–min fuzzy neural networks

Bearing performance degradation assessment has great significance to condition‐based maintenance (CBM). A novel degradation modeling method based on EMD‐SVD and fuzzy neural network (FNN) was proposed to identify and evaluate the degradation process of bearings in the whole life cycle accurately. Firstly, the vibration signals of bearings in known states were decomposed by empirical mode decomposition (EMD) to obtain the intrinsic mode functions (IMFs) containing feature information. Then, the selected key IMFs which contain the main features were decomposed by singular value decomposition (SVD). And the decomposed results were used as the training samples of FNN. At last, the output results of the tested data were normalized to the health index (HI) through learning and training of FNN, and then the performance degradation degree could be described by the distance between the test sample and the normal one. According to the case study, this modeling method could evaluate the performance degradation of bearings effectively and identify the early fault features accurately. This method also provided an important maintenance strategy for the CBM of bearings.

Designing a self-constructing fuzzy neural network controller for damping power system oscillations

Vehicle test has a great significance for the development of new vehicle products. It is necessary for a new type of vehicle stereotypes to conduct a lot of vehicle test. Some vehicle test, such as emission durability test, vehicle performance test, vehicle noise test, high and low temperature environment test, vehicle road test, vehicle bench test, is more suitable for the operation by robot. Robot driver is an intelligent robot that can realize automatic driving under harsh environment in vehicle test instead of a human driver without any modification. Because of the vehicle is not required to be modified, and the vehicle robot driver can be directly installed in the different vehicle cab. The drive way of the vehicle robot driver includes the hydraulic drive, the pneumatic drive and the servo electric drive. The hydraulic drive is steady, but it needs an oil cylinder. It is not easy for the pneumatic drive to accurate positioning and the real-time property. As a driving device of the robot driver, the servo electric drive needs a mechanism of rotary motion into linear motion. Electromagnetic linear motor can solve the shortcomings of three other drive styles. It can improve the transmission efficiency and transmission accuracy, and make the transmission mechanism simple. A control approach of speed in an electromagnetic direct drive robot driver based on fuzzy neural network is proposed in this paper, in order to realize the accurate speed tracking of different driving test cycle conditions. The electromagnetic direct drive robot driver adopts an electromagnetic linear actuator as the driving device in this paper. The throttle mechanical leg, the brake mechanical leg, the clutch mechanical leg and the shift manipulator are directly driven through the electromagnetic linear actuator. The control system structure and the coordinated motion control model of the electromagnetic direct drive robot driver are given. On the basis of this, the speed control model based on fuzzy neural network of the electromagnetic direct drive robot driver is designed. The shift manipulator displacement, the throttle mechanical leg displacement, the clutch mechanical leg displacement and the brake mechanical leg are the input variables of the fuzzy neural network model, and the vehicle speed of the test vehicle is the output variable of the fuzzy neural network model. The number of the input variable membership functions is three, and the type of the input variable membership functions is gbellmf. The fuzzy neural network training algorithm adapts the hybrid learning algorithm combing with back propagation algorithm and least square method. The proposed control method of the electromagnetic direct drive robot driver is experimentally proved and compared with other control methods and with human driver performances. Actual vehicle test results and error comparison analysis show that the vehicle speed tracking accuracy of robot driver using the proposed approach is higher than that of robot driver using PID control method and human driver. Besides, the proposed method has good adaptability under all kinds of driving test cycle, which can ensure the accuracy and effectiveness of vehicle test.

In view of the limitation of fault diagnosis methods in substation intelligent patrol system, a fault diagnosis method based on multi-sensors information fusion is proposed. In the field of fault diagnosis, this method can deal with uncertain and imprecise information by using fuzzy theory, and has a high self-study capability based on neural network. Collecting samples of data through establishing many sensors in the scene of the intelligent patrol system, and then through the BP algorithm of fuzzy neural network training to achieve accurate fault diagnosis function of the intelligent patrol system. By comparing the result of an example, it shows that, compared with using single information, using multi-sensors information as the diagnosis method is more accurate and reliable in the intelligent patrol system.

In order to automatically control the quality of the welding process, this paper concentrates on the problem of three dimensional surface reconstructions for the welding pool. As the relationships between size and shape of the welding pool are very complex and nonlinear, we utilize the fuzzy neural network to solve the proposed problem. In the fuzzy neural network, the input vector with 48 dimensions is made up of three parts: 1) welding parameters, 2) welding pool size parameters, and 3) shape parameters. In particular, the size of the welding pool negative side is regarded as the output. In the experiment, the welding process is implemented using direct current electrode negative GTAW, and then we suppose that the weld pool rotates when torch orientation, imaging plane, laser projector, and camera are fixed. Experimental results demonstrate that 1) the speed of the fuzzy neural network training process is fairly quick, and 2) the proposed three dimensional surface reconstruction method is robust under current disturbances.

A solution to the problem of training fuzzy neural networks is considered. A method of parametric identification of fuzzy neural models that is based on a probabilistic approach when searching for the values of adjustable parameters is proposed. The method allocates the most resource-intensive stages among nodes of a parallel computing system, which reduces the time it takes to adjust the parameters. It is proposed to take into account information on the training sample when forming the initial set of solutions and significance of terms of features, which brings the initial points closer to optimal and accelerates the optimization process.

El propósito de este artículo es presentar estrategias generales de entrenamiento para redes neuronales de números difusos utilizadas en el aprendizaje de sistemas a partir de información lingüística. Se exponen brevemente las principales tendencias en el entrenamiento de este tipo de sistemas y con base en ellas se proponen nuevas estrategias. La primera de ellas se basa en la retropropagación del error cuadrático medio en todos los a-cortes para pesos crisp. La segunda hace uso de un algoritmo genético con codificación real para redes con pesos crisp. La tercera consiste en la retropropagación del error en el valor promedio y la ambigüedad en todos los a-cortes para pesos difusos. Por último, se presenta una basada en la retropropagación de una medida difusa del error para redes con pesos difusos. Se realiza una etapa experimental en la que se implementan los algoritmos desarrollados junto con algunos de los más representativos reportados en el estado del arte, permitiendo identificar para qué conjuntos de datos particulares resulta útil cada una de las estrategias. Finalmente, se aplican dichas estrategias para la implementación de un sistema de evaluación de impacto ambiental en vertederos.