CN113743462A - HWOA-ELM-based error deflection angle identification method for mechanical arm end clamping - Google Patents

HWOA-ELM-based error deflection angle identification method for mechanical arm end clamping Download PDF

Info

Publication number
CN113743462A
CN113743462A CN202110876485.0A CN202110876485A CN113743462A CN 113743462 A CN113743462 A CN 113743462A CN 202110876485 A CN202110876485 A CN 202110876485A CN 113743462 A CN113743462 A CN 113743462A
Authority
CN
China
Prior art keywords
elm
individual
deflection angle
population
hwoa
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110876485.0A
Other languages
Chinese (zh)
Inventor
张立彬
支乐威
陈教料
阮贵航
胥芳
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.)
Zhejiang University of Technology ZJUT
Original Assignee
Zhejiang University of Technology ZJUT
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 Zhejiang University of Technology ZJUT filed Critical Zhejiang University of Technology ZJUT
Priority to CN202110876485.0A priority Critical patent/CN113743462A/en
Publication of CN113743462A publication Critical patent/CN113743462A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • 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
    • G06N3/045—Combinations of networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computing Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computational Linguistics (AREA)
  • Computational Mathematics (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Image Analysis (AREA)

Abstract

An error deflection angle identification method for mechanical arm tail end clamping based on HWOA-ELM is characterized in that a tail end position signal is used for changing deflection angle size and then assembling is carried out, and force data signals corresponding to different error deflection angles are collected; carrying out initialization setting on parameters; generating an initial parameter population individual of the ELM by using a logic mapping strategy; introducing an ELM initial parameter individual of opposite learning strategy optimization logic mapping; further optimizing the individual parameters with the fitness value lower than the average fitness through a wavelet mutation strategy; finding the best candidate solution in the final parameter individual by using WOA; optimizing population-updated individuals by adding gaussian perturbations; calculating the fitness value of the population individuals and updating the optimal parameters; constructing an error deflection angle identification model based on HWOA-ELM; and identifying the test set by using the HWOA-ELM, and comparing the identification tag with the actual tag to obtain the classification precision, thereby drawing a classification result graph. The deflection angle recognition method is high in deflection angle recognition accuracy.

Description

HWOA-ELM-based error deflection angle identification method for mechanical arm end clamping
Technical Field
The invention belongs to the field of machine learning and pattern recognition, and is suitable for the field of mechanical arm tail end clamping error deflection angle recognition during part assembly. In particular to an error declination identification method based on Hybrid Multi-strategy Whale Optimization Algorithm (HWOA) and Extreme Learning Machine (ELM).
Background
The parts assembly is widely applied in the industrial field, a large amount of manpower and material resources are consumed in the traditional manual parts assembly, and the success of the parts assembly greatly depends on the technology of assembly operators. The industrial production is urgently required to be shifted to automation and intellectualization, and the assembly by using industrial robots instead of the traditional manual assembly is gradually the trend of assembly automation and intellectualization. The industrial robot assembly can greatly reduce the manpower consumption, improve the labor productivity and promote the development of emerging industries, thereby improving the social and economic benefits and promoting the technological progress. The robot generally controls the tail end of the mechanical arm to perform assembly actions by using position information, but the assembly of parts is often unsuccessful due to the fact that the tail end of the mechanical arm is clamped and has an angle deviation. The analysis of the error deflection angle of the clamping of the tail end of the mechanical arm is beneficial to correctly sensing the posture of the part in the assembling process, so that the mechanical arm is guided to carry out the adjustment of the assembling action on the part with the wrong posture. The error deflection angle of the part can be obtained by analyzing the force and moment signals collected by the force sensor during assembly. The method for recognizing the deflection error of the end grip of the robot arm can improve the assembly efficiency of the industrial robot, and therefore, it is necessary to conduct research.
Machine learning is a subject of multi-domain intersection, and enables a robot to acquire new knowledge or skills to guide assembly actions through skill learning. The identification of the error deflection angle is mainly based on a classification algorithm in machine learning, and the error deflection angle between the part and the tail end of the mechanical arm is predicted by analyzing force and moment signals through the classification algorithm. Some existing simple machine learning algorithms such as decision trees, K-nearest neighbor algorithms, support vector machines, neural networks can predict the corresponding deflection category according to the force signal. However, since the force signals of different deflection angle classes have small changes, the classification accuracy of the algorithms is low, and the algorithms cannot be well applied to actual assembly tasks.
Therefore, it is necessary to design a method for identifying an error deflection angle between the end of the mechanical arm and the clamped part, and research on the field of part assembly of industrial robots is needed.
Disclosure of Invention
In order to solve the problem that the deflection angle identification precision of the tail end of a mechanical arm and a clamped part in the process of assembling the part by using the existing classification algorithm is low, the invention provides an error deflection angle identification method based on HWOA-ELM mechanical arm tail end clamping, which has higher deflection angle identification precision.
In order to achieve the purpose, the invention provides the following technical scheme:
an error deflection angle identification method for mechanical arm end clamping based on HWOA-ELM comprises the following steps:
step 1: through arm centre gripping part, fixed part is vertical direction, utilizes terminal position signal to change the declination size, assembles again, gathers the corresponding power data signal F ═ F { F of different error declinationsxi,Fyi,Fzi,Mxi,Myi,MziIn which Fxi,Fyi,FziRepresenting forces in the x, y, z directions, Mxi,Myi,MziRepresenting the moments along x, Y, z, and combining the force data signal F and the number Y of the corresponding deflection angle into an error deflection angle data set, which is further subdivided into a training set { F }1,Y1And test set F2,Y2}；
Step 2: carrying out initialization setting of parameters including hidden layer neuron number N of ELM and input weight omegakAnd input layer to hidden layer bias bkSearch range [ omega ] ofmax,ωmin]、[bmax, bmin]Of HWOAThe population number of individuals is M, the iteration times are t, and the maximum iteration times are tmax；
And step 3: generating initial parameter population individual h of ELM by using logic mapping strategyk,jWherein j is 1 or 2, hk,1＝ωk,hk,2＝bkAnd k represents the current individual of the parameter population, and the population individual of the parameter after logical mapping is represented as:
Figure BDA0003188574380000031
Figure BDA0003188574380000032
in the formula, r1、r2、rk-1Random numbers that are all 0 to 1;
and 4, step 4: the initial parameters of the ELM introducing the opponent learning strategy optimization logical mapping are expressed as:
Hk,j＝hmax,j+hmin,j-hk,j (3)
Figure BDA0003188574380000033
in the formula, Hk,jIs an opponent individual; when j is 1, hmax,jAnd hmin,jIs omegamaxAnd ωminWhen j is 2, hmax,jAnd hmin,jIs b ismaxAnd bmin；V(Hk,j) And V (h)k,j) The fitness function is used as the quality of the evaluation parameter individual;
and 5: and further optimizing the parameter individuals with the fitness values lower than the average fitness through a wavelet mutation strategy, wherein the mathematical model of the final parameter individuals is represented as follows:
Figure BDA0003188574380000034
in the formula, the intermediate variable d of the wavelet function values θ and θ is expressed as:
Figure BDA0003188574380000041
Figure BDA0003188574380000042
wherein o is a random number between (-2.5d,2.5 d); g is 10000; t and tmRespectively representing the current iteration times and the maximum iteration times;
step 6: finding the best candidate solution in the final parametric individual using WOA, setting the probability p to be a random value within [0,1], and the coefficient B expressed as:
B＝(2-2t/tm)·G (8)
wherein G represents a random number on (0,1), then WOA updates population individuals h according to probability p and coefficient Bj(t +1) is:
Figure BDA0003188574380000043
Figure BDA0003188574380000044
wherein j-1 and j-2 represent ω and b, respectively; o represents a random value in the range of (0, 2); c is the helical coefficient, typically 1; h isj f(t) and hj m(t) respectively representing the random individual and the optimal individual iterated to t times;
and 7: by optimizing the population-updated individuals by adding gaussian perturbations, the population-updated individuals are optimized to:
Figure BDA0003188574380000045
in the formula, η is a random value in (0, 1); gussat(μ,σ2) Is a Gaussian disturbance term after t iterations, obeys (mu, sigma)2) (ii) a gaussian distribution of; mu is 0, and sigma is equal to hj m(t)；
And 8: calculating the fitness value of the population individuals and updating the optimal parameter omegakAnd bk；
And step 9: the current iteration time t reaches the maximum iteration time tmaxIf yes, executing step 10; otherwise, returning to the step 8 if t is t + 1;
step 10: finishing the optimization of the HWOA for an input weight matrix and an offset matrix of the ELM, introducing the obtained optimal parameter matrix into the ELM, and constructing an error declination identification model based on the HWOA-ELM;
step 11: identification of test set { F Using HWOA-ELM2,Y2And comparing the identification label with the actual label to obtain the classification precision, thereby drawing a classification result graph.
The invention has the following beneficial effects:
the HWOA algorithm introduces various strategies, and the convergence speed and the convergence precision of the HWOA algorithm are greatly improved. Chaotic mapping and wavelet variation strategies are introduced, and distribution of WOA initial parameter populations is optimized; opposite learning strategies are introduced, and the convergence speed of the algorithm is improved; and a Gaussian disturbance strategy is introduced, so that the ability of the WOA to escape from the local optimum is enhanced.
The HWOA algorithm optimizes the input weight and bias of the ELM, so that the classification accuracy of the constructed error deflection angle recognition model clamped at the tail end of the mechanical arm is higher.
The HWOA algorithm is a parameter matrix for optimizing ELM in an off-line state, which does not influence the online classification speed of the error drift angle identification model.
Drawings
FIG. 1 is an error declination identification model based on HWOA-ELM.
Fig. 2 is a force signal for different angles.
Fig. 3 is a diagram showing the classification result of the error deflection angle of the end clamping of the robot arm.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
Referring to fig. 1 to 3, an error deflection angle identification method for mechanical arm end clamping based on HWOA-ELM includes the following steps:
step 1: the robot selects Mitsubishi industrial robot RV-2F, through arm centre gripping axle class part, the patchhole is in order to realize the assembly. The fixed part is in the vertical direction, the deflection angle is changed by using the tail end position signal, the force deflection angle is set to be 1 degree, 2 degrees, 3 degrees and 4 degrees, and the force data signals F (F is equal to { F) corresponding to different error deflection angles are acquired by using the six-dimensional force sensor 4F-FS001-W200xi,Fyi,Fzi,Mxi,Myi,MziAt a sampling frequency of 141Hz, wherein Fxi,Fyi,FziRepresenting forces in the x, y, z directions, Mxi,Myi,MziRepresenting the moments along x, Y, z, and combining the force data signal F and the number Y of the corresponding deflection angle into an error deflection angle data set, which is further subdivided into a training set { F1,Y1And test set F2,Y2}；
Step 2: carrying out initialization setting of parameters including the hidden layer neuron number 50 of the ELM and the input weight omegakAnd input layer to hidden layer bias bkAll search ranges of [ -1,1 [)]The number of population M of HWOA is 100, the number of iterations is t, and the maximum number of iterations is tmax＝100；
And step 3: generating initial parameter population individual h of ELM by using logic mapping strategyk,jWherein j is 1 or 2, hk,1＝ωk,hk,2＝bk. k represents the current individual of the parameter population, and the population individual of the parameter after logical mapping is represented as:
Figure BDA0003188574380000061
Figure BDA0003188574380000062
in the formula, r1、r2、rk-1Random numbers that are all 0 to 1; omegamaxAnd bmaxAre all 1, omegaminAnd bminAre all-1;
and 4, step 4: and (3) introducing an opposite learning strategy to obtain opposite initialized population individuals, then calculating the fitness value of each group of individuals of all the populations, and selecting the optimal individual of each group to form a new parameter population. The logic mapping optimized ELM initial parameter individual is expressed as:
Hk,j＝hmax,j+hmin,j-hk,j (3)
Figure BDA0003188574380000071
in the formula, Hk,jIs an opponent individual; when j is 1, hmax,jAnd hmin,jIs omegamaxAnd ωmin(ii) a When j is 2, hmax,jAnd hmin,jIs b ismaxAnd bmin；V(Hk,j) And V (h)k,j) The fitness function is represented as 1 in the embodiment, and the difference value of the precision is obtained after the parameter is input into the classifier, wherein the lower the fitness value is, the better the parameter individual is represented; if the generated opponent fitness value is lower than that of the original individual, selecting the opponent as a new population individual, otherwise, keeping the opponent fitness value unchanged;
and 5: calculating the average value of fitness of all the obtained individuals, further optimizing the parameter individuals with fitness values lower than the average fitness through a wavelet variation strategy to further improve population richness, and expressing the mathematical model of the final parameter individuals as follows:
Figure BDA0003188574380000072
in the formula, the intermediate variable d of the wavelet function values θ and θ is expressed as:
Figure BDA0003188574380000073
Figure BDA0003188574380000074
wherein o is a random number between (-2.5d,2.5 d); g is 10000; t and tmRespectively representing the current iteration times and the maximum iteration times;
step 6: finding the best candidate solution in the final parametric individual using WOA, setting the probability p to be a random value within [0,1], and the coefficient B expressed as:
B＝(2-2t/tm)·G (8)
wherein G represents a random number on (0,1), then WOA updates population individuals h according to probability p and coefficient Bj(t +1) is:
Figure BDA0003188574380000081
Figure BDA0003188574380000082
wherein j-1 and j-2 represent ω and b, respectively; o represents a random value in the range of (0, 2); c is the helical coefficient, typically 1; h isj f(t) and hj m(t) are respectively the random individual and the optimal individual iterated to t times.
And 7: the population updating individuals are optimized by increasing Gaussian disturbance, and the influence of random individuals on the population updating trend is improved, so that the algorithm is easier to jump out of local optimality, and then the population updating individuals are optimized as follows:
Figure BDA0003188574380000083
in the formula, η is a random value in (0, 1); gussat(μ,σ2) Is a Gaussian disturbance term after t iterations, obeys (mu, sigma)2) (ii) a gaussian distribution of; mu is 0, and sigma is equal to hj m(t)；
And 8: calculating the fitness value of the population individuals and updating the optimal parameter omegakAnd bk；
And step 9: the current iteration time t reaches the maximum iteration time tmaxIf yes, executing step 10; otherwise, returning to the step 8 if t is t + 1;
step 10: finishing the optimization of the HWOA for an input weight matrix and an offset matrix of the ELM, introducing the obtained optimal parameter matrix into the ELM, and constructing an error declination identification model based on the HWOA-ELM;
step 11: identification of test set { F Using HWOA-ELM2,Y2And comparing the identification label with the actual label to obtain the classification precision, thereby drawing a classification result graph.
Having described specific examples of the present invention in detail above, by combining HWOA and ELM, it is possible to identify the error deflection angle of the robot arm tip grip more accurately while ensuring fast identification of the error deflection angle. It is obvious that the invention is not limited to the above examples. All modifications attainable by one versed in the art from the present concepts by logical reasoning, analysis and other experiments herein are to be considered as within the scope of the present invention.

Claims (1)

1. An error deflection angle identification method for mechanical arm end clamping based on HWOA-ELM is characterized by comprising the following steps:
step 1: through arm centre gripping part, fixed part is vertical direction, utilizes terminal position signal to change the declination size, assembles again, gathers the power data signal F that different error declinations correspond ═ Fxi,Fyi,Fzi,Mxi,Myi,MziIn which Fxi,Fyi,FziRepresenting forces in the x, y, z directions, Mxi,Myi,MziWhich represents the position of the strip along x,y, z, forming an error deflection angle data set by the force data signal F and the number Y of the corresponding deflection angle, and further subdividing the error deflection angle data set into a training set { F1,Y1And test set F2,Y2}；
Step 2: carrying out initialization setting of parameters including hidden layer neuron number N of ELM and input weight omegakAnd input layer to hidden layer bias bkSearch range [ omega ] ofmax,ωmin]、[bmax,bmin]The population number M of HWOA, the iteration number is t, and the maximum iteration number is tmax；
And step 3: generating initial parameter population individual h of ELM by using logic mapping strategyk,jWherein j is 1 or 2, hk,1＝ωk,hk,2＝bkAnd k represents the current individual of the parameter population, and the population individual of the parameter after logical mapping is represented as:
Figure FDA0003188574370000011
Figure FDA0003188574370000012
in the formula, r1、r2、rk-1Random numbers that are all 0 to 1;
and 4, step 4: the individual ELM initial parameters for introducing the opposite learning strategy optimization logical mapping are expressed as follows:
Hk,j＝hmax,j+hmin,j-hk,j (3)
Figure FDA0003188574370000021
in the formula, Hk,jIs an opponent individual; when j is 1, hmax,jAnd hmin,jIs omegamaxAnd ωminWhen j is 2, hmax,jAnd hmin,jIs b ismaxAnd bmin；V(Hk,j) And V (h)k,j) The fitness function is used as the quality of the evaluation parameter individual;
and 5: and further optimizing the parameter individuals with the fitness values lower than the average fitness through a wavelet mutation strategy, and expressing the mathematical model of the final parameter individuals as follows:
Figure FDA0003188574370000022
in the formula, the intermediate variable d of the wavelet function values θ and θ is expressed as:
Figure FDA0003188574370000023
Figure FDA0003188574370000024
wherein o is a random number between (-2.5d,2.5 d); g is 10000; t and tmRespectively representing the current iteration times and the maximum iteration times;
step 6: finding the best candidate solution in the final parameter individual using WOA, setting the probability p to be a random value within [0,1], and the coefficient B is expressed as:
B＝(2-2t/tm)·G (8)
where G represents a random number on (0,1), then the WOA updates the population of individuals h according to a probability p and a coefficient Bj(t +1) is:
Figure FDA0003188574370000025
Figure FDA0003188574370000026
wherein j-1 and j-2 represent ω and b, respectively; o represents a random value in the range of (0, 2); c is the helical coefficient, typically 1; h isj f(t) and hj m(t) respectively representing the random individual and the optimal individual iterated to t times;
and 7: by optimizing the population-updated individuals by adding gaussian perturbations, the population-updated individuals are optimized to:
Figure FDA0003188574370000031
in the formula, η is a random value in (0, 1); gussat(μ,σ2) Is a Gaussian disturbance term after t iterations, obeys (mu, sigma)2) (ii) a gaussian distribution of; mu is 0, and sigma is equal to hj m(t)；
And 8: calculating the fitness value of the population individuals and updating the optimal parameter omegakAnd bk；
And step 9: the current iteration time t reaches the maximum iteration time tmaxIf yes, executing step 10; otherwise, returning to the step 8 if t is t + 1;
step 10: finishing the optimization of the HWOA for an input weight matrix and an offset matrix of the ELM, introducing the obtained optimal parameter matrix into the ELM, and constructing an error declination identification model based on the HWOA-ELM;
step 11: identification of test set { F Using HWOA-ELM2,Y2And comparing the identification label with the actual label to obtain the classification precision, thereby drawing a classification result graph.
CN202110876485.0A 2021-07-30 2021-07-30 HWOA-ELM-based error deflection angle identification method for mechanical arm end clamping Pending CN113743462A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110876485.0A CN113743462A (en) 2021-07-30 2021-07-30 HWOA-ELM-based error deflection angle identification method for mechanical arm end clamping

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110876485.0A CN113743462A (en) 2021-07-30 2021-07-30 HWOA-ELM-based error deflection angle identification method for mechanical arm end clamping

Publications (1)

Publication Number Publication Date
CN113743462A true CN113743462A (en) 2021-12-03

Family

ID=78729758

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110876485.0A Pending CN113743462A (en) 2021-07-30 2021-07-30 HWOA-ELM-based error deflection angle identification method for mechanical arm end clamping

Country Status (1)

Country Link
CN (1) CN113743462A (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110351241A (en) * 2019-05-31 2019-10-18 浙江工业大学 A kind of industrial network DDoS intruding detection system classification method based on GWA optimization
CN110363214A (en) * 2019-05-31 2019-10-22 浙江工业大学 A kind of contact condition recognition methods of the robotic asssembly based on GWA-SVM
CN112613493A (en) * 2021-01-11 2021-04-06 桂林电子科技大学 Rolling bearing fault diagnosis method based on multi-feature extraction and WOA-ELM

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110351241A (en) * 2019-05-31 2019-10-18 浙江工业大学 A kind of industrial network DDoS intruding detection system classification method based on GWA optimization
CN110363214A (en) * 2019-05-31 2019-10-22 浙江工业大学 A kind of contact condition recognition methods of the robotic asssembly based on GWA-SVM
CN112613493A (en) * 2021-01-11 2021-04-06 桂林电子科技大学 Rolling bearing fault diagnosis method based on multi-feature extraction and WOA-ELM

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
宋飞 等: "《融合多策略的鸟群算法及油层识别ELM模型优化》", 《计算机工程与应用》, vol. 58, no. 09, pages 279 - 287 *
张淑清 等: "《基于 ICEEMD 及 AWOA 优化 ELM 的机械故障诊断方法》", 《仪器仪表学报》, vol. 40, no. 11, pages 173 - 177 *

Similar Documents

Publication Publication Date Title
US11741356B2 (en) Data processing apparatus by learning of neural network, data processing method by learning of neural network, and recording medium recording the data processing method
Pirlo et al. Fuzzy-zoning-based classification for handwritten characters
CN112085252B (en) Anti-fact prediction method for set type decision effect
CN105425583B (en) The control method of penicillin production process based on coorinated training LWPLS
CN110363214B (en) GWA-SVM-based contact state recognition method for robot assembly
Zhang et al. Fast covariance matching with fuzzy genetic algorithm
CN114417720B (en) Multi-model predictive control method for pichia pastoris fermentation process
CN113012766B (en) Self-adaptive soft measurement modeling method based on online selective integration
CN110000781B (en) Development network-based mobile robot motion direction pre-decision method
Burke An unsupervised neural network approach to tool wear identification
CN116560313A (en) Genetic algorithm optimization scheduling method for multi-objective flexible job shop problem
Fukushima et al. Deep convolutional network neocognitron: improved interpolating-vector
CN113743462A (en) HWOA-ELM-based error deflection angle identification method for mechanical arm end clamping
CN109670538B (en) Method for identifying assembly contact state of non-rigid part
Yu et al. A dynamic multi-objective evolutionary algorithm based on polynomial regression and adaptive clustering
CN116665039A (en) Small sample target identification method based on two-stage causal intervention
Liu et al. Facial expression recognition approach based on least squares support vector machine with improved particle swarm optimization algorithm
Kitz et al. Neural dynamic assembly sequence planning
CN110210454B (en) Human body action pre-judging method based on data fusion
Elfelly et al. A multimodel approach for complex systems modeling based on classification algorithms
CN114372181A (en) Intelligent planning method for equipment production based on multi-mode data
CN110674883A (en) Active learning method based on k nearest neighbor and probability selection
CN113360851B (en) Industrial flow line production state detection method based on Gap-loss function
Yao et al. Design Optimization for Bellow Soft Pneumatic Actuators in Shape-Matching
Fukushima Add-if-silent rule for training multi-layered convolutional network neocognitron

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