WO2020140560A1 - 一种生产物流输送装备故障预警方法 - Google Patents

一种生产物流输送装备故障预警方法 Download PDF

Info

Publication number
WO2020140560A1
WO2020140560A1 PCT/CN2019/112249 CN2019112249W WO2020140560A1 WO 2020140560 A1 WO2020140560 A1 WO 2020140560A1 CN 2019112249 W CN2019112249 W CN 2019112249W WO 2020140560 A1 WO2020140560 A1 WO 2020140560A1
Authority
WO
WIPO (PCT)
Prior art keywords
early warning
production logistics
logistics transportation
transportation equipment
neuron
Prior art date
Application number
PCT/CN2019/112249
Other languages
English (en)
French (fr)
Inventor
钱晓明
楼佩煌
王鑫豪
Original Assignee
南京航空航天大学
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 南京航空航天大学 filed Critical 南京航空航天大学
Priority to US16/968,145 priority Critical patent/US11740619B2/en
Publication of WO2020140560A1 publication Critical patent/WO2020140560A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • 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
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00—Testing or monitoring of control systems or parts thereof
    • G05B23/02—Electric testing or monitoring
    • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
    • G05B23/0243—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
    • G05B23/0245—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model based on a qualitative model, e.g. rule based; if-then decisions
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00—Testing or monitoring of control systems or parts thereof
    • G05B23/02—Electric testing or monitoring
    • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
    • G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
    • G05B23/024—Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/088—Non-supervised learning, e.g. competitive learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • G06N5/043—Distributed expert systems; Blackboards
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • G06Q10/083—Shipping
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2223/00—Indexing scheme associated with group G05B23/00
    • G05B2223/02—Indirect monitoring, e.g. monitoring production to detect faults of a system

Definitions

  • the invention belongs to the technical application field of intelligent systems, and particularly relates to a method for early warning of production logistics transportation equipment failure.
  • the system can give real-time early warning of possible equipment failures, reduce production losses caused by equipment shutdown, and improve production efficiency.
  • the structure of smart equipment is more and more complicated, there are more and more parts, and the failure of a single part may cause a series of failures, so the early warning method of production logistics transportation equipment failure has strong practical significance.
  • the document "Initial Fault Detection and Condition Monitoring of Rolling Bearings Based on Martin System [Master Thesis], Lanzhou, Lanzhou University of Technology, 2016” analyzes the fault diagnosis technology of bearings.
  • the fault diagnosis technology can detect the fault type and Fault source, patent "A device fault early warning method and device, Chinese patent: CN109087008A, 2018-12-25” decomposes at least two long-term change trend values from the time series; the decomposed long-term change trend value Linear regression fitting of the values to obtain the fitting curve; according to the fitting curve and the preset warning value, determine the failure warning time point corresponding to the to-be-detected indicator.
  • Intelligent early warning method Chinese patent: CN109002031A, 2018-12-14" based on the relationship between different alarm events, establish an alarm cascade group, with the occurrence of alarm events as a trigger condition, automatically determine the same cascade group alarm within a certain time Whether they exist at the same time and generate correlation information between alarm events, however, they cannot assess the global status or performance of the device.
  • condition assessment is crucial. It not only reflects the overall degree of degradation of the equipment, but also provides a reference for the enterprise, but also provides the necessary basis for the next step of prediction and health management.
  • the present invention aims to provide a method for early warning of production logistics transportation equipment failure, to overcome the defects of the existing state diagnosis technology, and to realize the failure warning of production logistics transportation equipment.
  • the technical solution of the present invention is:
  • a method for early warning of production logistics transportation equipment failure includes the following steps:
  • Step 1 Calculate the feature vector of the historical normal operating state, divide the normal state data into various working conditions, obtain several cluster centers, and calculate the Euclidean distance between the current state and the cluster center to obtain the similarity trend;
  • Step 2 Construct the historical memory matrix, optimize the parameters of the LS-SVM regression model by improving the particle swarm algorithm, and calculate the residual values of the current state and the regression model;
  • Step three by combining the similarity trend and the residual value, a risk coefficient is obtained, the operation status of the equipment is evaluated, and a timely warning is given to the failure.
  • Step 1.1 initialization: create two nodes with weight vectors and zero value of local error
  • Step 1.2 input a vector to the neural network x, find the two neurons s and t closest to x, namely the nodes with weight vectors w s and w t ,
  • Step 1.3 Update the local error of the winner neuron s and add it to the squared distance between the vector w s and x:
  • step 1.4 the winner neuron s and all its topological neighbors are translated, the direction is the input vector x, and the distance is equal to the part ⁇ w and the entire ⁇ n :
  • Step 1.5 in steps of 1, increase the age of all connections coming out of the winner neuron s, and remove connections older than age max ; if this result in the neuron has no more divergent edges, then these Neuron removal
  • Step 1.6 If the current iteration number is a multiple of ⁇ and the limit size of the network has not been reached, insert a new neuron r as follows;
  • Step 1.7 Use fractional ⁇ to reduce all errors of neuron j
  • Step 1.8 If the stop condition is not met, continue to step two.
  • step two uses an improved particle swarm optimization algorithm to optimize the kernel function ⁇ and penalty coefficient c in the LS-SVM regression model.
  • step three is as follows:
  • Step 3.1 Calculate the residual value r i of the current state
  • Step 3.2 Calculate the similarity trend t i of the current state
  • Step 3.3 Calculate the risk factor d i .
  • step 1.6 is as follows:
  • Step 1.6.1 Determine the neuron u with a maximum local error
  • Step 1.6.2 Determine the neuron v with a maximum error in the nearest neighbors
  • Step 1.6.3 Create a "centered" node r between u and v:
  • Step 1.6.4 Replace the edge between u and v with the edge between u and r, v and r;
  • Step 1.6.5 Reduce the error of neurons u and v, set the error value of neuron r
  • step two specifically is:
  • Step 2.1.2 check whether the historical best fitness P b meets the constraint condition or whether the number of iterations reaches the maximum, if the constraint condition is still not met and the number of iterations is not the maximum, then proceed to step 2.1.3, otherwise map the result to LS -Kernel function ⁇ and penalty coefficient c of SVM model;
  • Step 2.1.3 adjust the speed and position of the particles, adjust the inertia weight.
  • step 2.1.3 the adaptive inertia weight method is used to adjust the inertia weight:
  • w min and w max are the minimum and maximum values of w; f is the fitness of the current particles, and f avg and f min are the average and minimum fitness values of all particles, respectively.
  • step 3.1 the specific process of calculating the residual value r i of the current state in step 3.1 is as follows:
  • y i is the true value in the sample set
  • f(x i ) is the predicted value of the LS-SVM regression model optimized by the improved particle swarm optimization algorithm.
  • step 3.3 the specific process of calculating the similarity trend t i of the current state in step 3.3 is as follows:
  • x i is the coordinate of the current state
  • X j is the coordinate of the jth cluster center.
  • step 3.3 the specific process of calculating the risk coefficient d i in step 3.3 is as follows:
  • a and b are weight factors, according to historical data, initialized to 0.5, 0.5.
  • the translation winner neuron s and all its topological neighbors refer to all neurons connected to the winner neuron s.
  • step 1.5 if the two best neurons s and t are connected, the age of the connection is set to zero, otherwise a connection is created between them.
  • a growth neural gas GNG algorithm is used to calculate the feature vector of the historical normal operating state.
  • the historical signal data obtained by the sensor is first subjected to feature extraction and dimensionality reduction processing to obtain a feature vector.
  • the growth neural gas (GNG) algorithm is used to divide the normal state data into a variety of working conditions to obtain a number of cluster centers, and calculate the Euclidean distance between the feature vector obtained from the current operating data and the cluster center
  • the similarity trend is obtained;
  • the historical memory matrix is constructed, and the parameter of the LS-SVM regression model is optimized by the improved particle swarm algorithm to calculate the residual value of the current state.
  • the risk coefficient is obtained, the equipment status is evaluated, and the equipment failure is pre-warned in advance.
  • FIG. 1 is a flowchart of an embodiment of the present invention.
  • the equipment used in the production of the automobile assembly line is taken as an example to explain the method for warning the failure of the transportation equipment for automobile assembly in the present invention.
  • the steps are as follows.
  • Step 1 Data collection: Use sensors to collect status data of the main equipment parts used in production, including vibration and acceleration signals of two bearings and reducers, belt displacement, etc.;
  • Step 2 Extract feature parameters: For different data, different feature extraction techniques are used for feature extraction.
  • Step 3 Data dimensionality reduction: take the average value of the effective value and peak value of vibration removal, and then synthesize a feature vector so that the dimension of the feature vector is 7; repeat the above steps to obtain multiple feature vectors with dimension 7;
  • Step 4 Construction of the growth neural gas (GNG) neural network model: using the growth neural gas (GNG) algorithm on the data of the historical normal operating state, the normal state data is divided into various working conditions, and several clustering centers are obtained. And calculate the Euclidean distance between the current state and the cluster center to get the similarity trend;
  • Step 5 LS-SVM regression model construction: construct a historical memory matrix, optimize the parameters of the LS-SVM regression model by improving the particle swarm algorithm, and calculate the residual value of the current state and the regression model;
  • Step 6 Calculation of risk coefficient: By combining the similarity trend and residual value, the risk coefficient is obtained, the operation status of the equipment is evaluated, and timely warning is given to the fault.
  • a method for early warning of production logistics transportation equipment failure includes the following steps:
  • Step 1 Calculate the feature vector of the historical normal operating state, divide the normal state data into various working conditions, obtain several cluster centers, and calculate the Euclidean distance between the current state and the cluster center to obtain the similarity trend;
  • Step 2 Construct the historical memory matrix, optimize the parameters of the LS-SVM regression model by improving the particle swarm algorithm, and calculate the residual values of the current state and the regression model;
  • Step three by combining the similarity trend and the residual value, a risk coefficient is obtained, the operation status of the equipment is evaluated, and a timely warning is given to the failure.
  • Step 1.1 initialization: create two nodes with weight vectors and zero value of local error
  • Step 1.2 input a vector to the neural network x, find the two neurons s and t closest to x, namely the nodes with weight vectors w s and w t ,
  • Step 1.3 Update the local error of the winner neuron s and add it to the squared distance between the vector w s and x:
  • step 1.4 the winner neuron s and all its topological neighbors are translated, the direction is the input vector x, and the distance is equal to the part ⁇ w and the entire ⁇ n :
  • Step 1.5 in steps of 1, increase the age of all connections coming out of the winner neuron s, and remove connections older than age max ; if this result in neurons has no more divergent edges, then these Neuron removal
  • Step 1.6 If the current iteration number is a multiple of ⁇ and the limit size of the network has not been reached, insert a new neuron r as follows;
  • Step 1.7 Use fractional ⁇ to reduce all errors of neuron j
  • Step 1.8 If the stop condition is not met, continue to step two.
  • step two uses an improved particle swarm optimization algorithm to optimize the kernel function ⁇ and penalty coefficient c in the LS-SVM regression model.
  • step three is as follows:
  • Step 3.1 Calculate the residual value r i of the current state
  • Step 3.2 Calculate the similarity trend t i of the current state
  • Step 3.3 Calculate the risk factor d i .
  • step 1.6 is as follows:
  • Step 1.6.1 Determine the neuron u with a maximum local error
  • Step 1.6.2 Determine the neuron v with a maximum error in the nearest neighbors
  • Step 1.6.3 Create a "centered" node r between u and v:
  • Step 1.6.4 Replace the edge between u and v with the edge between u and r, v and r;
  • Step 1.6.5 Reduce the error of neurons u and v, set the error value of neuron r
  • step two specifically is:
  • Step 2.1.2 check whether the historical best fitness P b meets the constraint condition or whether the number of iterations reaches the maximum, if the constraint condition is still not met and the number of iterations is not the maximum, then proceed to step 2.1.3, otherwise map the result to LS -Kernel function ⁇ and penalty coefficient c of SVM model;
  • Step 2.1.3 adjust the speed and position of the particles, adjust the inertia weight.
  • step 2.1.3 the adaptive inertia weight method is used to adjust the inertia weight:
  • w min and w max are the minimum and maximum values of w; f is the fitness of the current particles, and f avg and f min are the average and minimum fitness values of all particles, respectively.
  • step 3.1 the specific process of calculating the residual value r i of the current state in step 3.1 is as follows:
  • y i is the true value in the sample set
  • f(x i ) is the predicted value of the LS-SVM regression model optimized by the improved particle swarm optimization algorithm.
  • step 3.3 the specific process of calculating the similarity trend t i of the current state in step 3.3 is as follows:
  • x i is the coordinate of the current state
  • X j is the coordinate of the jth cluster center.
  • step 3.3 the specific process of calculating the risk coefficient d i in step 3.3 is as follows:
  • a and b are weight factors, according to historical data, initialized to 0.5, 0.5.
  • the translation winner neuron s and all its topological neighbors refer to all neurons connected to the winner neuron s.
  • step 1.5 if the two best neurons s and t are connected, the age of the connection is set to zero, otherwise a connection is created between them.
  • a growth neural gas GNG algorithm is used to calculate the feature vector of the historical normal operating state.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Strategic Management (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Marketing (AREA)
  • Development Economics (AREA)
  • Operations Research (AREA)
  • General Business, Economics & Management (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Automation & Control Theory (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本发明公开了一种生产物流输送装备故障预警方法。在将传感器获得的历史信号数据进行特征提取和降维处理，获得特征向量后，一方面运用生长型神经气(GNG)算法，将正常状态数据划分为多种工况，得到若干聚类中心，并计算由当前运行数据得到的特征向量与聚类中心的欧式距离从而得到相似度趋势；另一方面构建历史记忆矩阵，通过改进粒子群算法优化LS-SVM回归模型参数，计算当前状态的残差值。最终结合残差值与相似度趋势，得出风险系数，对装备状态做出评估，并对装备故障作出提前预警。本方法实现了生产物流输送装备的故障实时预警技术，为设备及时维护提供了参考，避免了因为设备故障停机造成的经济损失。

Description

一种生产物流输送装备故障预警方法 技术领域
本发明属于智能系统技术应用领域，特别涉及了一种生产物流输送装备故障预警方法。
背景技术
当前，智能装备在生产车间的应用越来越广泛，智能装备零部件多，内部构造复杂，依赖传统的人工经验去判断设备运行状态和做出故障预警已经不具备可行性。同时智能传感器采集到的生产数据，装备运行数据越来越精确，但是目前大部分数据只是被存储到数据库中，却没有利用起来。汽车总装车间，生产节拍快，产量大，汽车总装输送装备发生故障停机会造成巨大的经济损失。
为此，实现一套自动化的设备故障实时预警系统是非常迫切的需要。该系统可以实时对设备可能产生的故障做出预警，降低了因设备停机造成的生产损失，提高了生产效率。现在的智能设备结构越来越复杂，零部件越来越多，并且单个零件出现问题可能连锁导致一系列的故障，所以生产物流输送装备故障预警方法具备很强的现实意义。文献“基于马田系统的滚动轴承初始故障检测和状态监测[硕士学位论文]，兰州，兰州理工大学，2016”分析了轴承的故障诊断技术，对于机械生产设备，故障诊断技术可以探测到故障类型和故障源，专利“一种设备故障预警方法及装置，中国专利：CN109087008A，2018-12-25”从所述时间序列中分解出至少两个长期变化趋势数值；对分解出的所述长期变化趋势数值进行线性回归拟合,获取拟合曲线；根据所述拟合曲线和预设的预警值,确定所述待检测指标对应的故障预警时间点，专利“一种应用于监测系统设备故障诊断及智能预警的方法，中国专利：CN109002031A，2018-12-14”根据不同报警事件之间的关系,建立报警级联组,以报警事件产生为触发条件,自动判断同一级联组报警在一定时间内是否同时存在,产生报警事件之间关联信息，但是，它们不能评估设备的全局状态或性能。为了提高安全性和可靠性、状态评估是至关重要的。它不仅反映了设备的全局退化程度，为企业提供参考，同时也为下一步的预测和健康管理提供了必要的依据。
但是，现有的状态评估的研究主要集中在零件或部件单元，如轴承和一些电子系统，对于机械设备健康状态的全局评估缺乏充分的研究。考虑到机械设备的复杂性，反映设备的健康状态需要基于零件和部件来展开。由于每个零部件在一个设备中的重要性是不同的，从传感器收集到的状态特征应该给予不同的权重。但是当前对于状态评估的研究，缺乏权重决策的方法。常见的方法就是根据经验给予权重，但这些权重并不能够反映属性数据的变化率。
发明内容
为了解决上述背景技术提出的技术问题，本发明旨在提供一种生产物流输送装备故障预警 方法，克服现有状态诊断技术存在的缺陷，实现生产物流输送装备的故障预警。
为了实现上述技术目的，本发明的技术方案为：
一种生产物流输送装备故障预警方法，包括以下步骤：
步骤一，计算历史正常运行状态的特征向量，将正常状态数据划分为多种工况，得到若干聚类中心，并计算当前状态与聚类中心的欧式距离从而得到相似度趋势；
步骤二，构建历史记忆矩阵，通过改进粒子群算法优化LS-SVM回归模型参数，计算当前状态与回归模型的残差值；
步骤三，通过结合相似度趋势及残差值，得出风险系数，评估设备运行状态，并对故障作出及时预警。
进一步的，所述步骤一的具体过程如下：
步骤1.1，初始化：创建两个带权重向量的节点，以及局部误差的零值；
步骤1.2，向神经网络x输入一个向量，在最接近x的地方找到s和t两个神经元，即带有权重向量w s与w t的节点，||w s-x|| 2是所有节点中距离值最小、而||w t-x|| 2是第二小；
步骤1.3，更新赢家神经元s的局部误差，将其添加到向量w s与x的平方距离：
E s←E s+||w s-x|| 2         (1)
步骤1.4，平移赢家神经元s及其所有拓扑近邻点，方向是输入向量x，距离则等于部分∈ w和整个∈ n：
w s←w s+∈ w·(w s-x)        (2)
w n←w n+∈ n·(w n-x)        (3)
步骤1.5，以1为步幅，增加从赢家神经元s出来的所有连接的年龄，将年龄大于age max的连接移除；如果神经元中的这个结果没有更多的发散边缘，则亦将这些神经元移除；
步骤1.6，如果当前迭代的数量是λ的倍数，且尚未达到网络的限制尺寸，则如下插入一个新的神经元r；
步骤1.7，利用分式β减少神经元j的所有误差
E j←E j-E j·β            (4)
步骤1.8，如果未能满足停止条件，则继续步骤二。
进一步的，步骤二运用改进粒子群算法对LS-SVM回归模型中的核函数σ和惩罚系数c作 出优化。
更进一步的，步骤三的具体过程如下：
步骤3.1计算当前状态的残差值r i，
步骤3.2计算当前状态的相似度趋势t i，
步骤3.3计算风险系数d i。
更进一步的，步骤1.6的具体过程如下：
步骤1.6.1确定带有一个最大局部误差的神经元u；
步骤1.6.2于近邻点中确定u带有一个最大误差的神经元v；
步骤1.6.3于u和v中间创建一个“居中”的节点r：
Figure PCTCN2019112249-appb-000001
步骤1.6.4用u与r、v以及r之间的边，替代u与v之间的边；
步骤1.6.5减少神经元u与v的误差，设置神经元r的误差值
E u←E u·a                              (7)
E v←E v·a                              (8)
E r←E u                                (9)。
更进一步的，步骤二具体为：
步骤2.1.1，构建LS-SVM回归模型：引入拉格朗日函数对其求解，选择径向基函数K(x,x i)＝exp(-||x-x i|| 2/2σ 2)，其中σ为核宽度；整得到LS-SVM回归模型为：
Figure PCTCN2019112249-appb-000002
步骤2.1.2，检查历史最佳适应度P b是否满足约束条件或者迭代次数是否达到最大，如果仍未满足约束条件并且迭代次数不是最大，则进行步骤步骤2.1.3，否则将结果映射为LS-SVM模型的核函数σ和惩罚系数c；
步骤2.1.3，调整粒子的速度与位置，调整惯性权重。
更进一步的，步骤2.1.3中，运用自适应调整的惯性权重法，调整惯性权重：
Figure PCTCN2019112249-appb-000003
式中：w min、w max分别为w的最小值和最大值；f为当前粒子的适应度，f avg、f min分别为所有粒子的平均适应值和最小适应值。
更进一步的，步骤3.1计算当前状态的残差值r i的具体过程如下：
r i＝y i-f(x i)          (11)
式中：y i为样本集中的真实值，f(x i)为改进粒子群算法优化后的LS-SVM回归模型预测值。
更进一步的，步骤3.3计算当前状态的相似度趋势t i的具体过程如下：
Figure PCTCN2019112249-appb-000004
式中：x i为当前状态的坐标，X j为第j个聚类中心的坐标。
更进一步的，步骤3.3计算风险系数d i的具体过程如下：
d i＝ar i+bt i          (13)
式中：a和b为权重因子，根据历史数据，初始化为0.5,0.5。
更进一步的，所述平移赢家神经元s及其所有拓扑近邻点指与该赢家神经元s有连接的所有神经元。
作为一种优选，所述步骤1.5中，如果两个最佳神经元s与t已连接，则将其连接的年龄设为零，否则就在它们之间创建一个连接。
作为一种优选，所述步骤一种，运用生长型神经气GNG算法计算历史正常运行状态的特征向量。
采用上述技术方案带来的有益效果：
本发明首先将传感器获得的历史信号数据进行特征提取和降维处理，获得特征向量。对特征向量，一方面运用生长型神经气(GNG)算法，将正常状态数据划分为多种工况，得到若干聚类中心，并计算由当前运行数据得到的特征向量与聚类中心的欧式距离从而得到相似度趋势；另一方面构建历史记忆矩阵，通过改进粒子群算法优化LS-SVM回归模型参数，计算当前状态的残差值。最终结合残差值与相似度趋势，得出风险系数，对装备状态做出评估，并对 装备故障作出提前预警。
附图说明
图1是本发明实施例的流程图。
具体实施方式
以下将结合附图，对本发明的技术方案进行详细说明。
本实施例以汽车装配线生产中使用的装备为例，说明本发明的汽车总装用输送装备故障预警方法，如图1所示，其步骤如下。
步骤1、数据采集：利用传感器对生产中使用的装备主要零部件进行状态数据采集，包括两个轴承和减速机的振动加速度信号，皮带的位移等；
步骤2、提取特征参数：对于不同的数据采用不同的特征提取技术进行特征提取。
步骤3、数据降维：对去振动有效值和峰值取平均值，然后合成一个特征向量，使得特征向量的维数为7；重复上述步骤，得到多个维数为7的特征向量；
步骤4、生长型神经气(GNG)神经网络模型构建：对历史正常运行状态的数据，运用生长型神经气(GNG)算法，将正常状态数据划分为多种工况，得到若干聚类中心，并计算当前状态与聚类中心的欧式距离从而得到相似度趋势；
步骤5、LS-SVM回归模型构建：构建历史记忆矩阵，通过改进粒子群算法优化LS-SVM回归模型参数，计算当前状态与回归模型的残差值；
步骤6、风险系数计算：通过结合相似度趋势及残差值，得出风险系数，评估设备运行状态，并对故障作出及时预警。
一种生产物流输送装备故障预警方法，包括以下步骤：
步骤一，计算历史正常运行状态的特征向量，将正常状态数据划分为多种工况，得到若干聚类中心，并计算当前状态与聚类中心的欧式距离从而得到相似度趋势；
步骤二，构建历史记忆矩阵，通过改进粒子群算法优化LS-SVM回归模型参数，计算当前状态与回归模型的残差值；
步骤三，通过结合相似度趋势及残差值，得出风险系数，评估设备运行状态，并对故障作出及时预警。
进一步的，所述步骤一的具体过程如下：
步骤1.1，初始化：创建两个带权重向量的节点，以及局部误差的零值；
步骤1.2，向神经网络x输入一个向量，在最接近x的地方找到s和t两个神经元，即带有 权重向量w s与w t的节点，||w s-x|| 2是所有节点中距离值最小、而||w t-x|| 2是第二小；
步骤1.3，更新赢家神经元s的局部误差，将其添加到向量w s与x的平方距离：
E s←E s+||w s-x|| 2                               (1)
步骤1.4，平移赢家神经元s及其所有拓扑近邻点，方向是输入向量x，距离则等于部分∈ w和整个∈ n：
w s←w s+∈ w·(w s-x)                      (2)
w n←w n+∈ n·(w n-x)                   (3)
步骤1.5，以1为步幅，增加从赢家神经元s出来的所有连接的年龄，将年龄大于age max的连接移除；如果神经元中的这个结果没有更多的发散边缘，则亦将这些神经元移除；
步骤1.6，如果当前迭代的数量是λ的倍数，且尚未达到网络的限制尺寸，则如下插入一个新的神经元r；
步骤1.7，利用分式β减少神经元j的所有误差
E j←E j-E j·β                            (4)
步骤1.8，如果未能满足停止条件，则继续步骤二。
进一步的，步骤二运用改进粒子群算法对LS-SVM回归模型中的核函数σ和惩罚系数c作出优化。
更进一步的，步骤三的具体过程如下：
步骤3.1计算当前状态的残差值r i，
步骤3.2计算当前状态的相似度趋势t i，
步骤3.3计算风险系数d i。
更进一步的，步骤1.6的具体过程如下：
步骤1.6.1确定带有一个最大局部误差的神经元u；
步骤1.6.2于近邻点中确定u带有一个最大误差的神经元v；
步骤1.6.3于u和v中间创建一个“居中”的节点r：
Figure PCTCN2019112249-appb-000005
步骤1.6.4用u与r、v以及r之间的边，替代u与v之间的边；
步骤1.6.5减少神经元u与v的误差，设置神经元r的误差值
E u←E u·a                              (7)
E v←E v·a                              (8)
E r←E u                                (9)。
更进一步的，步骤二具体为：
步骤2.1.1，构建LS-SVM回归模型：引入拉格朗日函数对其求解，选择径向基函数K(x,x i)＝exp(-||x-x i|| 2/2σ 2)，其中σ为核宽度；整得到LS-SVM回归模型为：
Figure PCTCN2019112249-appb-000006
步骤2.1.2，检查历史最佳适应度P b是否满足约束条件或者迭代次数是否达到最大，如果仍未满足约束条件并且迭代次数不是最大，则进行步骤步骤2.1.3，否则将结果映射为LS-SVM模型的核函数σ和惩罚系数c；
步骤2.1.3，调整粒子的速度与位置，调整惯性权重。
更进一步的，步骤2.1.3中，运用自适应调整的惯性权重法，调整惯性权重：
Figure PCTCN2019112249-appb-000007
式中：w min、w max分别为w的最小值和最大值；f为当前粒子的适应度，f avg、f min分别为所有粒子的平均适应值和最小适应值。
更进一步的，步骤3.1计算当前状态的残差值r i的具体过程如下：
r i＝y i-f(x i)                          (11)
式中：y i为样本集中的真实值，f(x i)为改进粒子群算法优化后的LS-SVM回归模型预测值。
更进一步的，步骤3.3计算当前状态的相似度趋势t i的具体过程如下：
Figure PCTCN2019112249-appb-000008
式中：x i为当前状态的坐标，X j为第j个聚类中心的坐标。
更进一步的，步骤3.3计算风险系数d i的具体过程如下：
d i＝ar i+bt i            (13)
式中：a和b为权重因子，根据历史数据，初始化为0.5,0.5。
更进一步的，所述平移赢家神经元s及其所有拓扑近邻点指与该赢家神经元s有连接的所有神经元。
作为一种优选，所述步骤1.5中，如果两个最佳神经元s与t已连接，则将其连接的年龄设为零，否则就在它们之间创建一个连接。
作为一种优选，所述步骤一种，运用生长型神经气GNG算法计算历史正常运行状态的特征向量。
实施例仅为说明本发明的技术思想，不能以此限定本发明的保护范围，凡是按照本发明提出的技术思想，在技术方案基础上所做的任何改动，均落入本发明保护范围之内。

Claims (13)

  1. 一种生产物流输送装备故障预警方法，其特征在于，包括以下步骤：
    步骤一，计算历史正常运行状态的特征向量，将正常状态数据划分为多种工况，得到若干聚类中心，并计算当前状态与聚类中心的欧式距离从而得到相似度趋势；
    步骤二，构建历史记忆矩阵，通过改进粒子群算法优化LS-SVM回归模型参数，计算当前状态与回归模型的残差值；
    步骤三，通过结合相似度趋势及残差值，得出风险系数，评估设备运行状态，并对故障作出及时预警。
  2. 根据权利要求1所述生产物流输送装备故障预警方法，其特征在于：所述步骤一的具体过程如下：
    步骤1.1，初始化：创建两个带权重向量的节点，以及局部误差的零值；
    步骤1.2，向神经网络x输入一个向量，在最接近x的地方找到s和t两个神经元，即带有权重向量w s与w t的节点，||w s-x|| 2是所有节点中距离值最小、而||w t-x|| 2是第二小；
    步骤1.3，更新赢家神经元s的局部误差，将其添加到向量w s与x的平方距离：
    E s←E s+||w s-x|| 2       (1)
    步骤1.4，平移赢家神经元s及其所有拓扑近邻点，方向是输入向量x，距离则等于部分∈ w和整个∈ n：
    w s←w s+∈ w·(w s-x)        (2)
    w n←w n+∈ n·(w n-x)(3)
    步骤1.5，以1为步幅，增加从赢家神经元s出来的所有连接的年龄，将年龄大于age max的连接移除；如果神经元中的这个结果没有更多的发散边缘，则亦将这些神经元移除；
    步骤1.6，如果当前迭代的数量是λ的倍数，且尚未达到网络的限制尺寸，则如下插入一个新的神经元r；
    步骤1.7，利用分式β减少神经元j的所有误差
    E j←E j-E j·β        (4)
    步骤1.8，如果未能满足停止条件，则继续步骤二。
  3. 根据权利要求1所述生产物流输送装备故障预警方法，其特征在于：步骤二运用改进粒子群算法对LS-SVM回归模型中的核函数σ和惩罚系数c作出优化。
  4. 根据权利要求1所述生产物流输送装备故障预警方法，其特征在于：步骤三的具体过程如下：
    步骤3.1计算当前状态的残差值r i，
    步骤3.2计算当前状态的相似度趋势t i，
    步骤3.3计算风险系数d i。
  5. 根据权利要求2所述生产物流输送装备故障预警方法，其特征在于：步骤1.6的具体过程如下：
    步骤1.6.1确定带有一个最大局部误差的神经元u；
    步骤1.6.2于近邻点中确定u带有一个最大误差的神经元v；
    步骤1.6.3于u和v中间创建一个“居中”的节点r：
    Figure PCTCN2019112249-appb-100001
    步骤1.6.4用u与r、v以及r之间的边，替代u与v之间的边；
    步骤1.6.5减少神经元u与v的误差，设置神经元r的误差值
    E u←E u·a        (7)
    E v←E v·a       (8)
    E r←E u       (9)。
  6. 根据权利要求3所述生产物流输送装备故障预警方法，其特征在于：步骤二具体为：
    步骤2.1.1，构建LS-SVM回归模型：引入拉格朗日函数对其求解，选择径向基函数K(x,x i)＝ex p(-||x-x i|| 2/2σ 2)，其中σ为核宽度；整得到LS-SVM回归模型为：
    Figure PCTCN2019112249-appb-100002
    步骤2.1.2，检查历史最佳适应度P b是否满足约束条件或者迭代次数是否达到最大，如果仍未满足约束条件并且迭代次数不是最大，则进行步骤步骤2.1.3，否则将结果映射为LS-SVM模型的核函数σ和惩罚系数c；
    步骤2.1.3，调整粒子的速度与位置，调整惯性权重。
  7. 根据权利要求6所述生产物流输送装备故障预警方法，其特征在于：步骤2.1.3中，运用自适应调整的惯性权重法，调整惯性权重：
    Figure PCTCN2019112249-appb-100003
    式中：w min、w max分别为w的最小值和最大值；f为当前粒子的适应度，f avg、f min分别为所有粒子的平均适应值和最小适应值。
  8. 根据权利要求4所述生产物流输送装备故障预警方法，其特征在于：步骤3.1计算当前状态的残差值r i的具体过程如下：
    r i＝y i-f(x i)         (11)
    式中：y i为样本集中的真实值，f(x i)为改进粒子群算法优化后的LS-SVM回归模型预测值。
  9. 根据权利要求4所述生产物流输送装备故障预警方法，其特征在于：步骤3.3计算当前状态的相似度趋势t i的具体过程如下：
    Figure PCTCN2019112249-appb-100004
    式中：x i为当前状态的坐标，X j为第j个聚类中心的坐标。
  10. 根据权利要求4所述生产物流输送装备故障预警方法，其特征在于：步骤3.3计算风险系数d i的具体过程如下：
    d i＝ar i+bt i      (13)
    式中：a和b为权重因子，根据历史数据，初始化为0.5,0.5。
  11. 根据权利要求2所述生产物流输送装备故障预警方法，其特征在于：所述平移赢家神经元s及其所有拓扑近邻点指与该赢家神经元s有连接的所有神经元。
  12. 根据权利要求2所述生产物流输送装备故障预警方法，其特征在于：所述步骤1.5中，如果两个最佳神经元s与t已连接，则将其连接的年龄设为零，否则就在它们之间创建一个连接。
  13. 根据权利要求1至12任一项所述生产物流输送装备故障预警方法，其特征在于：所述步骤一种，运用生长型神经气GNG算法计算历史正常运行状态的特征向量。
PCT/CN2019/112249 2019-05-31 2019-10-21 一种生产物流输送装备故障预警方法 WO2020140560A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/968,145 US11740619B2 (en) 2019-05-31 2019-10-21 Malfunction early-warning method for production logistics delivery equipment

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910472866.5A CN110322048B (zh) 2019-05-31 2019-05-31 一种生产物流输送装备故障预警方法
CN201910472866.5 2019-05-31

Publications (1)

Publication Number Publication Date
WO2020140560A1 true WO2020140560A1 (zh) 2020-07-09

Family

ID=68119261

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/112249 WO2020140560A1 (zh) 2019-05-31 2019-10-21 一种生产物流输送装备故障预警方法

Country Status (3)

Country Link
US (1) US11740619B2 (zh)
CN (1) CN110322048B (zh)
WO (1) WO2020140560A1 (zh)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112183999A (zh) * 2020-09-24 2021-01-05 国网山东省电力公司电力科学研究院 一种配电主设备传感器可靠性评估指标特征提取方法
CN112330248A (zh) * 2020-11-02 2021-02-05 上海寻梦信息技术有限公司 物流末端配送方法、装置、计算机设备和存储介质
CN112348120A (zh) * 2020-12-01 2021-02-09 江苏南高智能装备创新中心有限公司 数控机床退化轨迹模式识别方法及装置
CN112859811A (zh) * 2021-01-14 2021-05-28 国网湖南省电力有限公司 火电机组辅机设备运行健康状态实时检测方法、系统及介质
CN113705858A (zh) * 2021-08-02 2021-11-26 西安交通大学 多目标区域的最短路径规划方法、系统、设备及存储介质
CN114897307A (zh) * 2022-04-13 2022-08-12 重庆交通大学 一种基于合作的物流集配网络联盟优化方法
CN116894247A (zh) * 2023-09-07 2023-10-17 贵州正易科技有限公司 一种保护计算机系统安全的方法及系统
CN116933197A (zh) * 2023-09-19 2023-10-24 安徽融兆智能有限公司 一种基于大数据的用电信息采集系统故障判别方法及系统

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109002031B (zh) * 2018-06-26 2020-06-30 卡斯柯信号有限公司 一种应用于监测系统设备故障诊断及智能预警的方法
CN110322048B (zh) 2019-05-31 2023-09-26 南京航空航天大学 一种生产物流输送装备故障预警方法
CN111062664A (zh) * 2019-12-13 2020-04-24 江苏佳利达国际物流股份有限公司 基于svm动态物流大数据预警分析及保护方法
CN111240300A (zh) * 2020-01-07 2020-06-05 国电南瑞科技股份有限公司 基于大数据的车辆健康状态评估模型构建方法
CN111488947B (zh) * 2020-04-28 2024-02-02 深圳力维智联技术有限公司 电力系统设备的故障检测方法与装置
CN111832993B (zh) * 2020-07-13 2023-06-30 深圳市今天国际物流技术股份有限公司 一种仓储物流系统的预测维护方法及相关组件
CN112052573B (zh) * 2020-08-26 2024-02-13 东南大学 基于有限时间观测器的分布式故障识别方法
CN113408371A (zh) * 2021-06-01 2021-09-17 武汉理工大学 一种早期故障诊断方法及装置
CN113570473B (zh) * 2021-06-25 2024-02-09 深圳供电局有限公司 设备故障监测方法、装置、计算机设备和存储介质
CN114088388A (zh) * 2021-12-10 2022-02-25 华润电力技术研究院有限公司 一种齿轮变速箱的故障诊断方法和故障诊断装置

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107506865A (zh) * 2017-08-30 2017-12-22 华中科技大学 一种基于lssvm优化的负荷预测方法及系统
CN109657847A (zh) * 2018-12-06 2019-04-19 华中科技大学 基于粒子群优化支持向量回归的工业生产中故障预测方法
CN110322048A (zh) * 2019-05-31 2019-10-11 南京航空航天大学 一种生产物流输送装备故障预警方法

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE19755133A1 (de) * 1997-12-11 1999-06-24 Siemens Ag Verfahren zur Überwachung von Bearbeitungsanlagen
WO2010011918A2 (en) * 2008-07-24 2010-01-28 University Of Cincinnati Methods for prognosing mechanical systems
CN101399672B (zh) * 2008-10-17 2011-03-02 章毅 一种多神经网络融合的入侵检测方法
CN104331635B (zh) * 2014-11-24 2017-03-29 国网吉林省电力有限公司信息通信公司 电力光纤通信光功率预测的方法
CN106443297B (zh) * 2016-09-07 2018-05-22 江南大学 光伏二极管箝位型三电平逆变器的决策树svm故障诊断方法
CN109002031B (zh) 2018-06-26 2020-06-30 卡斯柯信号有限公司 一种应用于监测系统设备故障诊断及智能预警的方法
CN109087008A (zh) 2018-08-07 2018-12-25 济南浪潮高新科技投资发展有限公司 一种设备故障预警方法及装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107506865A (zh) * 2017-08-30 2017-12-22 华中科技大学 一种基于lssvm优化的负荷预测方法及系统
CN109657847A (zh) * 2018-12-06 2019-04-19 华中科技大学 基于粒子群优化支持向量回归的工业生产中故障预测方法
CN110322048A (zh) * 2019-05-31 2019-10-11 南京航空航天大学 一种生产物流输送装备故障预警方法

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112183999A (zh) * 2020-09-24 2021-01-05 国网山东省电力公司电力科学研究院 一种配电主设备传感器可靠性评估指标特征提取方法
CN112330248A (zh) * 2020-11-02 2021-02-05 上海寻梦信息技术有限公司 物流末端配送方法、装置、计算机设备和存储介质
CN112348120A (zh) * 2020-12-01 2021-02-09 江苏南高智能装备创新中心有限公司 数控机床退化轨迹模式识别方法及装置
CN112859811A (zh) * 2021-01-14 2021-05-28 国网湖南省电力有限公司 火电机组辅机设备运行健康状态实时检测方法、系统及介质
CN112859811B (zh) * 2021-01-14 2022-01-28 国网湖南省电力有限公司 火电机组辅机设备运行健康状态实时检测方法、系统及介质
CN113705858A (zh) * 2021-08-02 2021-11-26 西安交通大学 多目标区域的最短路径规划方法、系统、设备及存储介质
CN113705858B (zh) * 2021-08-02 2023-07-11 西安交通大学 多目标区域的最短路径规划方法、系统、设备及存储介质
CN114897307A (zh) * 2022-04-13 2022-08-12 重庆交通大学 一种基于合作的物流集配网络联盟优化方法
CN116894247A (zh) * 2023-09-07 2023-10-17 贵州正易科技有限公司 一种保护计算机系统安全的方法及系统
CN116894247B (zh) * 2023-09-07 2023-11-14 贵州正易科技有限公司 一种保护计算机系统安全的方法及系统
CN116933197A (zh) * 2023-09-19 2023-10-24 安徽融兆智能有限公司 一种基于大数据的用电信息采集系统故障判别方法及系统
CN116933197B (zh) * 2023-09-19 2023-12-08 安徽融兆智能有限公司 一种基于大数据的用电信息采集系统故障判别方法及系统

Also Published As

Publication number Publication date
US11740619B2 (en) 2023-08-29
CN110322048A (zh) 2019-10-11
US20210041862A1 (en) 2021-02-11
CN110322048B (zh) 2023-09-26

Similar Documents

Publication Publication Date Title
WO2020140560A1 (zh) 一种生产物流输送装备故障预警方法
CN107941537B (zh) 一种机械设备健康状态评估方法
CN102361014B (zh) 大规模半导体制造过程的状态监控与故障诊断方法
JP5439265B2 (ja) 異常検知・診断方法、異常検知・診断システム、及び異常検知・診断プログラム
US20230385186A1 (en) Resilient estimation for grid situational awareness
CN106021826A (zh) 一种基于工况识别和相似性匹配的变工况下航空发动机整机剩余寿命预测方法
CN112415331B (zh) 基于多源故障信息的电网二次系统故障诊断方法
WO2022012137A1 (zh) 风力发电机组的监控方法、系统及计算机可读存储介质
Zheng et al. Fault diagnosis method based on supervised particle swarm optimization classification algorithm
Eickmeyer et al. Data Driven Modeling for System-Level Condition Monitoring on Wind Power Plants.
Yang et al. Fault early warning of wind turbine gearbox based on multi‐input support vector regression and improved ant lion optimization
Velásquez et al. A hybrid machine-learning ensemble for anomaly detection in real-time industry 4.0 systems
Al Bataineh et al. Autoencoder based semi-supervised anomaly detection in turbofan engines
KR20220164976A (ko) 머신러닝 기반 클린룸 관리 시스템 및 방법
WO2024007580A1 (zh) 一种基于混合聚类的电力设备并行故障诊断方法及装置
Ren et al. A clustering-based method for health conditions evaluation of aero-engines
Sepulvene et al. Analysis of machine learning techniques in fault diagnosis of vehicle fleet tracking modules
CN114492578A (zh) 一种基于bp神经网络的电气设备运行数据分类方法
Zhong et al. Research on fusion diagnosis method of thermal fault of marine diesel engine
CN111104971B (zh) 一种基于概率统计和支持向量机的发动机参数检测方法
Fangwei et al. Research and application progress of data mining technology in electric power system
Dai et al. A Novel Fault Diagnosis Method Integrating Reconstruction-based Principal Component Analysis with Probabilistic Neural Network on Condenser
KR102212022B1 (ko) 양수 수차의 건전성 자동 판정 방법 및 이를 위한 시스템
Dai et al. Health Assessment of Complex Product Component Combined with Class Interval Estimation
Yang et al. Denoising SOM: An Improved Fault Diagnosis Method for Quantum Power Communication Network

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19906669

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19906669

Country of ref document: EP

Kind code of ref document: A1