CN112881054A - Hoisting machinery fault diagnosis method and system - Google Patents

Hoisting machinery fault diagnosis method and system Download PDF

Info

Publication number
CN112881054A
CN112881054A CN202110088959.5A CN202110088959A CN112881054A CN 112881054 A CN112881054 A CN 112881054A CN 202110088959 A CN202110088959 A CN 202110088959A CN 112881054 A CN112881054 A CN 112881054A
Authority
CN
China
Prior art keywords
layer
convolution
signal
output
pooling
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110088959.5A
Other languages
Chinese (zh)
Other versions
CN112881054B (en
Inventor
梁敏健
戚政武
杨宁祥
林少杰
王璐
苏宇航
林晓明
刘德阳
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.)
Guangdong Inspection and Research Institute of Special Equipment Zhuhai Inspection Institute
Original Assignee
Guangdong Inspection and Research Institute of Special Equipment Zhuhai Inspection Institute
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 Guangdong Inspection and Research Institute of Special Equipment Zhuhai Inspection Institute filed Critical Guangdong Inspection and Research Institute of Special Equipment Zhuhai Inspection Institute
Priority to CN202110088959.5A priority Critical patent/CN112881054B/en
Publication of CN112881054A publication Critical patent/CN112881054A/en
Application granted granted Critical
Publication of CN112881054B publication Critical patent/CN112881054B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M99/00—Subject matter not provided for in other groups of this subclass
    • G01M99/004—Testing the effects of speed or acceleration
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66C—CRANES; LOAD-ENGAGING ELEMENTS OR DEVICES FOR CRANES, CAPSTANS, WINCHES, OR TACKLES
    • B66C15/00—Safety gear
    • B66C15/06—Arrangements or use of warning devices
    • 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
    • G06F17/15—Correlation function computation including computation of convolution operations
    • 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
    • G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Biomedical Technology (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Biophysics (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Neurology (AREA)
  • Molecular Biology (AREA)
  • Mechanical Engineering (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)

Abstract

The invention relates to a technical scheme of a hoisting machinery fault diagnosis method and a system, which comprises the following steps: initializing, namely associating a plurality of acceleration sensors with the positions to be detected respectively; acquiring data, namely acquiring a time domain signal from an acceleration sensor, and extracting time domain features from the time domain signal; data processing, namely performing LMD decomposition on the time domain signals to obtain a plurality of PF components with the highest kurtosis, and sequencing the PF components to obtain a data set; training a deep convolutional neural network, inputting a training set into a preset deep convolutional neural network for parameter adjusting training, and inputting a test set into the deep convolutional neural network for validity verification; and (4) fault diagnosis, namely, the trained deep convolutional neural network calculates the detection signal from each acceleration sensor to be detected and outputs the type of the mechanical fault. The invention has the beneficial effects that: the efficiency of hoisting machinery fault diagnosis is improved, and the accuracy of fault judgment is improved.

Description

Hoisting machinery fault diagnosis method and system
Technical Field
The invention relates to the technical field of computer and crane measurement, in particular to a method and a system for diagnosing faults of a hoisting machine.
Background
The crane is a common vehicle, and during the actual operation, accidents are often caused by faults, wherein mechanical faults of the crane are difficult to find, such as abrasion, gluing or pitting of gears in a reduction gearbox, abrasion of crane rollers and the like. The mechanical abrasion can be found in time only by checking the internal structure of the crane, the efficiency is low, the cost is high, the hidden danger identification depends on manual judgment, the subjective factor is large, and the hidden danger finding rate is low.
In order to improve the automation of fault diagnosis of hoisting machinery, the prior art combines a machine learning algorithm and a time-frequency analysis method. The original vibration signals generated by the parts to be measured (such as gears and rollers) in the crane are often non-stable and non-linear time-varying signals, and are decomposed into a plurality of signals with physical significance by a time-frequency analysis method for analysis. Common time-frequency analysis methods include Empirical Mode Decomposition (EMD) and Local-feature-scale Decomposition (LCD), but these two Decomposition algorithms have the problems of frequency aliasing and end-point effect, abrupt signal change caused by mediation, large calculation amount, and the like. The machine learning algorithm is based on shallow learning, the requirements on signal acquisition and processing are high, and the generalization capability of the model is poor.
Disclosure of Invention
The invention aims to solve at least one technical problem in the prior art, and provides a hoisting machine fault diagnosis method and system, which can improve the efficiency of hoisting machine fault diagnosis.
The technical scheme of the invention comprises a hoisting machinery fault diagnosis method, which comprises the following steps: initializing, namely associating a plurality of acceleration sensors with the positions to be detected respectively; acquiring data, namely acquiring a time domain signal from the acceleration sensor and extracting time domain features from the time domain signal; data processing, namely performing LMD decomposition on the time domain signal to obtain a plurality of PF components with the highest kurtosis, and sequencing the PF components to obtain a data set; training a deep convolutional neural network, dividing the data set into a training set and a test set according to a proportion, inputting the training set into a preset deep convolutional neural network for parameter adjustment training, and inputting the test set into the deep convolutional neural network for validity verification after parameter adjustment training; and (4) fault diagnosis, namely, the trained deep convolutional neural network calculates the detection signal from each acceleration sensor to be detected and outputs a corresponding mechanical fault type.
According to the hoisting machinery fault diagnosis method, the data processing comprises the following steps: equally dividing the time domain signal into a plurality of segments, wherein each segment comprises m data points; performing LMD decomposition on the time domain signal of each segment to obtain a plurality of PF components; and solving n PF components with the maximum kurtosis, and sequencing to form the data set, wherein the dimensionality of the data set is 1x m x n, m is greater than 0, and n is greater than or equal to 1.
According to the hoisting machinery fault diagnosis method, the LMD decomposition comprises the following steps: s310, assigning the time domain signal to an initial sequence y (t), and assigning the assigned sequence y (t) to a residual signal sequence r (t); s320, determining all local extreme points (t) of the sequence y (t)i,yi) 1, 2, 3, M, and averaging all adjacent local extrema points to obtain an average value
Figure BDA0002911725760000021
All mean value points miConnected by straight lines and then smoothed by a moving average method to obtain a local mean function m11(t); s330, calculating all the envelope estimation values,
Figure BDA0002911725760000022
all adjacent two envelope estimation values aiConnecting by straight lines, and smoothing by a moving average method to obtain an envelope estimation function a11(t); s340, calculating the local mean value m11(t) separating from the original signal y (t) to obtain a residual signal h11(t) wherein h11(t)＝y(t)-m11(t); s350, using an envelope estimation function a11(t) demodulating the residual signal to obtain a frequency modulated signal s11(t) in which
Figure BDA0002911725760000023
S360, for S11(t) repeating steps S320 to S350 j times as the original signal until a pure FM signal S is obtained1j(t), all envelope estimation functions generated in the iterative process are multiplied to obtain an envelope signal a of the PF component1(t), i.e. the instantaneous amplitude function of the PF component
Figure BDA0002911725760000031
S370 envelope signal a1(t) with a pure envelope signal s1j(t) to obtain a PF component PF of the original signal1＝a1(t)s1j(t); s380, calculating a signal difference sequence r1(t)＝y(t)-PF1R is to1(t) repeating the calculation k times as a new original signal according to S320 to 370, wherein the calculation mode is as follows:
Figure BDA0002911725760000032
up to rk(t) is a monotonic function and results in k PF components, the LMD decomposition ends, and the sequence
Figure BDA0002911725760000033
According to the hoisting machinery fault diagnosis method, the deep convolutional neural network comprises an input layer, a convolutional layer, a pooling layer, a full-link layer and an output layer, wherein the convolutional layer and the pooling layer respectively perform convolution operation and maximum pooling operation on the input layer for multiple times, and the output layer generates multiple crane mechanical fault types through a softmax function.
According to the hoisting machinery fault diagnosis method, the number of the convolution layers and the number of the pooling layers are 4, the convolution layers comprise a first convolution layer to a fourth convolution layer, the pooling layers comprise a first pooling layer to a fourth pooling layer, the first convolution layer selects 16 convolution kernels with the dimension of 1 × 63 × 6 to perform convolution operation with the input layer, the slip step length is set to be 2, the padding value is 31, a first convolution output with the dimension of 1 × 1024 × 16 is obtained, and the first convolution output is activated by using a ReLU function; a first pooling layer for pooling 1 × 2 maximum values of the first convolution output, outputting a first pooled output having dimensions of 1 × 512 × 16; a second convolution layer, selecting 32 convolution kernels with dimensions of 1 × 3 × 16 to perform convolution operation with the first pooled output, setting a slip step size to be 1, and obtaining a second convolution output with dimensions of 1 × 256 × 32 if a filling value is 1, and activating by using a ReLU function; a second pooling layer for pooling 1 × 2 maximum values of the second convolution output and outputting a second pooled output having a dimension of 1 × 128 × 32; a third convolution layer, selecting 32 convolution kernels with the dimension of 1 × 3 × 32 to perform convolution operation with the second pooled output, setting the slip step length to be 1, and obtaining a third convolution output with the dimension of 1 × 64 × 32 if the filling value is 1, and activating by using a ReLU function; a third pooling layer for pooling 1 × 2 maximum values of the third convolution output and outputting a third pooled output having a dimension of 1 × 32 × 32; a fourth convolution layer, selecting 64 convolution kernels with the dimension of 1 × 3 × 32 to perform convolution operation with the third pooled output, setting a slip step size to be 1, and obtaining a third convolution output with the dimension of 1 × 16 × 64 if a filling value is 1, and activating by using a ReLU function; and a fourth pooling layer for pooling the fourth convolution output by a maximum value of 1 × 2 and outputting a fourth pooled output having a dimension of 1 × 8 × 64.
According to the hoisting machinery fault diagnosis method, the type of the mechanical fault comprises at least one of bearing abrasion, bolt looseness, reduction gearbox gear abrasion, tooth breakage or pitting corrosion.
The technical scheme of the invention also comprises a crane mechanical fault diagnosis system which comprises acceleration sensors, a processor and a display, wherein one or more acceleration sensors are arranged at the position to be detected of the crane and used for acquiring detection information when the crane runs; the processor comprises a readable storage medium which is provided with an executable program, and the executable program realizes the execution of any hoisting machinery fault diagnosis method when executed, so as to obtain a diagnosis result; the display is used for outputting the diagnosis result of the processor.
The invention has the beneficial effects that: the detection signals from the acceleration sensor are subjected to LMD decomposition, so that the defects of multiple calculation times, large residual error, frequency mixing, end point effect and the like in the EMD decomposition and LCD decomposition processes are avoided, and the efficiency of fault diagnosis of the hoisting machinery is improved; the deep convolutional neural network is used, the structural limitation of the shallow neural network can be broken through, more complex data than the shallow neural network can be processed, effective features which cannot be extracted manually can be extracted in a self-adaptive mode, and the accuracy of fault judgment is improved.
Drawings
The invention is further described below with reference to the accompanying drawings and examples;
fig. 1 is a flowchart illustrating a fault diagnosis method of a hoisting machine according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a PF component decomposed from a bearing signal LMD according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a deep convolutional neural network according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of an input layer, convolutional layer, and pooling layer based on the deep convolutional neural network shown in FIG. 3;
FIG. 5 is a block diagram of a system according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to the present preferred embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to like elements throughout.
In the description of the present invention, the meaning of a plurality of means is one or more, the meaning of a plurality of means is two or more, and larger, smaller, larger, etc. are understood as excluding the number, and larger, smaller, inner, etc. are understood as including the number.
In the description of the present invention, the consecutive reference numbers of the method steps are for convenience of examination and understanding, and the implementation order between the steps is adjusted without affecting the technical effect achieved by the technical solution of the present invention by combining the whole technical solution of the present invention and the logical relationship between the steps.
In the description of the present invention, unless otherwise explicitly defined, terms such as set, etc. should be broadly construed, and those skilled in the art can reasonably determine the specific meanings of the above terms in the present invention in combination with the detailed contents of the technical solutions.
Referring to fig. 1, the present embodiment discloses a fault diagnosis method for a hoisting machine, which is applied to a fault diagnosis system for a hoisting machine having a main controller and a plurality of acceleration sensors, where the acceleration sensors are installed at positions to be detected of a crane, and the method includes the following steps:
s100, initializing, and associating a plurality of acceleration sensors with positions to be detected respectively;
for example, the acceleration sensor is mounted at a key portion such as a main machine and a speed reducer of each mechanism of the crane, and each acceleration sensor is associated with a mounting position.
S200, acquiring data, acquiring a time domain signal from an acceleration sensor, and extracting time domain features from the time domain signal, wherein the time domain features comprise an acceleration amplitude, an effective value and a standard deviation, and the extracted time domain features can be displayed as prompt information so as to assist a user in diagnosis and analysis;
s300, data processing, namely performing LMD (Local characteristic scale decomposition) on the time domain signals to obtain a plurality of PF (Intrinsic scale components) with the highest kurtosis, and sequencing the PF components to obtain a data set;
the LMD decomposition is that any one complex signal is assumed to be composed of different PF components, any two PF components are independent, and thus any one signal can be decomposed into the sum of a limited number of PF components, and the LMD decomposition is performed on time domain signals, so that the defects of multiple calculation times, large residual error, frequency mixing, end point effect and the like in the EMD decomposition and LCD decomposition processes are avoided, the calculation speed is improved, and the efficiency of fault diagnosis of the hoisting machinery is improved.
S400, training a deep convolutional neural network, dividing a data set into a training set and a test set according to a proportion, inputting the training set into a preset deep convolutional neural network for parameter adjustment training, and inputting the test set into the deep convolutional neural network for validity verification after parameter adjustment training;
the deep convolutional neural network is used, the structural limitation of the shallow neural network can be broken through, more complex data than the shallow neural network can be processed, effective features which cannot be extracted manually can be extracted in a self-adaptive mode, effectiveness verification is carried out through a test set after the deep convolutional neural network is trained, the deep convolutional neural network is guaranteed to be effective and feasible, and therefore accuracy of fault judgment is improved.
And S500, fault diagnosis, wherein the trained deep convolutional neural network is used for calculating the detection signal from each acceleration sensor to be detected and outputting the corresponding mechanical fault type.
For example, the acceleration sensor is installed on a speed reducer of a crane, the deep convolutional neural network calculates a detection signal from the acceleration sensor, and the deep convolutional neural network outputs a corresponding mechanical fault type according to a calculation result, such as bearing abrasion, bolt looseness, reduction gearbox gear abrasion, tooth breakage or pitting corrosion.
In the above step, the data processing specifically includes the following steps:
equally dividing the time domain signal into a plurality of segments, wherein each segment comprises m data points;
performing LMD decomposition on the time domain signal of each segment to obtain a plurality of PF components;
and (3) solving n PF components with the maximum kurtosis, and sequencing to form a data set, wherein the dimensionality of the data set is 1x m x n, m is greater than 0, and n is greater than or equal to 1.
In the above step, the LMD decomposition specifically includes the following steps:
s310, assigning the time domain signal to an initial sequence y (t), and assigning the assigned sequence y (t) to a residual signal sequence r (t);
s320, determining all local extreme points (including a maximum value and a minimum value) of the sequence y (t)i,yi) 1, 2, 3.. times.m, the average of all adjacent local extrema points is found:
Figure BDA0002911725760000071
all mean value points miConnected by straight lines and then smoothed by a moving average method to obtain a local mean function m11(t)
S330, all envelope estimation values are obtained
Figure BDA0002911725760000072
All adjacent two envelope estimation values aiConnecting by straight lines, and smoothing by a moving average method to obtain an envelope estimation function a11(t)。
S340 local mean m11(t) separating from the original signal y (t) to obtain a residual signal h11(t)
h11(t)＝y(t)-m11(t)
S350 envelope estimation function a11(t) demodulating the residual signal to obtain a frequency modulated signal s11(t)
Figure BDA0002911725760000073
S360 to S11(t) repeating the steps S320 to S350 j times as the original signal until a pure FM signal S is obtained1j(t), all envelope estimation functions generated in the iterative process are multiplied to obtain an envelope signal a of the PF component1(t), i.e. the instantaneous amplitude function of the PF component
Figure BDA0002911725760000074
S370 envelope signal a1(t) with a pure envelope signal s1j(t) to obtain a PF component of the original signal
PF1＝a1(t)s1j(t)
S380 computing a signal difference sequence r1(t)＝y(t)-PF1(ii) a Will r is1(t) repeating the calculation k times as a new original signal according to S320 to 370,
r1(t)＝y(t)-PF1(t)
r2(t)＝r1(t)-PF2(t)
...
rk(t)＝rk-1(t)-PFk(t)
up to rk(t) is a monotonic function, where k PF components are obtained and LMD decomposition ends
Figure BDA0002911725760000075
Please refer to fig. 2, which shows PF components of a bearing signal.
Referring to fig. 3, the deep convolutional neural network includes an input layer, a convolutional layer, a pooling layer, a fully-connected layer, and an output layer, wherein the convolutional layer and the pooling layer perform a plurality of convolutional operations and a maximum pooling operation on the input layer, respectively, the output layer generates a plurality of mechanical failure types of the crane through a softmax function, the softmax function is a normalized exponential function capable of "compressing" a K-dimensional vector z containing any real number into another K-dimensional real vector σ (z) such that the range of each element is between (0, 1), and the sum of all elements is 1.
Referring to fig. 4, the data set is as per 8: the scale of 2 is divided into a training set having 1 × 2048 × 6 PF components and a test set, wherein the training set serves as input parameters of the input layer 410 of the deep convolutional neural network. I.e., a six-channel data set, each channel is a 1X2048 point PF component map.
The number of the convolutional layers and the pooling layers is 4, the convolutional layers comprise a first convolutional layer to a fourth convolutional layer, the pooling layers comprise a first pooling layer to a fourth pooling layer, wherein,
the first convolution layer 421 selects 16 convolution kernels with dimensions of 1 × 63 × 6 to perform convolution operation with the input layer, sets a slip step size to be 2, and a padding value to be 31, then obtains a first convolution output with dimensions of 1 × 1024 × 16, and uses a ReLU function to activate, the ReLU (Rectified Linear Unit, Linear rectification function) function, also called a modified Linear Unit, is an activation function (activation function) commonly used in an artificial neural network,
the expression of the convolution operation is:
Figure BDA0002911725760000081
wherein k represents the kth convolutional layer, operator
Figure BDA0002911725760000082
Representing input features for layers
Figure BDA0002911725760000083
And weight matrix
Figure BDA0002911725760000084
Performing convolution operation to correspondingly obtain the mth output characteristic diagram
Figure BDA0002911725760000085
Wherein
Figure BDA0002911725760000086
Is an offset term, C is the number of layers, C is 1, 2, …, C. After operation
Figure BDA0002911725760000087
And (f) activating nonlinearly by the formula (x) max {0, x }, namely taking the maximum value of 0 or x, wherein x is an input parameter.
A first pooling layer 422 for pooling the first convolution output by a maximum value of 1 × 2, outputting a first pooled output having a dimension of 1 × 512 × 16;
the second convolution layer 423 selects 32 convolution kernels with dimensions of 1 × 3 × 16 to perform convolution operation with the first pooled output, sets the slip step size to be 1, and obtains a second convolution output with dimensions of 1 × 256 × 32 if the padding value is 1, and uses a ReLU function for activation;
a second pooling layer 424 for pooling the second convolution output by a maximum of 1 × 2, and outputting a second pooled output having a dimension of 1 × 128 × 32;
a third convolution layer 425, which selects 32 convolution kernels with dimensions of 1 × 3 × 32 to perform convolution operation with the second pooled output, sets a slip step size of 1 and a padding value of 1, obtains a third convolution output with dimensions of 1 × 64 × 32, and uses a ReLU function for activation;
a third pooling layer 426 for pooling the third convolution output by the maximum value of 1 × 2 and outputting a third pooled output having a dimension of 1 × 32 × 32;
a fourth convolution layer 427, which selects 64 convolution kernels with dimensions of 1 × 3 × 32 to perform convolution operation with the third pooled output, sets a slip step size of 1 and a padding value of 1, obtains a third convolution output with dimensions of 1 × 16 × 64, and activates the convolution output by using a ReLU function;
a fourth pooling layer 428 for pooling the fourth convolution output by a maximum of 1 × 2, outputting a fourth pooled output having dimensions of 1 × 8 × 64;
referring to fig. 3, a fully-connected layer, receiving a fourth pooled output with dimension 1 × 8 × 64 ═ 512, outputting a vector with dimension 300, and activating using a ReLU function; dropout can be added, 300-dimensional input vectors are received, 150-dimensional vectors are output, and ReLU function activation is adopted; DropOut produces only 80% of the output.
The output layer receives the vectors from the full connection layer and generates 5 mechanical fault types of the crane through a softmax function, wherein the mechanical fault types comprise at least one of bearing wear, bolt looseness, reduction gearbox gear wear, tooth breakage or pitting corrosion, and the activation function of the output layer is
Figure BDA0002911725760000091
xiIs the output value of the i-th node, x1,x2,...xCThe output values of the 1 st, 2 nd and … th nodes are C, and C is the number of classifications. .
Fig. 5 is a block diagram of a system according to an embodiment of the present invention, which includes acceleration sensors 100, a processor 200 and a display 300, where one or more acceleration sensors 100 are installed at a position to be detected of a crane, and are used for collecting detection information when the crane is in operation; the processor includes a readable storage medium 210, the readable storage medium 210 being provided with an executable program that when executed implements: initializing, namely associating a plurality of acceleration sensors with the positions to be detected respectively; acquiring data, namely acquiring a time domain signal from an acceleration sensor, and extracting time domain features from the time domain signal; data processing, namely performing LMD decomposition on the time domain signals to obtain a plurality of PF components with the highest kurtosis, and sequencing the PF components to obtain a data set; training a deep convolutional neural network, inputting a training set into a preset deep convolutional neural network for parameter adjusting training, and inputting a test set into the deep convolutional neural network for validity verification; fault diagnosis, namely calculating a detection signal from each acceleration sensor to be detected by the trained deep convolutional neural network, and outputting a mechanical fault type to obtain a diagnosis result; the display 300 is used to output the diagnostic results of the processor.
The embodiments of the present invention have been described in detail with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the gist of the present invention.

Claims (7)

1. A hoisting machinery fault diagnosis method is characterized by comprising the following steps:
initializing, namely associating a plurality of acceleration sensors with the positions to be detected respectively;
acquiring data, namely acquiring a time domain signal from the acceleration sensor and extracting time domain features from the time domain signal;
data processing, namely performing LMD decomposition on the time domain signal to obtain a plurality of PF components with the highest kurtosis, and sequencing the PF components to obtain a data set;
training a deep convolutional neural network, dividing the data set into a training set and a test set according to a proportion, inputting the training set into a preset deep convolutional neural network for parameter adjustment training, and inputting the test set into the deep convolutional neural network for validity verification after parameter adjustment training;
and (4) fault diagnosis, namely, the trained deep convolutional neural network calculates the detection signal from each acceleration sensor to be detected and outputs a corresponding mechanical fault type.
2. The hoisting machine fault diagnosis method according to claim 1, characterized in that the data processing comprises:
equally dividing the time domain signal into a plurality of segments, wherein each segment comprises m data points;
performing LMD decomposition on the time domain signal of each segment to obtain a plurality of PF components;
and solving n PF components with the maximum kurtosis, and sequencing to form the data set, wherein the dimensionality of the data set is 1x m x n, m is greater than 0, and n is greater than or equal to 1.
3. Hoisting machine fault diagnosis method according to claim 1 or 2, characterized in that the LMD decomposition comprises:
s310, assigning the time domain signal to an initial sequence y (t), and assigning the assigned sequence y (t) to a residual signal sequence r (t);
s320, determining all local extreme points (t) of the sequence y (t)i,yi) 1, 2, 3, M, and averaging all adjacent local extrema points to obtain an average value
Figure FDA0002911725750000011
All mean value points miConnected by straight lines and then smoothed by a moving average method to obtain a local mean function m11(t)；
S330, calculating all the envelope estimation values,
Figure FDA0002911725750000021
all adjacent two envelope estimation values aiConnecting by straight lines, and smoothing by a moving average method to obtain an envelope estimation function a11(t)；
S340, calculating the local mean value m11(t) separating from the original signal y (t) to obtain a residual signal h11(t) wherein h11(t)＝y(t)-m11(t)；
S350, using an envelope estimation function a11(t) demodulating the residual signal to obtain a frequency modulated signal s11(t) in which
Figure FDA0002911725750000022
S360, for S11(t) repeating steps S320 to S350 j times as the original signal until a pure FM signal S is obtained1j(t), all envelope estimation functions generated in the iterative process are multiplied to obtain an envelope signal a of the PF component1(t), i.e. the instantaneous amplitude function of the PF component
Figure FDA0002911725750000023
S370 envelope signal a1(t) with a pure envelope signal s1j(t) to obtain a PF component PF of the original signal1＝a1(t)s1j(t)；
S380, calculating a signal difference sequence r1(t)＝y(t)-PF1R is to1(t) repeating the calculation k times as a new original signal according to S320 to 370, wherein the calculation mode is as follows:
Figure FDA0002911725750000024
up to rk(t) is a monotonic function and results in k PF components, the LMD decomposition ends, and the sequence
Figure FDA0002911725750000025
4. The hoisting machine fault diagnosis method according to claim 1, wherein the deep convolutional neural network comprises an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the convolutional layer and the pooling layer perform a plurality of convolution operations and a maximum pooling operation on the input layer respectively, and the output layer generates a plurality of crane mechanical fault types through a softmax function.
5. The hoisting mechanical fault diagnosis method according to claim 4, wherein the number of said convolutional layers and said pooling layers is 4, each of said convolutional layers including a first convolutional layer to a fourth convolutional layer, and each of said pooling layers including a first pooling layer to a fourth pooling layer, wherein,
a first convolution layer, selecting 16 convolution kernels with the dimension of 1 × 63 × 6 to perform convolution operation with the input layer, setting a slip step length to be 2, and setting a filling value to be 31, then obtaining a first convolution output with the dimension of 1 × 1024 × 16, and activating by using a ReLU function;
a first pooling layer for pooling 1 × 2 maximum values of the first convolution output, outputting a first pooled output having dimensions of 1 × 512 × 16;
a second convolution layer, selecting 32 convolution kernels with dimensions of 1 × 3 × 16 to perform convolution operation with the first pooled output, setting a slip step size to be 1, and obtaining a second convolution output with dimensions of 1 × 256 × 32 if a filling value is 1, and activating by using a ReLU function;
a second pooling layer for pooling 1 × 2 maximum values of the second convolution output and outputting a second pooled output having a dimension of 1 × 128 × 32;
a third convolution layer, selecting 32 convolution kernels with the dimension of 1 × 3 × 32 to perform convolution operation with the second pooled output, setting the slip step length to be 1, and obtaining a third convolution output with the dimension of 1 × 64 × 32 if the filling value is 1, and activating by using a ReLU function;
a third pooling layer for pooling 1 × 2 maximum values of the third convolution output and outputting a third pooled output having a dimension of 1 × 32 × 32;
a fourth convolution layer, selecting 64 convolution kernels with the dimension of 1 × 3 × 32 to perform convolution operation with the third pooled output, setting a slip step size to be 1, and obtaining a third convolution output with the dimension of 1 × 16 × 64 if a filling value is 1, and activating by using a ReLU function;
and a fourth pooling layer for pooling the fourth convolution output by a maximum value of 1 × 2 and outputting a fourth pooled output having a dimension of 1 × 8 × 64.
6. The hoisting machine fault diagnosis method according to claim 1, characterized in that: the mechanical fault type comprises at least one of bearing abrasion, bolt looseness, reduction gearbox gear abrasion, tooth breakage or pitting corrosion.
7. The fault diagnosis system for the hoisting machinery is characterized by comprising acceleration sensors, a processor and a display, wherein one or more acceleration sensors are arranged at positions to be detected of a crane and used for acquiring detection information when the crane runs; the processor comprises a readable storage medium provided with an executable program, and the executable program realizes the hoisting machinery fault diagnosis method of any one of claims 1 to 6 when executed, and obtains a diagnosis result; the display is used for outputting the diagnosis result of the processor.
CN202110088959.5A 2021-01-22 2021-01-22 Hoisting machinery fault diagnosis method and system Active CN112881054B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110088959.5A CN112881054B (en) 2021-01-22 2021-01-22 Hoisting machinery fault diagnosis method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110088959.5A CN112881054B (en) 2021-01-22 2021-01-22 Hoisting machinery fault diagnosis method and system

Publications (2)

Publication Number Publication Date
CN112881054A true CN112881054A (en) 2021-06-01
CN112881054B CN112881054B (en) 2023-04-14

Family

ID=76050394

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110088959.5A Active CN112881054B (en) 2021-01-22 2021-01-22 Hoisting machinery fault diagnosis method and system

Country Status (1)

Country Link
CN (1) CN112881054B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114212688A (en) * 2022-02-22 2022-03-22 杭州未名信科科技有限公司 Motion control method and device of intelligent tower crane
CN114526898A (en) * 2022-01-25 2022-05-24 广东省特种设备检测研究院珠海检测院 Method and system for detecting looseness of bolt of mechanical part

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104390781A (en) * 2014-11-26 2015-03-04 中国矿业大学 Gear fault diagnosis method based on LMD and BP neural network
CN104832418A (en) * 2015-05-07 2015-08-12 北京航空航天大学 Hydraulic pump fault diagnosis method based on local mean conversion and Softmax
CN106875041A (en) * 2017-01-16 2017-06-20 广东电网有限责任公司揭阳供电局 A kind of short-term wind speed forecasting method
CN106908241A (en) * 2017-02-23 2017-06-30 北京工业大学 A kind of bearing fault method of discrimination being combined with Wavelet Denoising Method based on LMD
CN108999902A (en) * 2018-08-13 2018-12-14 广东省特种设备检测研究院珠海检测院 A kind of measurement method and device improving crane brake downslide accuracy of measurement
CN111751133A (en) * 2020-06-08 2020-10-09 南京航空航天大学 Intelligent fault diagnosis method of deep convolutional neural network model based on non-local mean embedding
CN111860184A (en) * 2020-06-23 2020-10-30 广东省特种设备检测研究院珠海检测院 Escalator mechanical fault diagnosis method and system
CN112014108A (en) * 2020-08-08 2020-12-01 中车长春轨道客车股份有限公司 Bearing fault diagnosis method based on LMD and improved PSO (particle swarm optimization) optimized BP (Back propagation) neural network

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104390781A (en) * 2014-11-26 2015-03-04 中国矿业大学 Gear fault diagnosis method based on LMD and BP neural network
CN104832418A (en) * 2015-05-07 2015-08-12 北京航空航天大学 Hydraulic pump fault diagnosis method based on local mean conversion and Softmax
CN106875041A (en) * 2017-01-16 2017-06-20 广东电网有限责任公司揭阳供电局 A kind of short-term wind speed forecasting method
CN106908241A (en) * 2017-02-23 2017-06-30 北京工业大学 A kind of bearing fault method of discrimination being combined with Wavelet Denoising Method based on LMD
CN108999902A (en) * 2018-08-13 2018-12-14 广东省特种设备检测研究院珠海检测院 A kind of measurement method and device improving crane brake downslide accuracy of measurement
CN111751133A (en) * 2020-06-08 2020-10-09 南京航空航天大学 Intelligent fault diagnosis method of deep convolutional neural network model based on non-local mean embedding
CN111860184A (en) * 2020-06-23 2020-10-30 广东省特种设备检测研究院珠海检测院 Escalator mechanical fault diagnosis method and system
CN112014108A (en) * 2020-08-08 2020-12-01 中车长春轨道客车股份有限公司 Bearing fault diagnosis method based on LMD and improved PSO (particle swarm optimization) optimized BP (Back propagation) neural network

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
宋平岗等: "形态滤波优化算法用于滚动轴承故障诊断", 《振动、测试与诊断》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114526898A (en) * 2022-01-25 2022-05-24 广东省特种设备检测研究院珠海检测院 Method and system for detecting looseness of bolt of mechanical part
CN114526898B (en) * 2022-01-25 2023-11-07 广东省特种设备检测研究院珠海检测院 Method and system for detecting looseness of mechanical part bolt
CN114212688A (en) * 2022-02-22 2022-03-22 杭州未名信科科技有限公司 Motion control method and device of intelligent tower crane

Also Published As

Publication number Publication date
CN112881054B (en) 2023-04-14

Similar Documents

Publication Publication Date Title
CN112881054B (en) Hoisting machinery fault diagnosis method and system
CN110428416B (en) Liquid level visual detection method and device
CN112013285B (en) Method and device for detecting pipeline leakage point, storage medium and terminal
US11544554B2 (en) Additional learning method for deterioration diagnosis system
WO2022222026A1 (en) Medical diagnosis missing data completion method and completion apparatus, and electronic device and medium
McHugh et al. The effect of uncertainty in patient classification on diagnostic performance estimations
CN111179235A (en) Image detection model generation method and device, and application method and device
DE102020214234A1 (en) Device and method for determining hotspot temperatures of a rotor and a stator
CN106939840A (en) Method and apparatus for determining the gas mass flow in internal combustion engine
CN109669849B (en) Complex system health state assessment method based on uncertain depth theory
CN115937126A (en) Fatigue testing method and device for automobile chassis, electronic equipment and storage medium
CN113569432B (en) Simulation detection method and system for liquid-air-tight element
CN114738205A (en) Method, device, equipment and medium for monitoring state of floating fan foundation
CN114519370A (en) Left ventricular hypertrophy detection method and system based on deep learning early-arrest mechanism
CN111860184A (en) Escalator mechanical fault diagnosis method and system
CN108833024B (en) Multi-channel wireless distributed field vehicle brake data transmission method
CN113063589B (en) Gear microscopic error vibration prediction method based on neural network
CN109946096B (en) High-speed train air pipe fault diagnosis method based on model space
CN111506045A (en) Fault diagnosis method based on single-value intelligent set correlation coefficient
JP7272143B2 (en) Inspection machine management system and inspection machine management method
CN114036648B (en) Ship body structure overall stress state acquisition method based on local stress correlation method
CN115859201B (en) Chemical process fault diagnosis method and system
US11669082B2 (en) Online fault localization in industrial processes without utilizing a dynamic system model
CN107622516A (en) The after-treatment system and post-processing approach of electrical impedance tomography art image
CN117077537A (en) Energy consumption determination and model training method and device for engineering mechanical equipment

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant