CN109359524A - A kind of loading machine operating mode's switch model construction and recognition methods - Google Patents
A kind of loading machine operating mode's switch model construction and recognition methods Download PDFInfo
- Publication number
- CN109359524A CN109359524A CN201811043486.1A CN201811043486A CN109359524A CN 109359524 A CN109359524 A CN 109359524A CN 201811043486 A CN201811043486 A CN 201811043486A CN 109359524 A CN109359524 A CN 109359524A
- Authority
- CN
- China
- Prior art keywords
- identification
- signal data
- feature
- loading machine
- identification signal
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
-
- E—FIXED CONSTRUCTIONS
- E02—HYDRAULIC ENGINEERING; FOUNDATIONS; SOIL SHIFTING
- E02F—DREDGING; SOIL-SHIFTING
- E02F9/00—Component parts of dredgers or soil-shifting machines, not restricted to one of the kinds covered by groups E02F3/00 - E02F7/00
- E02F9/26—Indicating devices
-
- 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
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Signal Processing (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Mining & Mineral Resources (AREA)
- Civil Engineering (AREA)
- Structural Engineering (AREA)
- Operation Control Of Excavators (AREA)
Abstract
The invention discloses a kind of loading machine operating mode's switch model construction and recognition methods, arrange corresponding sensor to acquire the source signals such as torque, pressure, gear, braking on loading machine first, and standardize to data, zero signal that floats is removed, interpolation polishing is given to missing values, noise reduction filtering processing is carried out to collected signal;Secondly, selecting the higher characteristic attribute of contribution degree from the multiattribute data of loading machine using Principal Component Analysis, and extracted using feature of the statistical analysis method to principal component;Then, loading machine operating condition sample is established, the relationship maps for establishing load signal between regime mode of presorting using the data mining algorithm of supervised learning form operating mode's switch model by mass data sample training;The feature extracting method of principal component analysis and KNN algorithm are combined, the range formula in KNN algorithm is improved, is allowed to more meet operating mode's switch, improves the accuracy and efficiency of operating mode's switch algorithm.
Description
Technical field
The present invention relates to operating mode's switch methods, and in particular to a kind of loading machine operating mode's switch model construction and recognition methods.
Background technique
With China's economic development, production, sales volume and the ownership of engineering truck are quickly improved, and engineering machinery undergoes an unusual development fast
Speed.The product of 95% or more engineering machinery uses hydraulic power, to obtain large torque, to meet large inertia load requirements, due to
Operating environment is severe, and working condition is complicated and changeable and equipment automatization, the level of informatization constantly improve, and how to ensure engineering
Mechanical reliable, efficient operation, is technical problem urgently to be resolved at present, in order to solve these problems, it is necessary to loading machine
The loading spectrum of operation is analyzed, the extraction including signal characteristic, the division of sessions and the identification of working condition, wherein filling
The operating condition of carrier aircraft includes: unloaded advance, spading, fully loaded retrogressing, fully loaded advance, unloading and unloaded retrogressing.
Existing operating mode's switch method is mainly that the big cavity pressure signal of rotating bucket, the swing arm when passing through detection loading machine operation are big
The variable quantity of cavity pressure signal is determined the behavior more than variable quantity threshold value by way of setting variable quantity threshold value, is somebody's turn to do
Kind of method depends on the setting of threshold value, and the value of threshold value is frequently depend upon experience, thus this kind of method there are recognition correct rates not
Height is unable to satisfy work requirements.
Summary of the invention
It is existing to solve the purpose of the present invention is to provide a kind of loading machine operating mode's switch model construction and recognition methods
The problems such as loading machine operating mode's switch method recognition correct rate in technology is not high.
In order to realize above-mentioned task, the invention adopts the following technical scheme:
A kind of loading machine operating mode's switch model building method, method the following steps are included:
The multiple groups identification signal data of step 1, acquisition loading machine under different operating conditions, obtain identification signal data set;Institute
The corresponding operating condition label of each group of identification signal data in the identification signal data set stated, obtains identification operating condition tally set;
The operating condition label includes: unloaded advance, spading, fully loaded retrogressing, fully loaded advance, unloading and unloaded retrogressing;
Step 2 pre-processes every group of identification signal data in the identification signal data set, is pre-processed
Identification signal data set afterwards;
The pretreatment specifically includes following procedure:
The identification signal data set is normalized between 0 to 1 by step 21, obtains the second identification signal data set;
Step 22 after carrying out removing trend term to the second identification signal data set, utilizes 3 σ method rejecting abnormalities
Newton interpolating method interpolation vacancy item is recycled after data, obtains third identification signal data set;
Step 23 is filtered the third identification signal data set using Wavelet Package Denoising Method, obtains
Pretreated identification signal data set;
In the Wavelet Package Denoising Method, Selection of Wavelet Basis db9-6 wavelet basis;
Step 3 is handled the pretreated identification signal data set using the feature extracting method of dimensionality reduction,
Obtain identification feature collection;
The identification feature collection includes multiple feature samples, is acquired in the quantity and step 1 of the feature samples
The group number of identification signal data is identical, and each feature samples include I identification feature amount, and I is positive integer;
Obtain the contribution rate of feature identified amount: i-th of identification of p-th of feature samples is special in multiple feature samples
Sign amountWith i-th of identification feature amount of q-th of feature samplesContribution rate it is identical, that is, obtain I identification feature amount
Contribution rate, i ∈ [1, I], p and q are positive integer, p ≠ q;
Step 4 is trained using the identification feature collection as input using the identification operating condition tally set as output
KNN model obtains loading machine operating mode's switch model, in the KNN model, p-th of feature samples and q-th feature samples it
Between distance Dis:.
Wherein, CiFor the contribution rate of i-th of identification feature amount.
Further, the step 4, the value of K value is any positive integer within 5 in KNN model.
Further, the step 3, using Principal Component Analysis to the pretreated identification signal data set
It is handled, obtains identification signal feature set.
A kind of loading machine operating mode's switch method, the method include: using in any one of claim 1-3 claim
The operating mode's switch model is to the loading by step 1- step 3 processing described in any one of claim 1-3 claim
The signal data to be identified of machine is identified.
The present invention has following technical characterstic compared with prior art:
1, the preprocess method of signal data provided by the invention is more accurate, so that the accuracy of identification also mentions therewith
Height is filtered comprising standardization, removing trend term, interpolation missing values and filtering processing etc. by using wavelet packet optimal base decomposition tree
Wave method remains high frequency detail while removing a large amount of interference signals, can retain a large amount of signal characteristics and be used for operating mode's switch,
Relative to prior art Butterworth filter method, Fourier Transform Filtering, wavelet transform filtering method etc.;
If 2, the existing judgment variable only with single pressure or torque as identification, accuracy rate are lower;
If all data collected are identified that operand is excessive, overlong time is calculated, delay is serious, is unfavorable for operating condition
In identification, condition model method for building up provided by the invention and identification, using the feature extracting method of dimensionality reduction, choose contribution rate compared with
Input of the high identification feature as KNN identification model;
3, the dimensionality reduction feature extracting method of Principal Component Analysis and KNN algorithm are combined, to the distance in KNN algorithm
Formula is improved, and is allowed to more meet operating mode's switch, improves the accuracy and efficiency of operating mode's switch algorithm.
Detailed description of the invention
Fig. 1 is loading machine operating mode's switch model building method flow chart provided by the invention.
Specific embodiment
It is the specific embodiment that inventor provides below, to be further explained explanation to technical solution of the present invention.
Embodiment one
Present embodiment discloses a kind of loading machine operating mode's switch model building method, method the following steps are included:
The multiple groups identification signal data of step 1, acquisition loading machine under different operating conditions, as identification signal data set;Institute
The corresponding operating condition label of each group of identification signal data in the identification signal data set stated, obtains identification operating condition tally set;
The operating condition of loading machine refers to working condition of loading machine under conditions of having direct relation with its movement, generally,
The operating condition of loading machine includes spading, fully loaded transport and unloading.
In the present embodiment, the operating condition of loading machine has been carried out to careful division, to guarantee the accuracy rate of judgement, loading machine
Operating condition includes unloaded advance, spading, fully loaded retrogressing, fully loaded advance, unloading and unloaded retrogressing;
Identification signal data of the loading machine under different operating conditions include: loading machine front axle torque, loading machine front axle revolving speed, dress
Carrier aircraft reared torque, the pressure of working barrel, steering pump pressure, the flow of working barrel, the flow of steering pump, engine speed, system
Dynamic signal, throttle signal and gearbox signal etc..
In this multi-signal, due to brake signal, throttle signal, turns to pump discharge and gearbox letter at work pump discharge
Number subjectivity it is too strong, theoretically analyze, accelerator open degree, brake signal and gearbox signal belong to artificial manipulation factor, main
The property seen is too strong;Working barrel is only capable of reacting the speed of hydraulic cylinder work with the flow of steering pump, is similarly subjected to the shadow of human factor
Sound is larger.In loading machine operating condition intelligent recognition, once it joined the behavior signal of driver, although can significantly judge work
Condition, but operation of the driver without fault is needed, it is otherwise easy misrecognition, therefore be not included in the range of identification signal.
The operating condition label includes: unloaded advance, spading, fully loaded retrogressing, fully loaded advance, unloading and unloaded retrogressing;
According to collected driving behavior data, after the pre-treatment, the division of operating condition segment is carried out to it, is shown in Table 1.
The judgment basis of 1 loading machine operating condition of table
Judgment basis in table 1 has universality.Judge that S1 zero load is advanced and the fully loaded advance of S4 according to gear;The S2 stage is
There is braking in forward, followed by along with significantly accelerator open degree, it is therefore an objective to when closing on material heap to spading
Preparation, the impact of vehicle is excessive if not braking, and significantly accelerator open degree is the work in spading;In S2 rank
Excessive due to loading at the end of section, speed is almost 0, sometimes with a slight brake signal, therefore retreats gear letter
Number judgment basis started for the S3 stage;The S5 stage is unloading phase, and accelerator releasing is usually understood when close to dumper, touches on the brake and keeps away
Exempt to bump against, and braking time at this time is slightly longer, since it is desired that carrying out discharging and shift;After the S6 stage terminates for brake signal,
Before throttle signal, rather than after gear shifting signal, because of the case where not yet unloading end occasionally there are shift end.
Therefore in the present embodiment, in identification signal data set include loading machine front axle torque, loading machine front axle revolving speed,
Loading machine reared torque, the pressure of working barrel, steering pump pressure, engine speed }, loading machine is acquired in 6 kinds of different operating conditions
Under, the value of above-mentioned 6 identification signal data, that is to say, that the corresponding one group of identification signal data of each operating condition, one group of identification letter
Number includes 6 numerical value.Operating condition label is operating condition title, and in the present embodiment, 1- zero load advance, 2- spading, 3- are fully loaded
It retreats, the fully loaded advance of 4-, 5- is unloaded and 6- zero load retreats.
Step 2 pre-processes each group of identification signal data in the identification signal data set, obtains pre- place
Identification signal data set after reason;All identification signal data in the pretreated identification data set 0 to 1 it
Between;
For improve algorithm operating rate, guarantee identification as a result, being pre-processed to identification signal data, comprising:
The identification signal data set is normalized between 0 to 1 by step 21, obtains the second identification signal data set;
In the present embodiment, in order to improve the efficiency of algorithm, by the unit conversion of identification signal data at corresponding immeasurable
Guiding principle.By taking engine speed as an example, maximum-is carried out to it and is most standardized, its numerical value is mapped in [0,1] section.
Step 22, after carrying out removing trend term to the second identification signal data set, to abnormal data therein into
Interpolation vacancy item again after row is rejected obtains third identification signal data set;
In this step, since identification signal is acquired by sensor, the interference of meeting amplifying ambient.Collected vibration
Dynamic signal often deviates baseline in time series, generates a linear general trend (null offset), the trend with
The process of time change is known as trend term.The period of trend term is far longer than the frequency of sample, can make correlation analysis, power spectrum
Very big distortion is caused when analysis, or even will cause the distortion of signal.When therefore being analyzed after long-time measuring signal, need
Trend term is separated from data.
As a preferred embodiment, it is simple and precision is high using the algorithm of least square method removing trend term, both
The growth trend of approximately linear can be eliminated, and the trend of higher order polynomial can be eliminated.
In this step, identification signal is after removing trend term, however it remains abnormal signal, its main feature is that randomness
By force, amplitude is big, the period is indefinite etc..If the data comprising exceptional value are calculated with not rejected, calculating will affect
The result of analysis.
Common interpolation method includes Lagrange's interpolation and Newton interpolating method etc..
As a preferred embodiment, being rejected using 3 σ principles in the identification signal data set by removing trend term
Abnormal data after recycle Newton interpolating method interpolation vacancy item, obtain third identification signal data set.
Due under great amount of samples data, usual Normal Distribution, and 3 σ principles be defined in measured data with it is flat
Mean bias is more than that the value of three times standard deviation is exceptional value.Under 3 σ principles, the probability of data exception be P (| x- μ | > 3 σ)≤
0.003, belong to minimum probability event, and the overall section being mainly distributed is (+3 σ of μ -3 σ, μ).
In the pretreatment stage of data, rejectings should give for abnormal data, but ignorance missing values and rejecting is different
Constant value can abandon a large amount of wastes hidden information in record, cause to acquire information, it is therefore desirable to by missing values and after rejecting
Abnormal point carry out interpolation, when needing to increase interpolation knot, the basic function of Lagrange's interpolation will change therewith, calculate
With it is more inconvenient in practice, therefore select Newton interpolating method.
Step 23 is filtered the third identification signal data set, obtains pretreated identification signal
Data set.
Since the interference of the signals such as noise is so that the collected data of institute are there are random error, the fluctuation of error is influenced whether
True value is filtered place to third identification signal data set so needing to be filtered noise signal, therefore in this step
Reason, the method being commonly filtered include that Fast Fourier Transform (FFT), Butterworth filtering processing, wavelet analysis, wavelet packet are gone
It makes an uproar.
In the present embodiment, by many experiments compare more than filter processing method superiority-inferiority, by signal-to-noise ratio,
Square error, peak error judge filter effect, the results are shown in Table 2.
Table 2 denoises effect assessment
As can be known from Table 2, wavelet analysis is similar to the signal-to-noise ratio of wavelet packet analyzing denoising, root-mean-square error, peak error,
It is superior to Butterworth denoising.Although the denoising method evaluation result of wavelet transformation is slightly better than wavelet packet denoising, wavelet packet
Decomposition it is finer, the high frequency detail of part can be retained, the operation is more convenient, therefore embodiment party as one preferred
Formula is filtered the third identification signal data set using Wavelet Package Denoising Method, obtains pretreated knowledge
Level signal data set;
In the Wavelet Package Denoising Method, Selection of Wavelet Basis db9-6 wavelet basis.
It, will { loading machine front axle torque, loading machine front axle revolving speed, the torsion of loading machine rear axle by the processing of step 21- step 23
Square, working barrel pressure, turn to pump pressure, engine speed this six kinds of signal identification datas pre-process to 0-1,
And it ensure that the stationarity of identification signal data.
But in the present embodiment, for loading machine front axle torque, loading machine front axle revolving speed, loading machine reared torque, work
Make the pressure pumped, turn to the engine power obtained after this 6 kinds of signal identification datas calculating of pump pressure, engine speed, transmission
Although the parameters such as system output power, steering pump and work pump power have certain correlation with original variable, more can
Working condition is intuitively reacted, derivative variable is belonged to, also as the identification signal in the present embodiment.
By above step, each group of pretreated identification data include loading in pretreated identification data set
Machine front axle torque, loading machine front axle revolving speed, loading machine reared torque, working barrel pressure, turn to pump pressure, working barrel function
Rate, the power of steering pump, engine speed, engine torque, output shaft power, engine power this 11 identification signals.
Step 3 is handled the pretreated identification signal data set using the feature extracting method of dimensionality reduction,
Obtain identification feature collection;
The identification feature collection includes multiple feature samples, is acquired in the quantity and step 1 of the feature samples
The group number of identification signal data is identical, and each feature samples include I identification feature amount, and I is positive integer;
Obtain the contribution rate of feature identified amount: i-th of identification of p-th of feature samples is special in multiple feature samples
Sign amountWith i-th of identification feature amount of q-th of feature samplesContribution rate it is identical, that is, obtain I identification feature amount
Contribution rate, i ∈ [1, I], p and q are positive integer, p ≠ q;
Due to by pretreated identification signal data be also high dimensional data, wherein comprising with judgment model relationship degree not
By force, even incoherent attribute calculates if carrying out sample coefficient, Euclidean distance etc. under high dimension, will cause dimension calamity
It is relatively large to will lead to operand if these identification signals are input in identification model for difficulty, it is therefore desirable to therefrom choose
Part main signal improves accuracy as much as possible, alleviates an important channel of dimension disaster while reducing operand
It is that important feature is screened in attribute.Although domain expert can pick out useful attribute, ignore part association attributes
Or retains uncorrelated attribute and be likely to lead to the reduction of operating condition intelligent algorithm quality.
The feature extracting method of existing dimensionality reduction includes Principal Component Analysis, LBP feature extracting method etc..
As a preferred embodiment, using Principal Component Analysis to the pretreated identification signal data
Collection is handled, and identification signal feature set is obtained.
In the present embodiment, feature extraction is carried out using Principal Component Analysis to 11 identification signals of input, by dimensionality reduction
Dimension when being set as 3, can by 11 identification signals drops at 3 identification feature amounts, be respectively front axle torque, reared torque and
Main pump power.
When carrying out dimensionality reduction feature extraction using the method for PCA, the contribution rate of each feature identified amount also can be corresponding
Acquisition, in the present embodiment, the contribution rate of each feature identified amount is shown in Table 3.
3 principal component analysis contribution rate of table
It include the identification feature amount of identical value volume and range of product, in the present embodiment, each feature sample under each feature samples
Include 3 identification feature amounts in this, is front axle torque, reared torque and main pump power respectively, and each feature samples are corresponding
One operating condition label, that is to say, that 3 kinds of identification feature amounts are respectively included under 6 kinds of operating condition labels, such as: feature samples 1 are [preceding
Torque afterwards, reared torque, main pump power]=[0.362,0.861,0.153], the corresponding operating condition of this feature samples is fully loaded for 3-
It retreats;Feature samples 2 are [front and back torque, reared torque, main pump power]=[0.421,0.937,0.268], this feature sample
This corresponding operating condition is 1- zero load advance.
Step 4 is trained using the identification feature collection as input using the identification operating condition tally set as output
KNN model obtains loading machine operating mode's switch model, in the KNN model, p-th of feature samples and q-th feature samples it
Between distance Dis are as follows:
Wherein, p and q is positive integer,I-th kind of identification feature amount of sample is identified for p-th, 1≤i≤I, I are spy
The sum of identification feature amount in sign sample, I > 0,I-th kind of identification feature amount of sample, C are identified for q-thiKnow for i-th kind
The contribution rate of other characteristic quantity.
In the present embodiment, the 1st identification sample [front and back torque, reared torque, the main pump function concentrated for identification feature
Rate]=[0.5,0.3,0.2] and the 2nd identification sample [front and back torque, reared torque, main pump power]=[0.6,0.4,
0.1], the distance between they Dis are as follows:
In the present embodiment, KNN recognizer is merged with PCA dimension reduction method, improves the accurate of recognizer
It spends and improves algorithm recognition efficiency.
In the present embodiment, another of KNN recognizer focuses on true defining K value, if K value is smaller, model
It can become complicated and more sensitive to neighbouring training points, be easy to appear over-fitting;When K value is larger, model then can be excessively simple
It is single, it can also be played a role apart from farther away training points, be easy poor fitting.
As a preferred embodiment, K=[1,5], the value of K value is the integer between 1 to 5.
KNN algorithm when doing regime mode intelligent recognition, only it is related to a small amount of adjacent cost, can evade falling sample it
Between the problem of being unevenly distributed, and judged by limited neighbouring sample, be more suitable for the more sample set of juxtaposition.
But disadvantage is that calculation amount is larger, therefore the present invention extracts the highest principal component of contribution rate using Principal Component Analysis, is rejected
Little attribute is acted on, to achieve the purpose that reduce amount of storage and calculation amount.
KNN algorithm has good accuracy rate in terms of low dimensional, but calculates when high-dimensional complex, needs to account for
With higher memory, therefore the present invention first carries out principal component analysis before obtaining KNN model, extracts the main category in data set
Property, reduce operand.
Embodiment two
The invention also discloses a kind of loading machine operating mode's switch method, the method includes:
Using operating mode's switch model described in embodiment one to the dress by the processing of step 1- step 3 described in embodiment one
The signal data to be identified of carrier aircraft is identified.
In the present embodiment, signal data to be identified is [loading machine front axle torque, loading machine front axle revolving speed, after loading machine
Axis torque, the pressure of working barrel turn to pump pressure, engine speed]=[1450,1280,2870,8.8,16.9,2430], warp
Cross embodiment one kind step 1-3 handled after, the identification feature collection of acquisition is [front axle torque, reared torque, main pump function
Rate]=[0.451,0.942,0.287], after being identified using operating mode's switch model, recognition result is 2- spading.
Claims (4)
1. a kind of loading machine operating mode's switch model building method, method the following steps are included:
The multiple groups identification signal data of step 1, acquisition loading machine under different operating conditions, obtain identification signal data set;Described
The corresponding operating condition label of each group of identification signal data in identification signal data set, obtains identification operating condition tally set;
It is characterized by:
The operating condition label includes: unloaded advance, spading, fully loaded retrogressing, fully loaded advance, unloading and unloaded retrogressing;
Step 2 pre-processes every group of identification signal data in the identification signal data set, obtains pretreated
Identification signal data set;
The pretreatment specifically includes following procedure:
The identification signal data set is normalized between 0 to 1 by step 21, obtains the second identification signal data set;
Step 22 after carrying out removing trend term to the second identification signal data set, utilizes 3 σ method rejecting abnormalities data
Newton interpolating method interpolation vacancy item is recycled afterwards, obtains third identification signal data set;
Step 23 is filtered the third identification signal data set using Wavelet Package Denoising Method, obtains pre- place
Identification signal data set after reason;
In the Wavelet Package Denoising Method, Selection of Wavelet Basis db9-6 wavelet basis;
Step 3 is handled the pretreated identification signal data set using the feature extracting method of dimensionality reduction, is obtained
Identification feature collection;
The identification feature collection includes multiple feature samples, the identification acquired in the quantity and step 1 of the feature samples
The group number of signal data is identical, and each feature samples include I identification feature amount, and I is positive integer;
Obtain the contribution rate of feature identified amount: i-th of identification feature amount of p-th of feature samples in multiple feature samplesWith i-th of identification feature amount of q-th of feature samplesContribution rate it is identical, i ∈ [1, I], p and q are positive integer, p
≠ q obtains the contribution rate of I identification feature amount;
Step 4 trains KNN mould using the identification operating condition tally set as output using the identification feature collection as input
Type obtains loading machine operating mode's switch model, in the KNN model, between p-th of feature samples and q-th of feature samples
Distance Dis:
Wherein, CiFor the contribution rate of i-th of identification feature amount.
2. loading machine operating mode's switch model building method as described in claim 1, which is characterized in that the step 4, KNN
The value of K value is any positive integer within 5 in model.
3. loading machine operating mode's switch model building method as described in claim 1, which is characterized in that the step 3 uses
Principal Component Analysis handles the pretreated identification signal data set, obtains identification signal feature set.
4. a kind of loading machine operating mode's switch method, which is characterized in that the method includes: using any one of claim 1-3
Operating mode's switch model described in claim is to by step 2- step 3 described in any one of claim 1-3 claim
The signal data to be identified of the loading machine of processing is identified.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811043486.1A CN109359524B (en) | 2018-09-07 | 2018-09-07 | Loader condition identification model construction and identification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811043486.1A CN109359524B (en) | 2018-09-07 | 2018-09-07 | Loader condition identification model construction and identification method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109359524A true CN109359524A (en) | 2019-02-19 |
CN109359524B CN109359524B (en) | 2021-06-22 |
Family
ID=65350578
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811043486.1A Active CN109359524B (en) | 2018-09-07 | 2018-09-07 | Loader condition identification model construction and identification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109359524B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110847277A (en) * | 2019-12-19 | 2020-02-28 | 天津工程机械研究院有限公司 | Intelligent switching system and control method for power mode of loader engine |
CN110886338A (en) * | 2019-12-19 | 2020-03-17 | 天津工程机械研究院有限公司 | Semi-automatic shovel loading control system and method for loader |
CN112700085A (en) * | 2020-12-11 | 2021-04-23 | 华南理工大学 | Association rule based method, system and medium for optimizing steady-state operation parameters of complex system |
CN113565162A (en) * | 2021-07-27 | 2021-10-29 | 徐州徐工挖掘机械有限公司 | Excavator, accurate control method and system thereof, vehicle-mounted terminal and storage medium |
CN113882457A (en) * | 2021-09-27 | 2022-01-04 | 徐州徐工挖掘机械有限公司 | Hydraulic excavator working condition identification and statistics method and system and hydraulic excavator |
CN113954843A (en) * | 2021-11-12 | 2022-01-21 | 燕山大学 | Real-time working condition identification method for hydraulic mechanical stepless speed change loader |
WO2023284443A1 (en) * | 2021-07-15 | 2023-01-19 | 上海三一重机股份有限公司 | Method and apparatus for determining working condition of excavator |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103839057A (en) * | 2014-03-28 | 2014-06-04 | 中南大学 | Antimony floatation working condition recognition method and system |
CN104794492A (en) * | 2015-04-27 | 2015-07-22 | 重庆大学 | Online machine tool equipment machining and running state recognizing method based on power feature models |
CN106055778A (en) * | 2016-05-27 | 2016-10-26 | 大连楼兰科技股份有限公司 | Remote damage-assessment system and method established based on artificial intelligence for different types of vehicles |
CN106096593A (en) * | 2016-07-22 | 2016-11-09 | 厦门大学 | A kind of recognition methods of the effective operation section loading mechanical shovel process of assembling |
CN106778533A (en) * | 2016-11-28 | 2017-05-31 | 国网上海市电力公司 | PCA KSICA energy-storage system typical condition recognition methods based on kernel function |
CN107908874A (en) * | 2017-11-15 | 2018-04-13 | 上海华兴数字科技有限公司 | Operating mode's switch method and apparatus and engineering mechanical device |
CN108052975A (en) * | 2017-12-12 | 2018-05-18 | 浙江大学宁波理工学院 | It is a kind of that real-time working condition Forecasting Methodology is run based on the vehicle of core pivot and neutral net |
CN108280543A (en) * | 2018-01-17 | 2018-07-13 | 杭州安脉盛智能技术有限公司 | The adaptive equipment health degree appraisal procedure of operating mode based on classification Regression Model |
-
2018
- 2018-09-07 CN CN201811043486.1A patent/CN109359524B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103839057A (en) * | 2014-03-28 | 2014-06-04 | 中南大学 | Antimony floatation working condition recognition method and system |
CN104794492A (en) * | 2015-04-27 | 2015-07-22 | 重庆大学 | Online machine tool equipment machining and running state recognizing method based on power feature models |
CN106055778A (en) * | 2016-05-27 | 2016-10-26 | 大连楼兰科技股份有限公司 | Remote damage-assessment system and method established based on artificial intelligence for different types of vehicles |
CN106096593A (en) * | 2016-07-22 | 2016-11-09 | 厦门大学 | A kind of recognition methods of the effective operation section loading mechanical shovel process of assembling |
CN106778533A (en) * | 2016-11-28 | 2017-05-31 | 国网上海市电力公司 | PCA KSICA energy-storage system typical condition recognition methods based on kernel function |
CN107908874A (en) * | 2017-11-15 | 2018-04-13 | 上海华兴数字科技有限公司 | Operating mode's switch method and apparatus and engineering mechanical device |
CN108052975A (en) * | 2017-12-12 | 2018-05-18 | 浙江大学宁波理工学院 | It is a kind of that real-time working condition Forecasting Methodology is run based on the vehicle of core pivot and neutral net |
CN108280543A (en) * | 2018-01-17 | 2018-07-13 | 杭州安脉盛智能技术有限公司 | The adaptive equipment health degree appraisal procedure of operating mode based on classification Regression Model |
Non-Patent Citations (2)
Title |
---|
RAUNAK BORWANKAR等: ""An Optical Surface Inspection and Automatic Classification Technique Using the Rotated Wavelet Transform"", 《IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT》 * |
张泽宇: ""液力变矩器与发动机的功率匹配与转速感应控制"", 《机械科学与技术》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110847277A (en) * | 2019-12-19 | 2020-02-28 | 天津工程机械研究院有限公司 | Intelligent switching system and control method for power mode of loader engine |
CN110886338A (en) * | 2019-12-19 | 2020-03-17 | 天津工程机械研究院有限公司 | Semi-automatic shovel loading control system and method for loader |
CN112700085A (en) * | 2020-12-11 | 2021-04-23 | 华南理工大学 | Association rule based method, system and medium for optimizing steady-state operation parameters of complex system |
WO2023284443A1 (en) * | 2021-07-15 | 2023-01-19 | 上海三一重机股份有限公司 | Method and apparatus for determining working condition of excavator |
CN113565162A (en) * | 2021-07-27 | 2021-10-29 | 徐州徐工挖掘机械有限公司 | Excavator, accurate control method and system thereof, vehicle-mounted terminal and storage medium |
CN113882457A (en) * | 2021-09-27 | 2022-01-04 | 徐州徐工挖掘机械有限公司 | Hydraulic excavator working condition identification and statistics method and system and hydraulic excavator |
CN113954843A (en) * | 2021-11-12 | 2022-01-21 | 燕山大学 | Real-time working condition identification method for hydraulic mechanical stepless speed change loader |
Also Published As
Publication number | Publication date |
---|---|
CN109359524B (en) | 2021-06-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109359524A (en) | A kind of loading machine operating mode's switch model construction and recognition methods | |
CN109635677B (en) | Compound fault diagnosis method and device based on multi-label classification convolutional neural network | |
CN109766930B (en) | Method for predicting residual life of mine mechanical equipment based on DCNN model | |
CN111562108A (en) | Rolling bearing intelligent fault diagnosis method based on CNN and FCMC | |
Liang et al. | Rolling bearing fault diagnosis based on one-dimensional dilated convolution network with residual connection | |
CN112254964A (en) | Rolling bearing fault diagnosis method based on rapid multi-scale convolution neural network | |
CN206504869U (en) | A kind of rolling bearing fault diagnosis device | |
Li et al. | Extending attribute information for small data set classification | |
CN112257530B (en) | Rolling bearing fault diagnosis method based on blind signal separation and support vector machine | |
CN111539374B (en) | Rail train bearing fault diagnosis method based on multidimensional data space | |
CN113923104B (en) | Network fault diagnosis method, equipment and storage medium based on wavelet neural network | |
CN112434662B (en) | Tea leaf scab automatic identification algorithm based on multi-scale convolutional neural network | |
Zhang et al. | Multi-model ensemble deep learning method for intelligent fault diagnosis with high-dimensional samples | |
CN116593157A (en) | Complex working condition gear fault diagnosis method based on matching element learning under small sample | |
CN110147648A (en) | Automobile sensor fault detection method based on independent component analysis and sparse denoising self-encoding encoder | |
CN116481821A (en) | Engine fault early warning method and system based on big data analysis management | |
CN110159554A (en) | Centrifugal pump fault diagnostic method based on pivot analysis and sequential probability ratio test | |
CN109237007A (en) | A kind of transmission system of wheel loader efficiency optimization control method | |
CN114169377A (en) | G-MSCNN-based fault diagnosis method for rolling bearing in noisy environment | |
Meruliya et al. | Image processing for fruit shape and texture feature extraction-review | |
CN115809435A (en) | Simulator-based automobile operation fault identification method | |
CN115273791A (en) | Industrial equipment fault detection method and device based on acoustic feature driving | |
Luo et al. | A novel fault detection scheme using improved inherent multiscale fuzzy entropy with partly ensemble local characteristic-scale decomposition | |
CN114714145A (en) | Method for enhancing, comparing, learning and monitoring tool wear state by using Gelam angular field | |
Chu et al. | Deepapple: Deep learning-based apple detection using a suppression mask r-cnn |
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 |