CN116484286A - Building machine-oriented equipment process flow state real-time identification method - Google Patents
Building machine-oriented equipment process flow state real-time identification method Download PDFInfo
- Publication number
- CN116484286A CN116484286A CN202310598984.7A CN202310598984A CN116484286A CN 116484286 A CN116484286 A CN 116484286A CN 202310598984 A CN202310598984 A CN 202310598984A CN 116484286 A CN116484286 A CN 116484286A
- Authority
- CN
- China
- Prior art keywords
- time
- building machine
- state
- acceleration time
- acceleration
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 70
- 230000008569 process Effects 0.000 title claims abstract description 42
- 230000001133 acceleration Effects 0.000 claims abstract description 120
- 238000010276 construction Methods 0.000 claims abstract description 25
- 239000013598 vector Substances 0.000 claims abstract description 21
- 238000005086 pumping Methods 0.000 claims abstract description 13
- 238000012544 monitoring process Methods 0.000 claims abstract description 12
- 238000013145 classification model Methods 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 7
- 238000010801 machine learning Methods 0.000 claims abstract description 6
- 238000005070 sampling Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000003595 spectral effect Effects 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 230000009194 climbing Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 229910001294 Reinforcing steel Inorganic materials 0.000 description 2
- 230000002411 adverse Effects 0.000 description 2
- 238000009415 formwork Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 229910000831 Steel Inorganic materials 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2123/00—Data types
- G06F2123/02—Data types in the time domain, e.g. time-series data
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention discloses a building machine-oriented real-time identification method for equipment process flow states, and relates to the technical field of building engineering construction. The system aims at solving the problems that in the construction process of the existing building machine for high-rise and super-high-rise buildings, the monitoring means of the working procedure is lacking and the working state identification accuracy is low. The steps are as follows: dividing the building machine process flow into four working states of shutdown, construction, lifting and pumping, acquiring acceleration time course signals on a main stress component in the running process of the building machine in real time by adopting a triaxial vibration sensor, analyzing time domain and frequency domain characteristics of the acceleration time course signals, training combined characteristic vectors of the building machine in the four working states by a machine learning method, acquiring current actually measured acceleration time course signals and corresponding characteristic vectors, and comparing the current actually measured acceleration time course signals and the trained classification model in real time so as to rapidly and accurately identify four key working states of shutdown, construction, lifting and pumping of the current building machine.
Description
Technical Field
The invention relates to the technical field of building engineering construction, in particular to a real-time identification method for equipment process flow states of a building machine.
Background
The building machine is a concrete structure high-altitude operation platform widely applied in high-rise and super-high-rise building engineering construction, can carry other construction equipment such as a tower crane, a distributing machine, a construction lifter attachment device and the like, and is used for carrying out collaborative climbing construction along with the vertical construction running water beat of a building structure main body, so that the building machine has good overall construction efficiency. However, the structure and the process of the multi-equipment integrated building machine are complex, the safety risk in the climbing process is high, and effective building machine safety state monitoring measures are required to be adopted.
At present, the safety state monitoring of the building machine is mainly focused on the aspects of mechanical state, environmental state and the like of structural members, but the safety monitoring of the working procedure is less. Each time a concrete structure main body is constructed, the building machine is required to go through a plurality of working procedures such as climbing, binding reinforcing steel bars, opening and closing the mould, pouring concrete and the like, the parts and the components to be monitored in different working procedures are stressed, the real-time working state of the building machine cannot be effectively known by means of a traditional modeling mode and an information model updating means, and the accuracy of identifying the working state of the building machine is not high due to a plurality of construction interference factors.
Disclosure of Invention
The system aims at solving the problems that in the construction process of the existing building machine for high-rise and super-high-rise buildings, the monitoring means of the working procedure is lacking and the working state identification accuracy is low. The invention aims to provide a real-time identification method for equipment process flow states of a building machine.
The technical scheme adopted for solving the technical problems is as follows: the real-time identification method for the technological process state of the building machine comprises the following steps:
s1: according to the actual technological process of building machine, it is divided into shutdown and constructionFour working states of lifting and pumping are respectively implemented by S A ，S B 、S C 、S D A representation; a triaxial vibration sensor is arranged at the center of a main horizontal stress layer of the building machine and is tightly fixed on a main stress member of the building machine;
s3: under four working states of the building machine, respectively reading acceleration data of the triaxial vibration sensor in real time and establishing an acceleration time course curve, dividing the acceleration time course curve into n sections according to a time period delta t, so that n sections of acceleration time course signals are respectively obtained along the X, Y, Z direction under each working state, and the acceleration time course signals in three directions are a respectively x 、a y 、a z ；
S4: for one working state, in each time period delta t, the triaxial vibration sensor has m readings in each direction, and the ith reading is a respectively xi 、a yi 、a zi I=1, 2,3 … m, combining the measured acceleration time-course signals in three directions into a section of acceleration time-course signal R:
R＝{R i }，i＝1，2，3...m
s5: extracting root mean square value R of each section of acceleration time-course signal R rms Average value ofPeak-to-peak value R pp Three time domain features;
s6: carrying out Fourier transform on the data of each section of acceleration time-course signal R after the mean value is removed to obtain frequency domain data Y R And calculate and get the entropy value R H ；
S7: obtaining a combined eigenvector F= (F) of n sections of acceleration time-course signals R acquired by the triaxial vibration sensor k K=1, 2,3 … n, as follows:
s is respectively obtained according to the actual process flow of the building machine A ，S B 、S C 、S D The combined characteristic vector F of the acceleration time interval signals R in four working states;
s8, training S by using classifier A ，S B 、S C 、S D The combined feature vector F in four working states is used for deploying a classification model which is trained in an information monitoring system of the industrial personal computer;
s9: the building machine enters a real-time identification stage, reads an acceleration time course curve every a time period deltat, and obtains a current actually measured acceleration time course signal R according to steps S3-S7 c And corresponding feature vector F c Comparing the model with the classification model trained in the step S8 in real time, judging the current working state of the building making machine, and respectively outputting the current actually measured acceleration time interval signals R c S of (2) A ，S B 、S C 、S D Results of four working states;
wherein n is the total number of segments of the acceleration time course curve measured in a single direction by the single triaxial vibration sensor according to the time period deltat;
m is the number of sampling points in a single direction of the single triaxial vibration sensor in the time period delta t;
Δt is a sampling interval time period set by the triaxial vibration sensor, and is a constant;
a x 、a y 、a z acceleration time course signals measured in real time in the direction of the single triaxial vibration sensor X, Y, Z respectively;
a xi 、a yi 、a zi acceleration values measured at the i-th moment in the direction of the triaxial vibration sensor X, Y, Z within a time period deltat respectively;
r is an acceleration time course signal combined in the direction of the triaxial vibration sensor X, Y, Z in a time period delta t;
R rms a root mean square value of R;
is the average value of R;
R pp peak-to-peak value for R;
R H entropy value of R;
R i the acceleration value at the ith moment in R;
f is a combined eigenvector of n sections of acceleration time course signals R;
F c for the measured acceleration time-course signal R c Is described in (a) the combined feature vector of (b);
R c the acceleration time-course signal is actually measured in the identification process of the running state of the building machine.
The real-time identification method for the equipment process flow state of the building machine comprises the steps of firstly dividing the building machine process flow into four working states of shutdown, construction, lifting and pumping, adopting a triaxial vibration sensor to acquire acceleration time path signals on a main stress component in the running process of the building machine in real time, analyzing time domain and frequency domain characteristics of the acceleration time path signals, training combined characteristic vectors in four working states of the building machine by a machine learning method, acquiring the current actually measured acceleration time path signals and corresponding characteristic vectors, comparing the current actually measured acceleration time path signals and the corresponding characteristic vectors with a trained classification model in real time, further rapidly and accurately identifying four key working states of the current building machine, namely shutdown, construction, lifting and pumping, accurately reflecting the dynamic change of the working procedure state of the on-site building machine, enhancing the autonomous discrimination capability of the working state of the building machine, improving the intelligent degree of the building machine, effectively solving the problems of lack of a real-time monitoring means of the working state and low identification accuracy in the traditional formwork equipment construction process, and utilizing the time domain characteristics acquired by the limited triaxial vibration sensor to better identify the time domain characteristics and the characteristics of the traditional formwork equipment in the working process, thereby avoiding the problem that the artificial noise is not good in the working state identification is affected by the fact that the working state is difficult to identify the working state.
Further, the step S9 further includes a step S10: setting a time delta T which is more than or equal to 4 deltaT, taking readings of the acceleration time-course signal including a time period Δt preceding the current time every time period Δt, if M is respectively identified for the time amount Δt SA 、M SB 、M SC 、M SD The number of secondary operating states, within the following amount of time Δt, outputs the operating state by:
when M SA >M SB ≥0，M SC ＝0，M SD When=0, output state S A ；
When M SB >M SA ≥0，M SC ＝0，M SD When=0, output state S B ；
When M SA ≥0，M SB ≥0，M SC ≥1，M SD When=0, output state S C ；
When M SA ≥0，M SB ≥0，M SC ＝0，M SD When not less than 1, output state S D ；
When M SA ≥0，M SB ≥0，M SC ≥1，M SD Outputting a state E when the output value is more than or equal to 1;
wherein M is SA 、M SB 、M SC 、M SD S in the single triaxial vibration sensor within the time delta T A 、S B 、S C 、S D The number of four output states;
e is an abnormal state.
Further, the step S9 further includes: an amount of time DeltaT is set, and DeltaT is greater than or equal to (Deltat+4s), and readings of the acceleration time course signal per second over a period of time that includes the current time are taken every time period Deltat.
Further, the step S4 further includes: an interference threshold epsilon is set, and epsilon >0,
if max (|a) xi ∣,︱a yi ∣,︱a zi ∣)>ε，
Then the acceleration time course signal a is deleted x 、a y 、a z Corresponding a of (a) xi 、a yi 、a zi Values.
Further, what is said isThe step S7 further includes: setting an offset characteristic value R cr Setting an offset delta of the acceleration time-course signal R, i.eWhenever (R) i -δ)×(R i-1 -δ)<At 0, R cr Adding once;
R cr the offset characteristic value of the acceleration time interval signal R;
delta is the offset of the acceleration time interval signal R;
k is setOffset factor, k.epsilon. (0, 1).
Further, in the step S1, N triaxial vibration sensors are disposed at different main stress portions of the building machine, and according to the steps S3 to S7, the corresponding working state of each triaxial vibration sensor is determined, and the output states are as follows:
when N is SA ＝N，N SB ＝0，N SC ＝0，N SD When=0, output state S A ；
When N is SA ≥0，N SB ≥1，N SC ＝0，N SD When=0, output state S B ；
When N is SA ≥0，N SB ≥0，N SC ≥1，N SD When=0, output state S C ；
When N is SA ≥0，N SB ≥0，N SC ＝0，N SD When not less than 1, output state S D ；
When N is SA ≥0，N SB ≥0，N SC ≥1，N SD Outputting a state E when the output value is more than or equal to 1;
n is the total number of triaxial vibration sensors arranged on the main stress part of the single building machine, and N is more than or equal to 2;
N SA 、N SB 、N SC 、N SD s in N triaxial vibration sensors respectively A 、S B 、S C 、S D The number of sensors corresponding to the four output states.
Further, in the step S5, a root mean square value R of the acceleration time-course signal R is calculated rms Average value ofPeak-to-peak value R pp The calculation formula of (2) is as follows:
R pp ＝max(R i )-min(R i )
further, the step S6 includes the following steps:
s601: performing Fourier transform (FFT) on the data of each section of acceleration time-course signal R after the mean value is removed to obtain frequency domain data Y R And calculates the power spectral density S R ：
S602: the power spectral density S derived from step S601 R Calculating probability density P of each frequency point i Then according to the probability density P i Calculating entropy value R H The calculation formulas are respectively as follows:
further, the classifier in the step S8 may select a KNN, SVM machine learning algorithm to classify.
Drawings
FIG. 1 is a schematic diagram of a triaxial vibration sensor installed on a main stress member of a building machine in an embodiment of a real-time identification method for a building machine-oriented equipment process flow state according to the present invention;
FIGS. 2 to 5 illustrate S in an embodiment of the invention A ，S B 、S C 、S D Schematic diagrams of acceleration time course data of X-axis directions measured by three-axis vibration sensors under four working states;
FIGS. 6-8 illustrate S in an embodiment of the invention A ，S B 、S C 、S D Comparing the characteristic values under the four working states with a scatter diagram;
FIG. 9 is a diagram of S in an embodiment of the invention A ，S B 、S C 、S D And the classification of the four working states predicts the confusion matrix. The labels in the figures are as follows:
building a core tube 1; building machine 2; a triaxial vibration sensor 3.
Detailed Description
The invention is described in further detail below with reference to the drawings and the specific examples. It should be noted that the drawings are in a very simplified form and are all to a non-precise scale, merely for convenience and clarity in aiding in the description of embodiments of the invention.
The invention relates to a building machine-oriented equipment process flow state real-time identification method, which is described by referring to fig. 1 to 9, and comprises the following specific steps:
s1: according to the actual technological process of building machine 2, it is divided into four working states of stopping, construction, lifting and pumping, and uses S respectively A ，S B 、S C 、S D A representation; wherein, the shutdown state S A A state in which the building machine 2 is stationary and is placed on the building core tube 1, and no personnel or equipment are active on the building machine 2; construction state S B The working states of binding reinforcing steel bars, detaching and assembling templates and the like are normally carried out for the building machine; lifting state S C The building machine is in an action state of ascending or descending through a mechanical power system of the building machine; pumping state S D Finger partA concrete spreader disposed on the building machine delivers concrete to the floor through the pumping equipment; a triaxial vibration sensor 3 is arranged at the center of a main horizontal stress layer of the building machine 2, the triaxial vibration sensor 3 is tightly attached to and fixed on a main stress member of the building machine 2, the sampling frequency f of the triaxial vibration sensor 3 is not lower than 100Hz, and each triaxial vibration sensor measures acceleration data of the self point in three directions X, Y, Z in real time and sends the data to a monitoring information system in an industrial personal computer through an acquisition instrument;
s3: as shown in fig. 2 to 5, in the building machine 2S A ，S B 、S C 、S D Under four working states, the acceleration data of the triaxial vibration sensor 3 are respectively read in real time, an acceleration time course curve is established, the acceleration time course curve is divided into n sections according to a time period delta t, n sections of acceleration time course signals are respectively obtained along the X, Y, Z direction under each working state, and the acceleration time course signals in three directions are respectively a x 、a y 、a z ；
S4: for one working state, in each time period deltat, the triaxial vibration sensor 3 has m readings in each direction, and the ith reading is a respectively xi 、a yi 、a zi I=1, 2,3 … m, combining the measured acceleration time-course signals in three directions into a section of acceleration time-course signal R:
R＝{R i }，i＝l，2，3...m
s5: extracting root mean square value R of each section of acceleration time-course signal R rms Average value ofPeak-to-peak value R pp Three time domain features;
s6: performing Fourier transform (FFT) on the data of each section of acceleration time-course signal R after the mean value is removed to obtain frequency domain data Y R And calculate and get the entropy value R H ；
S7: obtain the combined eigenvector F= F of the n sections of acceleration time-course signals R acquired by the triaxial vibration sensor 3 k K=1, 2,3 … n, as follows:
s is respectively obtained according to the actual process flow of the building machine 2 A ，S B 、S C 、S D The combined characteristic vector F of the acceleration time interval signals R in four working states;
s8, training S by using classifier A ，S B 、S C 、S D The combined feature vector F in four working states is used for deploying a classification model which is trained in an information monitoring system of the industrial personal computer;
s9: the building machine enters a real-time identification stage, reads an acceleration time course curve every a time period deltat, and obtains a current actually measured acceleration time course signal R according to steps S3-S7 c And corresponding feature vector F c Comparing the model with the classification model obtained by training in the step S8 in real time, judging the current working state of the building machine 2, and respectively outputting the current actually measured acceleration time interval signals R c S of (2) A ，S B 、S C 、S D Results of four working states;
wherein n is the total number of segments of the acceleration time course curve measured in one direction by the single triaxial vibration sensor 3 according to the time period Δt;
m is the number of sampling points in a single direction of the single triaxial vibration sensor in the time period delta t;
f is the sampling frequency of the triaxial vibration sensor;
Δt is a sampling interval period set by the triaxial vibration sensor 3, and is a constant, such as 10s, 20s, 30s, etc.;
a x 、a y 、a z acceleration time course signals measured in real time in the direction of the single triaxial vibration sensor X, Y, Z respectively;
a xi 、a yi 、a zi acceleration values measured at the i-th moment in the direction of the triaxial vibration sensor X, Y, Z within a time period deltat respectively;
r is an acceleration time course signal combined in the direction of the triaxial vibration sensor X, Y, Z in a time period delta t;
R rms a root mean square value of R;
is the average value of R;
R pp peak-to-peak value for R;
R c entropy value of R;
R cr the offset characteristic value of the acceleration time interval signal R;
R i the acceleration value at the ith moment in R;
f is a combined eigenvector of n sections of acceleration time course signals R;
R c the acceleration time course signal is actually measured in the operation state identification process of the building machine;
F c for the measured acceleration time-course signal R c Is described.
According to the real-time identification method for the technological process state of the equipment facing the building machine, firstly, the technological process of the building machine 2 is divided into four working states of shutdown, construction, lifting and pumping, the triaxial vibration sensor 3 is adopted to acquire acceleration time course signals on a main stressed member in the running process of the building machine 2 in real time, time domain and frequency domain characteristics of the acceleration time course signals are analyzed, combined characteristic vectors of the building machine 2 in the four working states are trained through a machine learning method, the current actually measured acceleration time course signals and the corresponding characteristic vectors are obtained, the current actually measured acceleration time course signals and the trained classification model are compared in real time, further, four key working states of shutdown, construction, lifting and pumping of the current building machine 2 are rapidly and accurately identified, the dynamic change of the working state of the building machine 2 in the field is accurately reflected, the autonomous discrimination capability of the working state of the building machine 2 is enhanced, the intelligent degree of the building machine 2 is improved, the safety of the equipment running is guaranteed, the problem that the real-time monitoring means of the working state in the traditional building machine is lack and the identification accuracy is low is effectively solved, and the method is shown by practice, the fact that the time domain limited acceleration sensor 3 is utilized, the vibration sensor can well acquire the working state data of the building machine 2 and the key state is better, and the accuracy is better than the artificial noise can be identified, and the key state is better is prevented from being influenced.
In order to improve the reliability of the identification of the working state of the building machine, the step S9 further includes a step S10, which specifically includes the following steps:
s10: setting a time quantity delta T, wherein delta T is more than or equal to 4 delta T, if delta t=10 seconds, delta T at least comprises 40 seconds, an acceleration time course curve of the time quantity delta T is divided into at least 4 sections according to the time section delta T, readings of acceleration time course signals of the first 10 seconds including the current moment are obtained every 10 seconds, and the readings of each time section delta T are not overlapped; if M is identified in the respective time amount DeltaT SA 、M SB 、M SC 、M SD The number of secondary operating states, within the following amount of time Δt, outputs the operating state by:
when M SA >M SB ≥0，M SC ＝0，M SD When=0, output state S A ；
When M SB >M SA ≥0，M SC ＝0，M SD When=0, output state S B ；
When M SA ≥0，M SB ≥0，M SC ≥1，M SD When=0, output state S C ；
When M SA ≥0，M SB ≥0，M SC ＝0，M SD When not less than 1, output state S D ；
When M SA ≥0，M SB ≥0，M SC ≥1，M SD Outputting a state E when the output value is more than or equal to 1;
wherein M is SA 、M SB 、M SC 、M SD S in the single triaxial vibration sensor within the time delta T A 、S B 、S C 、S D Four kinds of transfusionNumber of out states. The climbing and pouring are mutually exclusive, the climbing and pouring cannot occur simultaneously in the practical process, and if the recognition results appear in the pouring and climbing within the time delta T, the output state E is an output abnormal state.
The step S9 further includes setting a time amount Δt, where Δt is greater than or equal to (Δt+4s), and if Δt=10 seconds, Δt at least includes 14 seconds, and obtaining readings including acceleration time interval signals of every second within 10 seconds before the current time every 10 seconds, thereby obtaining a working state of the building machine every second, where the current time period overlaps with the readings of 9 seconds between the previous time period, so that the time amount Δt is greatly reduced, and recognition efficiency is improved.
The step S4 further includes: an interference threshold epsilon is set, and epsilon >0,
if max (|a) xi ∣,︱a yi ∣,︱a zi ∣)>ε，
Then the acceleration time course signal a is deleted x 、a y 、a z Corresponding a of (a) xi 、a yi 、a zi And the value is obtained, so that adverse effects of interference factors such as walking of people nearby the triaxial vibration sensor are avoided.
The step S7 further includes: because of the non-synchronism of the jacking oil cylinders, the measured value of the triaxial vibration sensor 3 in the lifting state of the building machine 2 can deviate, and the deviation characteristic value R is set for improving the characteristic recognition rate of the lifting state of the building machine 2 and avoiding miscalculating due to construction interference noise cr Which is the number of times the waveform of the acceleration time-course signal intersects the time axis, gives the offset delta of one acceleration time-course signal R, namelyFor the acceleration time-course signal R, each time (R i -δ)×(R i-1 -δ)<At 0, R cr Adding once;
R cr the offset characteristic value of the acceleration time interval signal R;
delta is the offset of the acceleration time interval signal R;
k is setOffset factor, k.epsilon. (0, 1).
In order to improve the accuracy of sensing different parts of the building frame body, in the step S1, N (N is more than or equal to 2) triaxial vibration sensors 3 are arranged at different main stress parts of the building machine 2, and according to the steps S3-S7, the corresponding working state of each triaxial vibration sensor 3 is judged, and the output state is as follows:
when N is SA ＝N，N SB ＝0，N SC ＝0，N SD When=0, output state S A ；
When N is SA ≥0，N SB ≥1，N SC ＝0，N SD When=0, output state S B ；
When N is SA ≥0，N SB ≥0，N SC ≥1，N SD When=0, output state S C ；
When N is SA ≥0，N SB ≥0，N SC ＝0，N SD When not less than 1, output state S D ；
When N is SA ≥0，N SB ≥0，N SC ≥1，N SD Outputting a state E when the output value is more than or equal to 1;
n is the total number of the triaxial vibration sensors 3 arranged on the main stress part of the single building machine;
N SA 、N SB 、N SC 、N SD s in the N triaxial vibration sensors 3 respectively A 、S B 、S C 、S D The number of sensors corresponding to the four output states.
In the step S5, the root mean square value R of each acceleration time-course signal R rms Average value ofPeak-to-peak value R pp The calculation formula of (2) is as follows:
R pp ＝max(R i )-min(R i )
the step S6 includes the steps of:
s601: performing Fourier transform (FFT) on the data of each section of acceleration time-course signal R after the mean value is removed to obtain frequency domain data Y R And calculates the power spectral density S R ：
S602: the power spectral density S derived from step S601 R Calculating probability density P of each frequency point i Then according to the probability density P i Calculating entropy value R H The calculation formulas are respectively as follows:
the classifier in the step S8 may select a machine learning algorithm such as KNN, SVM, etc. to classify.
As shown in fig. 1, in this embodiment, a certain super high-rise building project is taken as an example, the project adopts a frame core tube structure system, the height of a building core tube 1 is 350.10m, the overground part is 70 layers, the building core tube 1 is constructed by adopting a building machine, a triaxial vibration sensor 3 is arranged at the web position of a section steel beam of a main stress plane of the building machine 2 and is used for collecting vibration acceleration data in real time, the sampling frequency f=125 of the triaxial vibration sensor 3, Δt takes 30 seconds, and the number m of single unidirectional sampling points of the triaxial vibration sensor 3 in the Δt time period is 3750.
Applied according to standard layersThe working state of the building machine 2 in a standard layer is divided into four conditions of shutdown, construction, lifting and pumping by S A ，S B 、S C 、S D And (3) representing. FIGS. 2 to 5 show S A ，S B 、S C 、S D Acceleration time-course data a of X-axis direction measured by the triaxial vibration sensor 3 in four working states x The amplitude and waveform characteristic change difference of the acceleration time domain information under each working state can be primarily seen.
In order to compare the related data features, each working state intercepts n=53 sections of acceleration time-course signals, and obtains a combined feature vector F of 53 acceleration time-course signals through steps S3-S7, where ε=200, k=0.3, and table 1 below gives S corresponding to the acceleration time-course signals A ，S B 、S C 、S D Typical eigenvalue calculations for the four operating states.
Table 1 typical characteristic values of building machine in four working states
Fig. 6 to 8 show characteristic values versus scatter diagrams of 53 acceleration time-course signals in each working state of the building machine, and four working states can be well distinguished through root mean square values, average values, peak-to-peak values, entropy values and offset characteristic values.
Training the combined feature vector through a KNN classifier (the number of adjacent points is K=10), deploying the trained classification model in an information monitoring system of an industrial personal computer to obtain a current actually measured acceleration time course signal and a corresponding feature vector, comparing and verifying the current actually measured acceleration time course signal and the trained classification model in real time, and judging the current working state of the building machine.
In the actual construction process, 2017500 data points are intercepted for verification according to the actual measurement results of four working states of the building machine 2, namely, each state has n=538 acceleration time-course curves. Fig. 9 shows a classification prediction confusion matrix of the building machine four working state verification results. The verification results show that the recall rates of the four working states of shutdown, construction, lifting and pumping are 99.4%, 100.0%, 84.9% and 99.3%, the overall recognition rate reaches 95.9%, and the requirement of the working state recognition accuracy of the building machine of the high-rise building is met. The practical measurement data shows that the key working state of the building machine 2 can be well identified by only utilizing the acceleration time domain and frequency domain characteristic data acquired by the limited triaxial vibration sensor 3, and the adverse effect of noise generated by peripheral manual activities can be avoided.
The above description is only illustrative of the preferred embodiments of the present invention and is not intended to limit the scope of the present invention, and any alterations and modifications made by those skilled in the art based on the above disclosure are intended to fall within the scope of the claims.
Claims (9)
1. The real-time identification method for the technological process state of the building machine is characterized by comprising the following steps:
s1: according to the actual technological process of building machine, it is divided into four working states of stopping, construction, lifting and pumping, and uses S respectively A ，S B 、S C 、S D A representation; a triaxial vibration sensor is arranged at the center of a main horizontal stress layer of the building machine and is tightly fixed on a main stress member of the building machine;
s3: under four working states of the building machine, respectively reading acceleration data of the triaxial vibration sensor in real time and establishing an acceleration time course curve, dividing the acceleration time course curve into n sections according to a time period delta t, so that n sections of acceleration time course signals are respectively obtained along the X, Y, Z direction under each working state, and the acceleration time course signals in three directions are a respectively x 、a y 、a z ；
S4: for one working state, in each time period delta t, the triaxial vibration sensor has m readings in each direction, and the ith reading is a respectively xi 、a yi 、a zi I=1, 2,3 … m, the measured acceleration time course signals of three directions are calculatedMerging into a section of acceleration time interval signal R:
R＝{R i }，i＝123...m
s5: extracting root mean square value R of each section of acceleration time-course signal R rms Average value ofPeak-to-peak value R pp Three time domain features;
s6: carrying out Fourier transform on the data of each section of acceleration time-course signal R after the mean value is removed to obtain frequency domain data Y R And calculate and get the entropy value R H ；
S7: obtaining a combined eigenvector F= (F) of n sections of acceleration time-course signals R acquired by the triaxial vibration sensor k K=1, 2,3 … n, as follows:
s is respectively obtained according to the actual process flow of the building machine A ，S B 、S C 、S D The combined characteristic vector F of the acceleration time interval signals R in four working states;
s8, training S by using classifier A ，S B 、S C 、S D The combined feature vector F in four working states is used for deploying a classification model which is trained in an information monitoring system of the industrial personal computer;
s9: the building machine enters a real-time identification stage, reads an acceleration time course curve every a time period deltat, and obtains a current actually measured acceleration time course signal R according to steps S3-S7 c And corresponding feature vector F c Comparing the model with the classification model trained in the step S8 in real time, judging the current working state of the building making machine, and respectively outputting the current actually measured acceleration time interval signals R c S of (2) A ，S B 、S C 、S D Results of four working states;
wherein n is the total number of segments of the acceleration time course curve measured in a single direction by the single triaxial vibration sensor according to the time period deltat;
m is the number of sampling points in a single direction of the single triaxial vibration sensor in the time period delta t;
Δt is a sampling interval time period set by the triaxial vibration sensor, and is a constant;
a x 、a y 、a z acceleration time course signals measured in real time in the direction of the single triaxial vibration sensor X, Y, Z respectively;
a xi 、a yi 、a zi acceleration values measured at the i-th moment in the direction of the triaxial vibration sensor X, Y, Z within a time period deltat respectively;
r is an acceleration time course signal combined in the direction of the triaxial vibration sensor X, Y, Z in a time period delta t;
R rms a root mean square value of R;
is the average value of R;
R pp peak-to-peak value for R;
R H entropy value of R;
R i the acceleration value at the ith moment in R;
f is a combined eigenvector of n sections of acceleration time course signals R;
F c for the measured acceleration time-course signal R c Is described in (a) the combined feature vector of (b);
R c the acceleration time-course signal is actually measured in the identification process of the running state of the building machine.
2. The method for identifying the state of the equipment process flow for the building machine in real time according to claim 1, wherein the step S9 further comprises the step S10 of: setting a time delta T which is more than or equal to 4 delta T at intervalsThe interval Δt takes readings of the acceleration time-course signal including the time period Δt preceding the current time, if M is respectively identified for the amount of time Δt SA 、M SB 、M SC 、M SD The number of secondary operating states, within the following amount of time Δt, outputs the operating state by:
when M SA >M SB ≥0，M SC ＝0，M SD When=0, output state S A ；
When M SB >M SA ≥0，M SC ＝0，M SD When=0, output state S B ；
When M SA ≥0，M SB ≥0，M SC ≥1，M SD When=0, output state S C ；
When M SA ≥0，M SB ≥0，M SC ＝0，M SD When not less than 1, output state S D ；
When M SA ≥0，M SB ≥0，M SC ≥1，M SD Outputting a state E when the output value is more than or equal to 1;
wherein M is SA 、M SB 、M SC 、M SD S in the single triaxial vibration sensor within the time delta T A 、S B 、S C 、S D The number of four output states;
e is an abnormal state.
3. The method for identifying the state of the equipment process flow for the building machine in real time according to claim 1, wherein the step S9 further comprises: an amount of time DeltaT is set, and DeltaT is greater than or equal to (Deltat+4s), and readings of the acceleration time course signal per second over a period of time that includes the current time are taken every time period Deltat.
4. The method for identifying the state of the equipment process flow for the building machine in real time according to claim 1, wherein the step S4 further comprises: an interference threshold epsilon is set, and epsilon >0,
if max (|a) xi ∣,︱a yi ∣,︱a zi ∣)>ε，
Then the acceleration time course signal a is deleted x 、a y 、a z Corresponding a of (a) xi 、a yi 、a zi Values.
5. The method for identifying the state of the equipment process flow for the building machine in real time according to claim 1, wherein the step S7 further comprises: setting an offset characteristic value R cr Setting an offset delta of the acceleration time-course signal R, i.eWhenever (R) i -δ)×(R i-1 -δ)<At 0, R cr Adding once;
R cr the offset characteristic value of the acceleration time interval signal R;
delta is the offset of the acceleration time interval signal R;
k is setOffset factor, k.epsilon. (0, 1).
6. The method for identifying the state of the equipment process flow for the building machine in real time according to claim 1, wherein in the step S1, N triaxial vibration sensors are arranged at different main stress portions of the building machine, and according to the steps S3 to S7, the corresponding working state of each triaxial vibration sensor is determined, and the output state is as follows:
when N is SA ＝N，N SB ＝0，N SC ＝0，N SD When=0, output state S A ；
When N is SA ≥0，N SB ≥1，N SC ＝0，N SD When=0, output state S B ；
When N is SA ≥0，N SB ≥0，N SC ≥1，N SD When=0, output state S C ；
When N is SA ≥0，N SB ≥0，N SC ＝0，N SD When not less than 1, output state S D ；
When N is SA ≥0，N SB ≥0，N SC ≥1，N SD Outputting a state E when the output value is more than or equal to 1;
n is the total number of triaxial vibration sensors arranged on the main stress part of the single building machine, and N is more than or equal to 2;
N SA 、N SB 、N SC 、N SD s in N triaxial vibration sensors respectively A 、S B 、S C 、S D The number of sensors corresponding to the four output states.
7. The building machine-oriented equipment process state real-time identification method according to claim 1, wherein the method comprises the following steps: in the step S5, the root mean square value R of each acceleration time-course signal R rms Average value ofPeak-to-peak value R pp The calculation formula of (2) is as follows:
R pp ＝max(R i )-min(R i )
8. the method for identifying the state of the equipment process flow for the building machine in real time according to claim 1, wherein the step S6 comprises the steps of:
s601: performing Fourier transform (FFT) on the data of each section of acceleration time-course signal R after the mean value is removed to obtain frequency domain data Y R And calculates the power spectral density S R ：
S602: the power spectral density S derived from step S601 R Calculating probability density P of each frequency point i Then according to the probability density P i Calculating entropy value R H The calculation formulas are respectively as follows:
9. the building machine-oriented equipment process state real-time identification method according to claim 1, wherein the method comprises the following steps: the classifier in the step S8 may select KNN, SVM machine learning algorithm to classify.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310598984.7A CN116484286A (en) | 2023-05-25 | 2023-05-25 | Building machine-oriented equipment process flow state real-time identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310598984.7A CN116484286A (en) | 2023-05-25 | 2023-05-25 | Building machine-oriented equipment process flow state real-time identification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116484286A true CN116484286A (en) | 2023-07-25 |
Family
ID=87225253
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310598984.7A Pending CN116484286A (en) | 2023-05-25 | 2023-05-25 | Building machine-oriented equipment process flow state real-time identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116484286A (en) |
-
2023
- 2023-05-25 CN CN202310598984.7A patent/CN116484286A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| de Almeida Cardoso et al. | Automated real-time damage detection strategy using raw dynamic measurements | |
| Pan et al. | A novel method of vibration modes selection for improving accuracy of frequency-based damage detection | |
| CN105865522A (en) | Bridge structure monitoring system | |
| CN105550506B (en) | Bridge wind vibration response DIRECT FORECASTING METHOD based on long-term measured data | |
| CN101504333B (en) | Progressive health monitoring method for cable structure median cord system based on strain monitoring | |
| Cheung et al. | The application of statistical pattern recognition methods for damage detection to field data | |
| Xu et al. | Prediction analysis of vortex-induced vibration of long-span suspension bridge based on monitoring data | |
| CN110017929A (en) | Vessel bump bridge load and damage method for synchronously recognizing based on minor structure sensitivity analysis | |
| CN110455517A (en) | A kind of tower health monitor method of wind power generating set | |
| CN104111133A (en) | Cable force monitoring system and monitoring method of wide-span pre-stress stay cable steel structure | |
| Entezami et al. | Improving feature extraction via time series modeling for structural health monitoring based on unsupervised learning methods | |
| KR101328889B1 (en) | Structural health monitoring system based on measured displacement | |
| WO2022267613A1 (en) | Passive excitation-based online calibration method for bridge structure strain monitoring system | |
| US20200249118A1 (en) | Systems and methods for measuring structural element deflections | |
| CN110188399B (en) | Dam safety monitoring single-measuring-point evaluation method based on multiple correlation sequences | |
| CN106446384B (en) | A kind of damnification recognition method of bridging crane main beam structure | |
| CN104807661B (en) | A kind of high-rise and tall and slender structure Dynamic testing evaluation on bearing capacity method | |
| EP3951344A1 (en) | Methods and systems for damage evaluation of structural assets | |
| KR101227776B1 (en) | Designing method for displacement estimation on construction using ann and estimating method for displacement of construction | |
| CN101699248A (en) | Progressive health monitoring method for identifying loose carrying cable based on angle monitoring | |
| CN116484286A (en) | Building machine-oriented equipment process flow state real-time identification method | |
| CN112926384B (en) | Automatic modal identification method based on power spectrum transfer ratio and support vector machine | |
| CN115950485A (en) | Full-automatic intelligent monitoring method in tunnel | |
| Chase et al. | Detection of Damage in OperatingWind Turbines by Signature Distances | |
| Shen et al. | A boom damage prediction framework of wheeled cranes combining hybrid features of acceleration and Gaussian process regression |
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 |