Equipment process flow state real-time identification method for skyscraper building machine
By installing triaxial vibration sensors on the building construction machine, the time and frequency domain characteristics of the acceleration time history signal are acquired and analyzed in real time. Combined with machine learning methods, the four key working states of the building construction machine are accurately identified, which solves the shortcomings of traditional monitoring methods and improves construction safety and intelligence.
Patent Information
- Application Number
- PCT/CN2023/138640
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-25
- Filing Date
- 2023-12-14
- Publication Date
- 2024-11-28
AI Technical Summary
In existing technologies, there is a lack of monitoring methods for the construction process of building machines, the accuracy of identifying the working status is low, and it is impossible to effectively know their real-time working status, which poses safety risks, especially in the construction of high-rise and super high-rise buildings.
A triaxial vibration sensor is used to acquire the acceleration time history signal of the main load-bearing component of the building machine in real time. By analyzing the time domain and frequency domain characteristics and using machine learning methods to train combined feature vectors, a classification model is established to identify four working states in real time: shutdown, construction, lifting, and pumping.
It improves the accuracy of identifying the working status of the building construction machine, enhances its intelligence, ensures construction safety, effectively avoids the impact of surrounding noise interference, and improves the reliability and accuracy of identification.
Smart Images

Figure CN2023138640_28112024_PF_FP_ABST
Abstract
Description
Real-time identification method of equipment process status for building construction machines Technical Field
[0001] The present invention relates to the technical field of building engineering construction, and in particular to a real-time identification method for equipment process status of a building construction machine. Background Art
[0002] Building cranes are aerial work platforms used extensively in the construction of high-rise and super-high-rise buildings. They can carry other construction equipment, such as tower cranes, concrete placing booms, and construction hoist attachments. They ascend and descend in sync with the vertical construction rhythm of the main building structure, resulting in excellent overall construction efficiency. However, the structure and process of these multi-device integrated building cranes are complex, posing significant safety risks during the ascent process. Therefore, effective safety monitoring measures for the building cranes are necessary.
[0003] Currently, safety monitoring of building construction machines primarily focuses on the mechanical and environmental conditions of structural components, while monitoring of their process flows is less common. For each floor of the concrete structure, the machine undergoes multiple processes, including climbing, tying rebar, opening and closing molds, and pouring concrete. Different processes require monitoring of different areas and components. Traditional modeling and information model update methods are unable to effectively capture the machine's real-time operating status. Furthermore, due to numerous construction interference factors, accurate identification of the machine's operating status is limited.
[0004] Summary of the Invention
[0005] The existing building construction machines for high-rise and super high-rise buildings have problems such as lack of process monitoring means and low accuracy in identifying working status during the construction process. The purpose of the present invention is to provide a real-time identification method for the equipment process status of building construction machines.
[0006] The technical solution adopted by the present invention to solve the technical problem is: a real-time identification method for the equipment process status of a building construction machine, the steps are as follows:
[0007] S1: According to the actual process flow of the building construction machine, it is divided into four working states: shutdown, construction, lifting, and pumping. A , S B 、S C 、S D Indicates that a three-axis vibration sensor is set at the center of the main horizontal load-bearing layer of the building construction machine, and the three-axis vibration sensor is closely fixed on the main load-bearing component of the building construction machine;
[0008] S3: Under the four working states of the building construction machine, the acceleration data of the three-axis vibration sensor are read in real time and the acceleration time history curve is established. The acceleration time history curve is divided into n segments according to the time period Δt, so that in each working state, n segments of acceleration time history signals are obtained along the X, Y, and Z directions respectively. The acceleration time history signals in the three directions are a x 、a y 、a z ;
[0009] S4: For a working state, in each time period Δt, the triaxial vibration sensor has m readings in each direction, and the i-th reading is a xi 、a yi 、a zi , i = 1, 2, 3…m, combine the measured acceleration time history signals in the three directions into an acceleration time history signal R: R={R i}, i=1,2,3...m
[0010] S5: Extract the root mean square value R of each acceleration time history signal R rms ,average value Peak-to-peak value R pp Three time domain features;
[0011] S6: Perform Fourier transform on the data of each acceleration time history signal R after removing the mean value to obtain the frequency domain data Y R , and calculate the entropy value R H ;
[0012] S7: Obtain the combined characteristic vector F = ﹛F of the n-segment acceleration time history signal R collected by the triaxial vibration sensor k ﹜, k=1,2,3…n, as follows:
[0013] According to the actual process flow of the building machine, S A , S B 、S C 、S D Combined characteristic vector F of acceleration time history signal R under four working states;
[0014] S8: Use classifier to train S A , S B 、S C 、S D The combined feature vector F under the four working conditions is obtained, and the trained classification model is deployed in the information monitoring system of the industrial computer;
[0015] S9: The building machine enters the real-time recognition stage, and reads a section of the acceleration time curve every time period Δt. According to steps S3 to S7, the current measured acceleration time signal R is obtained. c and the corresponding eigenvector F c , and compare it with the classification model trained in step S8 in real time, and judge the current working state of the building machine, and output the current measured acceleration time signal R c S A , S B 、S C 、S D Four working status results;
[0016] Where n is the total number of segments of the acceleration time history curve measured in one direction by a single triaxial vibration sensor according to the time period Δt;
[0017] m is the number of sampling points of a single triaxial vibration sensor in a single direction within the time period Δt;
[0018] Δt is the sampling interval time period set by the triaxial vibration sensor, which is a constant;
[0019] a x 、a y 、a z They are respectively the acceleration time history signals measured in real time in the X, Y, and Z directions by a single three-axis vibration sensor;
[0020] a xi 、a yi 、a zi are the acceleration values measured by the triaxial vibration sensor in the X, Y, and Z directions at the i-th moment within the time period Δt;
[0021] R is the acceleration time history signal of the triaxial vibration sensor in the X, Y, and Z directions within the time period Δt;
[0022] R rms is the root mean square value of R;
[0023] is the average value of R;
[0024] R pp is the peak-to-peak value of R;
[0025] R H is the entropy value of R;
[0026] R i is the acceleration value at the i-th moment in R;
[0027] F is the combined eigenvector of n acceleration time history signals R;
[0028] Fc is the measured acceleration time history signal R c The combined feature vector of
[0029] R c This is the acceleration time history signal measured during the building construction machine operation status identification process.
[0030] The present invention is a method for real-time identification of equipment process status of building construction machines. First, the process of the building construction machine is divided into four working states: shutdown, construction, lifting, and pumping. A three-axis vibration sensor is used to obtain the acceleration time history signal on the main load-bearing component during the operation of the building construction machine in real time, and its time domain and frequency domain characteristics are analyzed. The combined feature vectors of the four working states of the building construction machine are trained by machine learning methods to obtain the current measured acceleration time history signal and the corresponding feature vector, which are compared with the trained classification model in real time, thereby quickly and accurately identifying the shutdown, construction, lifting, and pumping of the current building construction machine. It sends four key working states, accurately reflects the dynamic changes of the process status of the on-site building construction machine, enhances the autonomous judgment ability of the building construction machine's working status, improves the intelligence level of the building construction machine, ensures the safety of equipment operation, and effectively solves the problems of lack of real-time monitoring means of the working status and low recognition accuracy during the construction process of traditional formwork equipment. Practice has shown that this method uses the acceleration time domain and frequency domain feature data collected by the limited three-axis vibration sensor to better identify the key working states of the building construction machine, and can avoid the adverse effects of noise generated by surrounding human activities, thereby improving the recognition accuracy.
[0031] Furthermore, after step S9, step S10 is further included: setting a time amount ΔT, ΔT≥4Δt, obtaining the reading of the acceleration time history signal of the time period Δt before the current moment every time period Δt, if M is identified within the time amount ΔT SA 、M SB 、M SC 、M SD The number of working states, in the next amount of time ΔT, output the working state by the following method:
[0032] When M SA >M SB ≥0, M SC =0,M SD =0, the output state is S A ;
[0033] When M SB >M SA ≥0, M SC =0,M SD =0, the output state is S B ;
[0034] When MSA ≥0, M SB ≥0, M SC ≥1, M SD =0, the output state is S C ;
[0035] When M SA ≥0, M SB ≥0, M SC =0,M SD ≥1, the output state S D ;
[0036] When M SA ≥0, M SB ≥0, M SC ≥1, M SD When ≥1, the output state is E;
[0037] Among them, M SA 、M SB 、M SC 、M SD They are respectively the S of a single triaxial vibration sensor within a time period ΔT A 、S B 、S C 、S D The number of the four output states;
[0038] E is an abnormal state.
[0039] Furthermore, the step S9 further includes: setting a time amount ΔT, and ΔT≥(Δt+4s), and obtaining a reading of the acceleration time history signal per second in a time period before the current moment every time period Δt.
[0040] Furthermore, the step S4 further includes: setting an interference threshold ε, and ε>0,
[0041] If max(︱a xi ∣,︱a yi ∣,︱a zi ∣)>ε,
[0042] Then delete the acceleration time history signal a x 、a y 、a z The corresponding a xi 、a yi 、a zi value.
[0043] Furthermore, the step S7 further includes: setting the offset characteristic value R cr , set an offset δ of the acceleration time history signal R, that is Whenever (R i -δ)×(Ri-1 -δ)<0, R cr Increase once;
[0044] R cr is the offset characteristic value of the acceleration time history signal R;
[0045] δ is the offset of the acceleration time history signal R;
[0046] k is set Offset factor, k∈(0,1).
[0047] Furthermore, in step S1, N three-axis vibration sensors are set at different main stress-bearing parts of the building construction machine. According to steps S3 to S7, the corresponding working state of each three-axis vibration sensor is determined, and the output state is as follows:
[0048] When N SA =N,N SB =0, N SC =0, N SD =0, the output state is S A ;
[0049] When N SA ≥0, N SB ≥1, N SC =0, N SD =0, the output state is S B ;
[0050] When N SA ≥0, N SB ≥0, N SC ≥1, N SD =0, the output state is S C ;
[0051] When N SA ≥0, N SB ≥0, N SC =0, N SD ≥1, the output state S D ;
[0052] When N SA ≥0, N SB ≥0, N SC ≥1, N SD When ≥1, the output state is E;
[0053] N is the total number of triaxial vibration sensors installed on the main stress-bearing parts of a single building construction machine, N≥2;
[0054] N SA 、N SB 、N SC 、N SDThey are S of N triaxial vibration sensors A 、S B 、S C 、S D The number of sensors corresponding to the four output states.
[0055] Furthermore, in step S5, the root mean square value R of each acceleration time history signal R rms ,average value Peak-to-peak value R pp The calculation formula is as follows: R pp =max(R i )-min(R i )
[0056] Furthermore, step S6 includes the following steps:
[0057] S601: Perform Fourier transform (FFT) on the data of each acceleration time history signal R after removing the mean value to obtain frequency domain data Y R , and calculate the power spectral density S R :
[0058] S602: The power spectrum density S obtained in step S601 R Calculate the probability density P for each frequency point i , then according to the probability density P i Calculate the entropy value R H , and the calculation formulas are:
[0059] Furthermore, the classifier in step S8 may select KNN or SVM machine learning algorithm for classification. BRIEF DESCRIPTION OF THE DRAWINGS
[0060] FIG1 is a schematic diagram showing a method for real-time identification of equipment process status of a building construction machine according to an embodiment of the present invention, in which a three-axis vibration sensor is installed on a main load-bearing component of the building construction machine;
[0061] Figures 2 to 5 are diagrams of an embodiment of the present invention. A , S B 、S C 、S D Schematic diagram of the X-axis acceleration time history data measured by the triaxial vibration sensor under four working conditions;
[0062] Figures 6 to 8 are diagrams of an embodiment of the present invention. A , S B 、S C 、SD Scatter plot comparing characteristic values under four working conditions;
[0063] FIG9 is an embodiment of the present invention S A , S B 、S C 、S D The confusion matrix of classification prediction for four working states. The numbers in the figure are as follows:
[0064] Building core 1; building machine 2; triaxial vibration sensor 3. DETAILED DESCRIPTION
[0065] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the drawings are all in very simplified form and not in exact proportions, and are only used for the purpose of conveniently and clearly explaining the embodiments of the present invention.
[0066] 1 to 9 illustrate the present invention for the equipment process state real-time identification method of the building construction machine, the specific steps are as follows:
[0067] S1: According to the actual process flow of the building construction machine 2, it is divided into four working states: shutdown, construction, lifting, and pumping. A , S B 、S C 、S D Indicates; Among them, the shutdown state S A The building construction machine 2 is stationary and placed on the building core tube 1, and there is no personnel or equipment activity on the building construction machine 2; the construction state S B The building machine is in normal working state of tying steel bars, disassembling formwork, etc.; the lifting state S C Refers to the state in which the building machine is moving up or down through its own mechanical power system; pumping state S D Refers to the operating state of the concrete placing boom deployed on the building construction machine, which is delivering concrete to the floors through pumping equipment. A three-axis vibration sensor 3 is installed at the center of the main horizontal load-bearing layer of the building construction machine 2. The three-axis vibration sensor 3 is tightly fixed to the main load-bearing component of the building construction machine 2. The sampling frequency f of the three-axis vibration sensor 3 is not less than 100Hz. Each three-axis vibration sensor measures the acceleration data of the X, Y, and Z directions of its own point in real time, and sends the data to the monitoring information system in the industrial computer through the data collector.
[0068] S3: As shown in Figures 2 to 5, in the S of the building machine 2 A , S B 、S C 、S DUnder the four working states, the acceleration data of the three-axis vibration sensor 3 are read in real time and the acceleration time history curve is established. The acceleration time history curve is divided into n segments according to the time period Δt, so that under each working state, n segments of acceleration time history signals are obtained along the X, Y, and Z directions respectively. The acceleration time history signals in the three directions are a x 、a y 、a z ;
[0069] S4: For a working state, in each time period Δt, the triaxial vibration sensor 3 has m readings in each direction, and the i-th reading is a xi 、a yi 、a zi , i = 1, 2, 3…m, combine the measured acceleration time history signals in the three directions into an acceleration time history signal R: R={R i}, i=1,2,3...m
[0070] S5: Extract the root mean square value R of each acceleration time history signal R rms ,average value Peak-to-peak value R pp Three time domain features;
[0071] S6: Perform Fourier transform (FFT) on the data of each acceleration time history signal R after removing the mean value to obtain the frequency domain data Y R , and calculate the entropy value R H ;
[0072] S7: Obtain the combined characteristic vector F of the n-segment acceleration time history signal R collected by the triaxial vibration sensor 3 =﹛F k ﹜, k=1,2,3…n, as follows:
[0073] According to the actual process flow of building machine 2, S A , S B 、S C 、S D Combined characteristic vector F of acceleration time history signal R under four working states;
[0074] S8: Use classifier to train S A , S B 、S C 、S D The combined feature vector F under the four working conditions is obtained, and the trained classification model is deployed in the information monitoring system of the industrial computer;
[0075] S9: The building machine enters the real-time recognition stage, and reads a section of the acceleration time curve every time period Δt. According to steps S3 to S7, the current measured acceleration time signal R is obtained. c and the corresponding eigenvector F c , and compare it with the classification model trained in step S8 in real time, and judge the current working state of the building construction machine 2, and output the current measured acceleration time history signal R c S A , S B 、S C 、S D Four working status results;
[0076] Wherein, n is the total number of segments of the acceleration time history curve measured in a single direction by a single triaxial vibration sensor 3 according to the time period Δt;
[0077] m is the number of sampling points of a single triaxial vibration sensor in a single direction within the time period Δt;
[0078] f is the sampling frequency of the triaxial vibration sensor;
[0079] Δt is the sampling interval time period set by the triaxial vibration sensor 3 and is a constant, such as 10s, 20s, 30s, etc.;
[0080] a x 、a y 、a z They are respectively the acceleration time history signals measured in real time in the X, Y, and Z directions by a single three-axis vibration sensor;
[0081] a xi 、a yi 、a zi are the acceleration values measured by the triaxial vibration sensor in the X, Y, and Z directions at the i-th moment within the time period Δt;
[0082] R is the acceleration time history signal of the triaxial vibration sensor in the X, Y, and Z directions within the time period Δt;
[0083] R rms is the root mean square value of R;
[0084] is the average value of R;
[0085] R pp is the peak-to-peak value of R;
[0086] R c is the entropy value of R;
[0087] R cr is the offset characteristic value of the acceleration time history signal R;
[0088] R i is the acceleration value at the i-th moment in R;
[0089] F is the combined eigenvector of n acceleration time history signals R;
[0090] R c It is the acceleration time history signal measured during the operation status identification process of the building construction machine;
[0091] F c is the measured acceleration time history signal R c The combined feature vector of .
[0092] The present invention is a method for real-time identification of equipment process status of building construction machines. First, the process of the building construction machine 2 is divided into four working states: shutdown, construction, lifting, and pumping. The three-axis vibration sensor 3 is used to obtain the acceleration time history signal on the main load-bearing component during the operation of the building construction machine 2 in real time, and its time domain and frequency domain characteristics are analyzed. The combined feature vectors of the four working states of the building construction machine 2 are trained by machine learning methods to obtain the current measured acceleration time history signal and the corresponding feature vector, which are compared with the trained classification model in real time, thereby quickly and accurately identifying the shutdown, construction, lifting, and pumping of the current building construction machine 2. It sends four key working states, accurately reflects the dynamic changes of the process status of the on-site building construction machine 2, enhances the autonomous judgment ability of the working status of the building construction machine 2, improves the intelligence level of the building construction machine 2, ensures the safety of equipment operation, and effectively solves the problems of lack of real-time monitoring means of the working status and low recognition accuracy during the construction process of traditional formwork equipment. Practice has shown that this method uses the limited acceleration time domain and frequency domain feature data collected by the three-axis vibration sensor 3 to better identify the key working states of the building construction machine 2, and can avoid the adverse effects of noise generated by surrounding human activities, thereby improving the recognition accuracy.
[0093] In order to improve the reliability of the working status identification of the building construction machine, step S10 is further included after the above step S9, and the specific content is as follows:
[0094] S10: Set a time amount ΔT, and ΔT ≥ 4Δt. If Δt = 10 seconds, then ΔT includes at least 40 seconds. The acceleration time history curve of the time amount ΔT is divided into at least 4 segments according to the time period Δt. The reading of the acceleration time history signal including the 10 seconds before the current moment is obtained every 10 seconds, and the readings of each time period Δt do not overlap. If M is identified in the time amount ΔT, SA 、M SB 、M SC 、M SD The number of working states, in the next amount of time ΔT, output the working state by the following method:
[0095] When M SA >M SB ≥0, M SC =0,M SD =0, the output state is S A ;
[0096] When M SB >M SA ≥0, M SC =0,M SD =0, the output state is S B ;
[0097] When M SA ≥0, M SB ≥0, M SC ≥1, M SD =0, the output state is S C ;
[0098] When M SA ≥0, M SB ≥0, M SC =0,M SD ≥1, the output state S D ;
[0099] When M SA ≥0, M SB ≥0, M SC ≥1, M SD When ≥1, the output state is E;
[0100] Among them, M SA 、M SB 、M SC 、M SD They are respectively the S of a single triaxial vibration sensor within a time period ΔT A 、S B 、S C 、S D The number of the four output states. Climbing and pouring are mutually exclusive states and will not occur simultaneously in practice. If both pouring and climbing have recognition results within the time ΔT, the output state is E, and E is the abnormal output state.
[0101] The step S9 also includes setting a time amount ΔT, and ΔT ≥ (Δt + 4s). If Δt = 10 seconds, then ΔT includes at least 14 seconds. Every 10 seconds, the reading of the acceleration time series signal for each second within 10 seconds before the current moment is obtained, and then the working status of the building construction machine per second is obtained. There is a 9-second overlap in readings between the current time period and the previous time period, which greatly reduces the time amount ΔT and improves the recognition efficiency.
[0102] The step S4 further includes: setting an interference threshold ε, where ε>0,
[0103] If max(︱a xi ∣,︱a yi ∣,︱a zi ∣)>ε,
[0104] Then delete the acceleration time history signal a x 、a y 、a z The corresponding a xi 、a yi 、a zi value, thereby avoiding the adverse effects of interference factors such as people walking near the triaxial vibration sensor.
[0105] The step S7 further includes: due to the asynchronous nature of the lifting cylinder, the measurement value of the three-axis vibration sensor 3 in the lifting state of the building machine 2 will be offset. In order to improve the feature recognition rate of the lifting state of the building machine 2 and avoid miscalculation due to construction interference noise, an offset feature value R is set. cr , which is the number of times the acceleration time-history signal waveform intersects the time axis. Given an offset δ of the acceleration time-history signal R, that is, For the acceleration time signal R, whenever (R i -δ)×(R i-1 -δ)<0, R cr Increase once;
[0106] R cr is the offset characteristic value of the acceleration time history signal R;
[0107] δ is the offset of the acceleration time history signal R;
[0108] k is set Offset factor, k∈(0,1).
[0109] In order to improve the accuracy of sensing at different parts of the building machine frame, in step S1, N (N≥2) triaxial vibration sensors 3 are set at different main stress-bearing parts of the building machine 2. According to steps S3 to S7, the corresponding working state of each triaxial vibration sensor 3 is determined, and the output state is as follows:
[0110] When N SA =N,N SB =0, N SC =0, N SD =0, the output state is S A ;
[0111] When N SA ≥0, N SB ≥1, N SC =0, N SD =0, the output state is S B ;
[0112] When N SA ≥0, N SB ≥0, N SC ≥1, N SD =0, the output state is S C ;
[0113] When N SA ≥0, N SB ≥0, N SC =0, N SD ≥1, the output state S D ;
[0114] When N SA ≥0, N SB ≥0, N SC ≥1, N SD When ≥1, the output state is E;
[0115] N is the total number of triaxial vibration sensors 3 installed on the main stress-bearing parts of a single building construction machine;
[0116] N SA 、N SB 、N SC 、N SD There are N three-axis vibration sensors 3 in S A 、S B 、S C 、S D The number of sensors corresponding to the four output states.
[0117] In step S5, the root mean square value R of each acceleration time signal R rms ,average value Peak-to-peak value R pp The calculation formula is as follows: R pp =max(R i )-min(R i )
[0118] The step S6 comprises the following steps:
[0119] S601: Perform Fourier transform (FFT) on the data of each acceleration time history signal R after removing the mean value to obtain frequency domain data Y R , and calculate the power spectral density S R :
[0120] S602: The power spectrum density S obtained in step S601 R Calculate the probability density P for each frequency point i, then according to the probability density P i Calculate the entropy value R H , and the calculation formulas are:
[0121] The classifier in step S8 may select a machine learning algorithm such as KNN or SVM for classification.
[0122] As shown in Figure 1, this embodiment takes a super high-rise construction project as an example. The project adopts a frame core tube structural system. The building core tube 1 is 350.10 meters high and has 70 floors above ground. The building core tube 1 is constructed using a building construction machine. A three-axis vibration sensor 3 is arranged at the web position of the steel beam on the main load-bearing plane of the building construction machine 2 to collect vibration acceleration data in real time. The sampling frequency of the three-axis vibration sensor 3 is f=125, and Δt is 30 seconds. Then, the number m of sampling points in a single direction of a single three-axis vibration sensor 3 within the Δt time period is 3750.
[0123] According to the standard floor construction process, the working state of the building machine 2 on a standard floor is divided into four states: shutdown, construction, lifting, and pumping, which are respectively represented by S A , S B 、S C 、S D Figures 2 to 5 show S A , S B 、S C 、S D Under four working conditions, the acceleration time history data a in the X-axis direction measured by the triaxial vibration sensor 3 x ,It can be preliminarily seen that the amplitude and waveform characteristics of the acceleration time domain information under various working states vary greatly.
[0124] In order to compare the relevant data features, n = 53 acceleration time history signals are intercepted for each working state. Through steps S3-S7, the combined feature vector F of 53 acceleration time history signals is obtained, where ε = 200 and k = 0.3. Table 1 below shows the S corresponding to the acceleration time history signal. A , S B 、S C 、S D Calculation results of typical eigenvalues for four working states.
[0125] Table 1 Typical characteristic values of building construction machine in four working states
[0126] Figures 6 to 8 show scatter plots comparing the characteristic values of 53 acceleration time history signals under each working state of the building construction machine. The four working states can be better distinguished by the root mean square value, average value, peak-to-peak value, entropy value, and offset characteristic value.
[0127] The combined feature vector is trained through the KNN classifier (the number of adjacent points K = 10), and the trained classification model is deployed in the information monitoring system of the industrial computer to obtain the current measured acceleration time-history signal and the corresponding feature vector. It is compared and verified in real time with the trained classification model, and the current working status of the building construction machine is determined.
[0128] During actual construction, based on the measured results of the four operating states of the building construction machine 2, 2,017,500 data points were collected for each state for verification, resulting in n = 538 acceleration time history curves for each state. Figure 9 shows the classification prediction confusion matrix for the verification results of the four operating states of the building construction machine. The verification results show that the recall rates for the four operating states of the building construction machine are 99.4%, 100.0%, 84.9%, and 99.3%, respectively, with an overall recognition rate of 95.9%, meeting the accuracy requirements for identifying the operating states of building construction machines for high-rise and super-high-rise buildings. Measured data examples demonstrate that using only the limited time-domain and frequency-domain acceleration feature data collected by a single triaxial vibration sensor 3, it is possible to effectively identify the key operating states of the building construction machine 2 while avoiding the adverse effects of noise generated by surrounding human activities.
[0129] The above description is only a description of the preferred embodiments of the present invention and does not limit the scope of the present invention. Any changes and modifications made by ordinary technicians in the field of the present invention based on the above disclosure shall fall within the scope of the claims.
Claims
1. A real-time identification method for equipment process status of a building construction machine, characterized in that: Here are the steps: S1: According to the actual process flow of the building construction machine, it is divided into four working states: shutdown, construction, lifting and pumping, respectively. A , S B , S C , S D Indicates that a three-axis vibration sensor is arranged at the center of the main horizontal stress-bearing layer of the building construction machine, and the three-axis vibration sensor is closely fixed on the main stress-bearing component of the building construction machine; S3: Under the four working states of the building construction machine, the acceleration data of the three-axis vibration sensor are read in real time and the acceleration time history curve is established. The acceleration time history curve is divided into n segments according to the time period Δt, so that in each working state, n segments of acceleration time history signals are obtained along the X, Y, and Z directions respectively. The acceleration time history signals in the three directions are a x 、a y 、a z ; S4: For a working state, in each time period Δt, the triaxial vibration sensor has m readings in each direction, and the i-th reading is a xi 、a yi 、a zi , i = 1, 2, 3…m, and combine the measured acceleration time-history signals in the three directions into one acceleration time-history signal R: S5: Extract the root mean square value R of each acceleration time history signal R rms ,average value Peak-to-peak value R pp Three time domain features; S6: Perform Fourier transform on the data of each acceleration time history signal R after removing the mean value to obtain the frequency domain data Y R , and calculate the entropy value R H ; S7: Obtain the combined characteristic vector F =﹛F of the n-segment acceleration time history signal R collected by the triaxial vibration sensor k ﹜, k=1,2,3…n, as follows: According to the actual process flow of the building construction machine, S A , S B , S C , S D The combined characteristic vector F of the acceleration time history signal R under four working conditions; S8: Use classifier to train S A , S B , S C , S D The combined feature vector F under four working conditions is obtained, and the trained classification model is deployed in the information monitoring system of the industrial computer; S9: The building construction machine enters the real-time identification stage, reads a section of the acceleration time curve every time period Δt, and obtains the current measured acceleration time signal R according to steps S3 to S7. c and the corresponding eigenvector F c , and compare it with the classification model trained in step S8 in real time, and judge the current working state of the building construction machine state, and output the currently measured acceleration time signal R c S A , S B , S C , S D Four working status results; Where n is the total number of segments of the acceleration time history curve measured in a single direction by a single triaxial vibration sensor according to the time period Δt; m is the number of sampling points of a single triaxial vibration sensor in a single direction within the time period Δt; Δt is the sampling interval time period set by the triaxial vibration sensor, which is a constant; a x 、a y 、a z They are respectively the acceleration time history signals measured in real time in the X, Y, and Z directions by a single three-axis vibration sensor; a xi 、a yi 、a zi are respectively the acceleration values measured by the triaxial vibration sensor in the X, Y, and Z directions at the i-th moment within the time period Δt; R is the acceleration time history signal of the triaxial vibration sensor in the X, Y, and Z directions combined within the time period Δt; R rms is the root mean square value of R; is the average value of R; R pp is the peak-to-peak value of R; R H is the entropy value of R; R i is the acceleration value at the i-th moment in R; F is the combined characteristic vector of n acceleration time history signals R; F c is the measured acceleration time signal R c The combined feature vector of R c It is the acceleration time history signal measured during the operation status identification process of the building construction machine.
2. The real-time identification method for equipment process status of a building construction machine according to claim 1 is characterized in that: The step S9 further includes a step S10: setting a time amount ΔT, ΔT≥4Δt, obtaining the reading of the acceleration time history signal of the time period Δt before the current moment every time period Δt, and if M is identified within the time amount ΔT SA 、M SB 、M SC 、M SD The number of working states, in the next amount of time ΔT, output the working state by the following method: When M SA >M SB ≥0, M SC =0,M SD =0, the output state is S A ; When M SB >M SA ≥0, M SC =0,M SD =0, the output state is S B ; When M SA ≥0, M SB ≥0, M SC ≥1, M SD =0, the output state is S C ; When M SA ≥0, M SB ≥0, M SC =0,M SD ≥1, the output state is S D ; When M SA ≥0, M SB ≥0, M SC ≥1, M SD When ≥1, the output state is E; Among them, M SA 、M SB 、M SC 、M SD They are respectively the S of a single triaxial vibration sensor within a time quantity ΔT A , S B , S C , S D The number of the four output states; E is an abnormal state.
3. The real-time identification method for equipment process status of a building construction machine according to claim 1 is characterized in that: The step S9 further includes: setting a time amount ΔT, and ΔT≥(Δt+4s), and obtaining a reading of the acceleration time history signal per second in a time period before the current moment every time period Δt.
4. The real-time identification method for equipment process status of a building construction machine according to claim 1 is characterized in that: The step S4 further includes: setting an interference threshold ε, where ε>0, If max(|a xi |, |a yi |, |a zi |) > ε, Then delete the acceleration time history signal a x 、a y 、a z The corresponding a xi 、a yi 、a zi value.
5. The real-time identification method of equipment process status for building construction machines according to claim 1 is characterized in that: The step S7 also includes: setting the offset characteristic value R cr , set an acceleration time signal R The offset δ is Whenever (R i -δ)×(R i-1 -δ)<0, R cr Increase once; R cr is the offset characteristic value of the acceleration time history signal R; δ is the offset of the acceleration time history signal R; k is set Offset factor, k∈(0,1).
6. The real-time identification method for equipment process status of a building construction machine according to claim 1 is characterized in that: In the step S1, N three-axis vibration sensors are set at different main stress-bearing parts of the building construction machine. According to the steps S3 to S7, the corresponding working state of each three-axis vibration sensor is determined, and the output state is as follows: When N SA =N,N SB =0,N SC =0,N SD =0, the output state is S A ; When N SA ≥0,N SB ≥1, N SC =0,N SD =0, the output state is S B ; When N SA ≥0,N SB ≥0,N SC ≥1, N SD =0, the output state is S C ; When N SA ≥0,N SB ≥0,N SC =0,N SD ≥1, the output state is S D ; When N SA ≥0,N SB ≥0,N SC ≥1, N SD When ≥1, the output state is E; N is the total number of triaxial vibration sensors installed on the main stress-bearing parts of a single building construction machine, N≥2; N SA 、N SB 、N SC 、N SD They are respectively S A , S B , S C , S D The number of sensors corresponding to the four output states.
7. The method for real-time identification of equipment process status for a building construction machine according to claim 1 is characterized in that: In step S5, the root mean square value R of each acceleration time history signal R rms ,average value Peak-to-peak value R pp The calculation formula is as follows: R pp =max(R i )-min(R i ) 8. The real-time identification method for equipment process status of a building construction machine according to claim 1 is characterized in that: The step S6 comprises the following steps: S601: Perform Fourier transform (FFT) on the data of each acceleration time history signal R after removing the mean value to obtain frequency domain data Y R , and calculate the power spectral density S R : S602: Power spectrum density S obtained in step S601 R Calculate the probability density P for each frequency point i , then according to the probability density P i Calculate the entropy value R H , and the calculation formulas are:
9. The method for real-time identification of equipment process status for a building construction machine according to claim 1, characterized in that: The classifier in step S8 may select KNN or SVM machine learning algorithm for classification.
Citation Information
Patent Citations
Equipment state detection method, device and system
CN115077685A
Centrifugal pump multi-fault identification method based on vibration intensity and statistical indexes
CN115310479A
Equipment process flow state real-time identification method for building machine
CN116484286A
Building building machine working state identification method based on multi-source sensing information
CN116628603A
Machine-tool-state determination system and machine-tool-state determination method
US20200026262A1
Cited By
Steel slag flow real-time detection method and system based on triaxial accelerometer
CN122361195A
