CN-122027983-A - Construction site personnel positioning electronic fence system

CN122027983ACN 122027983 ACN122027983 ACN 122027983ACN-122027983-A

Abstract

The invention provides a construction site personnel positioning electronic fence system which comprises a multi-source sensing fusion acquisition module, a dynamic environment adaptation calibration module, a three-dimensional fence layering modeling module and a space-time correlation boundary crossing judgment module. The invention aims to provide a high-precision, intelligent and strong-adaptability construction site personnel positioning electronic fence system, which solves the technical defects of the traditional system by means of multi-source sensing fusion, dynamic self-adaptive positioning calibration, three-dimensional fence layering modeling, space-time collaborative border crossing judgment and hierarchical touch execution technical design, improves the positioning accuracy and border crossing judgment rationality of the construction site personnel, realizes the full-dimension and intelligent safety management and control of the construction site personnel, adapts to the high-temperature, high-humidity, strong-electromagnetic and complex-terrain dynamic operation environment of the construction site, and reduces the safety accident rate caused by personnel border crossing of the construction site from the technical aspect.

Inventors

Chen Chuankan
SUN BIN
Ya Hanqing
WU YUNPENG
XU ZHIYAO
SONG YAZHOU
LIU QIAN

Assignees

沈阳鑫鼎恒科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260210

Claims (10)

1. A job site personnel location electronic fence system, comprising: The multi-source sensing fusion acquisition module is used for acquiring personnel positioning terminal acquisition data, environment sensing data and base station positioning data, performing synchronous access, cleaning, preliminary fusion and standardized encapsulation, and outputting multi-source positioning fusion original data; The dynamic environment adaptation calibration module is used for acquiring multi-source positioning fusion original data, acquiring a historical positioning calibration data set and a construction site real-time environment characteristic set, performing bidirectional interaction fusion calculation by adopting a variation modal decomposition-kernel extreme learning machine algorithm and a space-time attention mechanism-Gaussian process regression algorithm, completing dynamic drift correction and space-time error compensation of positioning data, and outputting corrected multi-source positioning fusion data.
2. The system of claim 1, further comprising a three-dimensional fence hierarchical modeling module for acquiring corrected multi-source positioning fusion data, simultaneously acquiring electronic fence basic boundary data, three-dimensional terrain data and operation area hierarchical division data, performing three-dimensional basic fence modeling, fence hierarchical division and positioning coordinate hierarchical matching, and outputting three-dimensional fence hierarchical matching data.
3. The construction site personnel positioning electronic fence system of claim 2, further comprising a space-time correlation boundary crossing judgment module, wherein the space-time correlation boundary crossing judgment module is used for acquiring three-dimensional fence layered matching data, acquiring personnel operation authority data sets, personnel history track data sets and network time service data, performing bidirectional interaction fusion calculation by adopting a fuzzy evidence theory algorithm and a time sequence graph rolling network algorithm, completing space-time cooperative boundary crossing judgment, and outputting space-time correlation boundary crossing judgment data.
4. The job site personnel location electronic fence system as set forth in claim 3, further comprising: and the hierarchical touch execution module is used for acquiring the space-time association boundary crossing judgment data, acquiring a boundary crossing hierarchical execution rule base, matching the corresponding execution grades and triggering matched execution measures.
5. The construction site personnel positioning electronic fence system according to claim 1 is characterized in that working logic of the variation modal decomposition-kernel extreme learning machine algorithm is that positioning data in multisource positioning fusion original data is subjected to variation modal decomposition, positioning effective signals and environment interference noise are separated, core features of the positioning data are extracted, a history positioning calibration data set and a construction site real-time environment feature set are combined, the positioning core features and the environment features are used as input, model training and calculation are conducted through the kernel extreme learning machine algorithm, positioning drift correction coefficients of the current construction site environment are dynamically predicted and adapted, and a core calculation basis is provided for drift correction of the positioning data.
6. The construction site personnel positioning electronic fence system according to claim 5 is characterized in that working logic of the space-time attention mechanism-Gaussian process regression algorithm is that positioning drift correction coefficients output by a variation modal decomposition-kernel extreme learning machine algorithm are received firstly, preliminary drift correction is carried out on multi-source positioning fusion original data, then the space-time attention mechanism is introduced, positioning data weights of different environment areas and different time nodes of a construction site are focused, space-time error compensation calculation is carried out on the preliminarily corrected positioning data through the Gaussian process regression algorithm in combination with a real-time environment feature set of the construction site, positioning deviation in space-time dimension is corrected, and positioning data after error compensation is output.
7. The construction site personnel positioning electronic fence system according to claim 6 is characterized in that the interaction logic of the variation modal decomposition-kernel extreme learning machine algorithm and the space-time attention mechanism-Gaussian process regression algorithm is that a positioning drift correction coefficient output by the variation modal decomposition-kernel extreme learning machine algorithm is used as a core input parameter for preliminary drift correction by the space-time attention mechanism-Gaussian process regression algorithm, and after space-time error compensation is completed by the space-time attention mechanism-Gaussian process regression algorithm, the calculated error compensation value is reversely fed back to the variation modal decomposition-kernel extreme learning machine algorithm to dynamically optimize model training parameters thereof, so that bidirectional interaction cooperation of the two algorithms is realized, and positioning calibration accuracy is improved.
8. The construction site personnel positioning electronic fence system according to claim 3 is characterized in that working logic of the fuzzy evidence theory algorithm is that positioning coordinates in three-dimensional fence layered matching data are used as core basis, a personnel operation authority data set is combined, space position relations between positioning coordinates and corresponding combined layered fences and personnel operation authorities are converted into fuzzy evidences, fusion calculation is conducted on various fuzzy evidences through evidence synthesis rules, uncertainty of space out-of-range is quantified, a space out-of-range credibility value is output, and probability that personnel positioning coordinates exceed the range of corresponding fences is reflected.
9. The construction site personnel positioning electronic fence system according to claim 8 is characterized in that working logic of the time sequence diagram convolution network algorithm is that three-dimensional fence layered matching data and personnel history track data sets are taken as input, time stamps marked by network time service data are combined, time rules and movement trends of historical operation tracks of personnel are mined through the time sequence diagram convolution network, whether current personnel positioning coordinates meet operation authority requirements in a time dimension is analyzed based on the time rules and the movement trends, uncertainty of time out-of-range is quantized, and time out-of-range reliability values are output.
10. The construction site personnel positioning electronic fence system according to claim 9 is characterized in that the interaction logic of the fuzzy evidence theory algorithm and the time sequence diagram rolling network algorithm is that a space out-of-range reliability value output by the fuzzy evidence theory algorithm is used as an auxiliary input characteristic of the time sequence diagram rolling network algorithm for analyzing time out-of-range risks and used for adjusting the weight of time dimension out-of-range judgment, the time out-of-range reliability value output by the time sequence diagram rolling network algorithm is reversely fed back to the fuzzy evidence theory algorithm, the synthetic weight of fuzzy evidence is optimized, misjudgment and missed judgment caused by single space dimension judgment are reduced, and time-space collaborative out-of-range soft judgment is achieved.

Description

Construction site personnel positioning electronic fence system Technical Field The invention relates to the technical field of personnel positioning and electronic fence, in particular to a construction site personnel positioning electronic fence system. Background The construction site personnel positioning electronic fence system is a core technical means of construction site safety management, acquires personnel position information through a positioning terminal, and combines an electronic fence boundary to realize automatic judgment and alarm of personnel out-of-range behaviors, so that the construction site personnel positioning electronic fence system is a key technology for preventing personnel from entering a high-risk area to cause safety accidents. At present, the traditional construction site personnel positioning electronic fence system still has a plurality of technical defects, and is difficult to adapt to the actual requirements of a complex construction site: the positioning acquisition layer adopts a single GPS/Beidou or UWB positioning source, so that the problems of weak signals, data loss and the like are easy to occur in areas such as tower crane shielding, foundation pit, strong electromagnetic and the like on a construction site, an effective environment adaptation calibration means is not available, the positioning drift phenomenon is serious, and the hidden error hidden danger is judged for the subsequent out-of-range; The modeling level of the electronic fence adopts a planar two-dimensional modeling mode, three-dimensional terrain features such as a foundation pit, a scaffold, floors, a high-altitude platform and the like on a construction site are ignored, out-of-range behaviors caused by height differences cannot be identified, and a control blind area exists; The out-of-range judgment level is only based on the spatial coordinates to carry out single yes/no two-classification hard judgment, the time authority and time trend rule of personnel operation are not combined, the uncertainty of spatial matching is not considered, misjudgment and missed judgment are easily generated due to the temporary movement of personnel and small errors of coordinates, and the invalid cost of site safety management is increased; On the algorithm application level, even if a small number of algorithms are introduced into the existing system, the existing system is mainly operated independently and subjected to unidirectional data transmission, interaction fusion and parameter optimization among the algorithms are avoided, and the overall positioning and judging accuracy of the system cannot be improved through algorithm cooperation. Disclosure of Invention The invention provides a construction site personnel positioning electronic fence system, which aims to provide a high-precision, intelligent and strong-adaptability construction site personnel positioning electronic fence system, and solves the technical defects of the traditional system by adopting the technical design of multi-source sensing fusion, dynamic self-adaptive positioning calibration, three-dimensional fence layering modeling, space-time collaborative border crossing judgment and grading touch execution, improves the positioning accuracy of the construction site personnel and the border crossing judgment rationality, realizes the full-dimension and intelligent safety management and control of the construction site personnel, adapts to the dynamic operation environment with high temperature, high humidity, strong electromagnetism and complex topography of the construction site, and reduces the safety accident occurrence rate caused by personnel border crossing of the construction site from the technical level. In order to achieve the above purpose, the invention adopts the following technical scheme: A construction site personnel positioning electronic fence system comprises a multisource sensing fusion acquisition module, a dynamic environment adaptation calibration module, a time-space correlation cross-border judgment module and a time-space correlation cross-border judgment module, wherein the multisource sensing fusion acquisition module is used for acquiring personnel positioning terminal acquisition data, environment sensing data and base station positioning data, synchronously accessing, cleaning, preliminary fusion and standardized encapsulation, outputting multisource positioning fusion original data, the dynamic environment adaptation calibration module is used for acquiring multisource positioning fusion original data, simultaneously acquiring historical positioning calibration data sets and construction site real-time environment characteristic sets, performing bidirectional interactive fusion calculation by adopting a variation modal decomposition-nuclear extreme learning algorithm and a time-space attention mechanism-Gaussian process regression algorithm, completing dynamic drift correction and time-space error compe