CN-120688734-B - Hoisting construction supervision optimization management system based on BIM

CN120688734BCN 120688734 BCN120688734 BCN 120688734BCN-120688734-B

Abstract

The invention discloses a hoisting construction supervision and optimization management system based on BIM, which comprises a sensing layer, a decision layer, an execution layer and an optimization layer, wherein the sensing layer is used for acquiring multi-source heterogeneous data in real time, the decision layer is used for generating candidate paths meeting crane kinematics constraint by utilizing an incremental RRT# algorithm, the upper layer is embedded into a distributed Q-learning frame, each crane is used as an independent intelligent body, collaborative learning is carried out through a shared experience pool, a path planning strategy is dynamically optimized according to environment sensing data and construction progress requirements, a fuzzy sliding mode control algorithm is developed, meanwhile, a crane motion compensation parameter is calculated by combining LSTM-transporter crane cart, trolley travel speed, lifting hook lifting speed and a wire rope disturbance prediction model, an optimal control instruction is generated in advance based on predicted crane travel and lifting data, the instruction generated by the decision layer is issued to construction equipment, and the BIM model is constructed, and the equipment execution result and structural safety monitoring data are fed back to the decision layer.

Inventors

LIU WENHUI
HU XIONGFENG
XU JUNYANG
WU YUECHAO
MA JINPENG

Assignees

中国电建集团华东勘测设计研究院有限公司

Dates

Publication Date: 20260505
Application Date: 20250609

Claims (5)

1. BIM-based hoisting construction supervision and optimization management system is characterized by comprising the following components: The sensing layer is used for deploying 5G+ edge computing nodes, collecting multi-source heterogeneous data of millimeter wave radar, UWB+ laser radar, miniature laser radar weather stations and fiber bragg grating sensor networks in real time, adopting a Kalman filtering-particle filtering fusion algorithm, constructing a space-time map of a dynamic obstacle and carrying out space-time alignment on the multi-sensor data, embedding the fiber bragg grating sensor networks and deploying digital twin mirror interfaces, and completing real-time monitoring of strain, inclination and steel structure stress of key components of the crane and synchronization of a BIM model and physical entity states; The decision layer is used for generating candidate paths meeting crane kinematics constraint by the bottom layer through an incremental RRT# algorithm and combining a non-uniform sampling strategy and an intelligent pruning mechanism, and comprises the following steps of; initializing, namely setting a starting point and a target point in a BIM model; Non-uniform sampling, namely sampling in a key area according to a risk thermodynamic diagram, and sampling probability Risk value according to location x And obstacle density Calculation, the calculation formula is expressed as Where Ω is the sampling space, the integral term is used to normalize the probability; Path expansion, namely randomly expanding paths from a starting point by utilizing an incremental RRT# algorithm, wherein each expansion considers the kinematic constraint of a crane; intelligent pruning, namely removing redundant paths, and reserving paths which are low in path cost and meet kinematic constraint; path optimization, namely smoothing the reserved path; outputting candidate paths, namely outputting candidate paths meeting all constraints; The upper layer is embedded with a distributed Q-learning framework, each crane is used as an independent intelligent body, collaborative learning is carried out through a shared experience pool, and a path planning strategy is dynamically optimized according to environment perception data and construction progress requirements, and the method comprises the following steps: Initializing, namely acquiring current state information from the BIM by the intelligent agent, and initializing a Q value table; Experience collection, namely collecting experience by each agent in the interaction process and storing the experience in a shared experience pool; the intelligent agent randomly extracts a batch of experience from the experience pool to learn, and updates a Q value table by using a Q-learning algorithm according to the extracted experience; path planning, namely selecting an optimal action by an agent to carry out path planning according to the learned Q value table; In the process of hoisting the large components of the machine set, the intelligent agent selects the optimal action according to the environmental perception data and the construction progress requirement, the intelligent agent executes the selected action, the self position is updated, and the intelligent agent receives the rewarding signal fed back by the environment according to the result after executing the action; Meanwhile, a fuzzy sliding mode control algorithm is developed, crane motion compensation parameters are calculated by combining an LSTM-converter crane cart, a crane travelling speed, a lifting hook lifting speed and a steel wire rope disturbance prediction model, model prediction control is introduced, and an optimal control instruction is generated in advance based on predicted travelling, lifting speed and steel wire rope disturbance data, and the method comprises the following steps: fuzzy sliding mode control algorithm by defining sliding mode surface function Representing a deviation of the system state from the desired state, the sliding mode surface function being represented as , wherein, Is the error in position or angle and, Is the rate of change of the error and, Is a design parameter and designs a control law Allowing the system state to reach the sliding mode surface in a limited time and move along the sliding mode surface to reach the balance point, wherein the control law is expressed as , wherein, Is an equivalent control item, which is used to control the operation of the device, The control item is switched, and the gain of the control item is adjusted by utilizing the fuzzy logic so as to adapt to different working conditions; An LSTM layer processing time sequence data, capturing short-term changes of moving, lifting or descending speeds and wire rope disturbance, a transducer encoder processing space data, capturing space correlation of moving, walking, lifting and descending speeds at different positions and a hoisted large piece when the wire rope is disturbed, training the model through historical crane moving, walking, lifting and lifting speeds and wire rope disturbance data, optimizing a loss function, and enabling the model to predict hoisting data of the large piece in a future period of time; The integrated model predictive control comprises the steps of inputting current lifting speed data, outputting future lifting speed and a steel wire rope disturbance predicted value, solving a finite time domain optimal control problem in each control period, solving the optimal control problem by using an optimization algorithm to obtain an optimal control sequence, and applying the first element of the optimal control sequence to a crane system so as to generate an optimal control instruction in advance based on an LSTM-transducer crane cart, a crane travelling speed, a lifting hook lifting speed and a steel wire rope disturbance predicted model, rolling optimization and feedback correction; the execution layer is used for sending the path planning instruction and the wind disturbance compensation control instruction generated by the decision layer to construction equipment through an OPC UA over TSN protocol; and the optimization layer is used for constructing a BIM model, integrating geometric information, construction progress, structural response and equipment state, and feeding back equipment execution results and structural safety monitoring data to the decision layer.
2. The hoisting construction supervision and optimization management system based on BIM according to claim 1, wherein the multisource heterogeneous data collected by the sensing layer comprises environment sensing data, structure sensing data, equipment state data and personnel and safety data, the environment sensing data comprises millimeter wave radar data, UWB+laser radar data and micro laser radar weather station data, the structure sensing data comprises fiber bragg grating sensor network data and digital twin mirror interface data, the equipment state data comprises crane operation state data and elevator operation state data, and the personnel and safety data comprises personnel positioning data and safety monitoring video data.
3. The hoisting construction supervision and optimization management system based on BIM according to claim 1, wherein the perception layer adopts a Kalman filtering-particle filtering fusion algorithm to perform the specific steps of constructing a dynamic obstacle space-time map and aligning the space time of multi-sensor data: the data preprocessing, namely cleaning the data and unifying the data formats, detecting and eliminating abnormal values by using a threshold method or a statistical method, defining a unified data structure, performing preliminary processing on the original sensor data, removing noise and abnormal values, and converting the data of different sensors into the unified format; kalman filtering, namely recursively estimating the state of the system by using a state transition matrix and an observation matrix through two steps of prediction and updating, and carrying out optimal state estimation on the linear Gaussian system; particle filtering, namely representing the state distribution of the system through a group of random samples, and approximating the real posterior distribution through resampling and weight updating to complete the state estimation of the nonlinear non-Gaussian system; And (3) fusing, namely performing state estimation on the linear part by using Kalman filtering, performing state estimation on the nonlinear part by using particle filtering, and fusing the results of the two filters by using weighted average or interactive multi-model algorithm.
4. The hoisting construction supervision and optimization management system based on BIM according to claim 1, wherein the sensing layer performs the specific steps of digital twin mirror synchronization: deploying a sensor, namely deploying a fiber bragg grating sensor network on a key component of the crane; The digital twin mirror image interface is integrated with the BIM model, wherein a unique identifier is allocated to each component of the BIM model, each sensor in the sensor network is allocated to a unique identifier, the association of sensor data and the BIM model components is carried out by establishing a mapping relation between the identifiers, and when the sensor data changes, the data is synchronized into the BIM model through the interface, so that the information of the model components is driven to update.
5. The hoisting construction supervision and optimization management system based on the BIM according to claim 1, wherein the hoisting construction supervision and optimization management system is issued to construction equipment through an OPC UA over TSN protocol, and comprises the following steps: coding and packaging the instruction, namely coding a path planning result and a compensation parameter into a data structure of OPC UA; constructing an OPC UA over TSN network transmission, namely constructing an OPC UA-based construction equipment information model, defining nodes and methods of equipment, sensors and actuators, configuring time synchronization and flow scheduling parameters of the TSN network, and sending encoded instructions to the construction equipment through an OPC UA over TSN protocol; and the equipment end instruction decoding and execution is that an OPC UA client is arranged on construction equipment, the instruction from a decision layer is received and decoded, and the OPC UA client analyzes instruction parameters in an OPC UA data structure according to the specific model and configuration of the equipment.

Description

Hoisting construction supervision optimization management system based on BIM Technical Field The invention relates to the technical field of hoisting safety of large components of pumped storage hydropower station units, in particular to a hoisting construction supervision and optimization management system based on BIM. Background The pumped storage power station factory building has a complex multi-layer space structure, and large components of the machine set such as a stator, a rotor, a rotating wheel, a top cover, a water inlet valve, a GCB (generator outlet switch) and the like are subjected to severe hoisting construction environments, so that the large hoisting operation becomes a link with highest risk concentration. The bridge crane in main workshop is a core crane for hoisting large parts, and is composed of a bridge frame (also called a cart), a lifting mechanism, a trolley, a cart traveling mechanism, a control room, a trolley conductive device (auxiliary sliding wire), a crane main power supply conductive device (main sliding wire) and the like. The cart runs longitudinally along the tracks laid on the elevated frames at both sides, and the trolley runs transversely along the tracks laid on the bridge frame. The traditional supervision mode depends on manual experience and offline planning, is difficult to cope with complex scenes of multivariable coupling such as dynamic movement, lifting speed and steel wire rope disturbance, equipment coordination and structural deformation of a crane, for example, collision accidents can be caused by swinging of a lifting hook caused by walking or lifting or stopping of a crane cart and a trolley, the construction efficiency can be reduced due to path conflict when a bridge crane of a main factory building and lifting cross operation between the same layer or different working layers are carried out, potential safety hazards are more likely to be caused by uncertainty of actions such as multi-working-face operation when a power station is installed for the first time or a unit is overhauled, a digital base is provided for construction supervision by maturation of a Building Information Model (BIM) technology, the defects of the traditional supervision in terms of instantaneity, accuracy and predictability can be exactly overcome, and the like, and an intelligent supervision system covering a 'perception-decision-execution-optimization' full chain can be constructed by fusing the BIM with the technical depth of the Internet of things, the artificial intelligence, so that the core functions such as dynamic path planning, the moving speed, the lifting speed and the lifting rope disturbance and the final safety precaution can be realized in a hydropower station, especially in a pumped storage power station, the core operation and the construction safety margin are remarkable. In the prior art, the improved RRT algorithm is tightly integrated with the BIM model. The BIM provides rich building information, the improved RRT algorithm can more accurately plan a path meeting actual construction requirements by calling the information, however, although the improved RRT algorithm considers equipment kinematics constraint, in civil construction of a pumping and storage station and electromechanical equipment installation, environment dynamic change is extremely fast (such as multiple working surfaces, temporary obstacles and the like), real-time response capability of the algorithm may be insufficient, construction equipment continues to move according to the original path, the actual working conditions may deviate, and collision risk is increased, so that the hoisting construction supervision and optimization management system based on the BIM is provided. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides a hoisting construction supervision and optimization management system based on BIM. In order to achieve the above purpose, the present invention adopts the following technical scheme: a hoist and mount construction supervision optimization management system based on BIM includes: The sensing layer is used for deploying 5G+ edge computing nodes, collecting multi-source heterogeneous data of millimeter wave radar, UWB+ laser radar, miniature laser radar weather stations and fiber bragg grating sensor networks in real time, adopting a Kalman filtering-particle filtering fusion algorithm, constructing a space-time map of a dynamic obstacle and carrying out space-time alignment on the multi-sensor data, embedding the fiber bragg grating sensor networks and deploying digital twin mirror interfaces, and completing real-time monitoring of strain, inclination and steel structure stress of key components of the crane and synchronization of a BIM model and physical entity states; The decision layer utilizes an incremental RRT# algorithm and combines a non-uniform sampling strategy and an intelligent pruning mechanism to generate