CN-121561348-B - Tunnel lining construction intelligent trolley and control system thereof

CN121561348BCN 121561348 BCN121561348 BCN 121561348BCN-121561348-B

Abstract

The invention relates to the technical field of intelligent control of tunnel construction and discloses an intelligent trolley for tunnel lining construction and a control system thereof. The system includes determining a state change point in time by event listening, intercepting multi-source sensor data therefrom and extracting features. The pattern matching module generates a quality difference metric by comparing the criteria to the anomaly pattern. A causal analysis framework with the operating frequency as an adjusting variable is innovatively introduced, the influence weight of each sensor data stream on the quality is dynamically evaluated, and key quality factors are accurately identified. And carrying out partition combination on the optimized feature set based on the construction waveform time sequence relationship, and generating a control instruction by adopting an attention mechanism. The scheme realizes the transition from static threshold judgment to dynamic causal inference, realizes self-adaptive control through time sequence context understanding, and improves the accuracy of quality diagnosis and the adaptability of the system to complex working conditions.

Inventors

YAN SU
WANG HONGYI
JIN LIUJIE
MA WEIBIN
HUANG MINGLI
SHI YUFENG
WANG BAIQUAN
LU YONGLONG
AN ZHELI
HUANG QUNGUANG
DANG DONG

Assignees

中铁十六局集团有限公司
华东交通大学
中国铁道科学研究院集团有限公司铁道建筑研究所
北京交通大学
中铁隧道勘察设计研究院有限公司
中国铁建大桥工程局集团有限公司

Dates

Publication Date: 20260508
Application Date: 20260121

Claims (10)

1. Control system of tunnel lining construction intelligent trolley, characterized in that, control system includes: The event monitoring module is used for monitoring occurrence signals of trolley operation events and identifying a reference time point of construction state change in a continuously acquired sensor data stream; the data acquisition and processing module is used for intercepting multi-source sensor data in a fixed time interval before and after a reference time point and carrying out transformation processing on the multi-source sensor data so as to derive a characteristic representation of a construction state; The pattern matching module is used for sending the characteristic representation into the pattern matching unit, evaluating the pattern consistency of the standard construction mode and the abnormal construction mode, and generating a quality difference measurement; The causal analysis module is used for using the trolley operating frequency as an adjusting variable, processing the quality difference measurement through a causal analysis framework and estimating the influence weight of each sensor data stream on the construction quality; The feature optimization module is used for integrating the construction resistance dynamic index and the trolley movement track index, optimizing the selection conditions of the influence weights, screening out feature representations corresponding to the high influence weights, and forming an optimized feature set; The instruction generation module is used for partitioning and combining the optimized feature set according to the time sequence alignment relation of the construction waveform, and generating a driving instruction by adopting an attention distribution mechanism.
2. The control system of the intelligent trolley for tunnel lining construction according to claim 1, wherein the monitoring of the occurrence signal of the trolley operation event comprises continuously receiving a control pulse signal of a trolley actuator, detecting edge change of the control pulse signal as an event trigger mark, recording an accurate time stamp of the event trigger mark, comparing the time stamp with a time difference of a displacement abrupt change point in a displacement sensor data sequence, and confirming that the time stamp is a reference time point of construction state change when the time difference is smaller than a set tolerance.
3. The control system of the intelligent trolley for tunnel lining construction according to claim 1, wherein the intercepting of the multi-source sensor data in the fixed time interval based on the reference time point comprises the steps of expanding preset time lengths forwards and backwards by taking the reference time point as a center, obtaining original readings of an acceleration sensor, a pressure sensor and a displacement sensor in the time window, carrying out moving average filtering on the original readings to inhibit noise, then calculating a data variance and a mean value of each sensor channel as preliminary statistics, carrying out standardization processing on the preliminary statistics to unify data scales of all channels, and then carrying out dimension reduction through principal component analysis to generate a characteristic representation of a construction state.
4. The control system of the intelligent trolley for tunnel lining construction according to claim 1, wherein the feature representation is sent to the pattern matching unit, the mode consistency between the standard construction mode and the abnormal construction mode is evaluated, and the quality difference metric is generated by constructing a double-branch structure of the pattern matching unit, wherein each branch comprises a convolution layer and a full connection layer, the feature representations of the standard construction mode and the abnormal construction mode are processed respectively, calculating Euclidean distances between output vectors of the two branches, converting the Euclidean distances into similarity scores through a sigmoid function, and adjusting weights of the feature representations according to the similarity scores so that the similarity scores in the standard mode are maximized, and the similarity scores in the abnormal mode are minimized, so that the quality difference metric is output.
5. The control system of the intelligent trolley for tunnel lining construction according to claim 1, wherein the quality difference measurement is processed through a causal analysis framework by using the trolley operation frequency as an adjustment variable, the influence weight of each sensor data stream on the construction quality is estimated by counting the occurrence times of trolley operation events in unit time, dividing the operation frequency into a low interval, a medium interval and a high interval, establishing a structural equation model of the causal analysis framework, taking the quality difference measurement as a dependent variable, the construction state as a processing variable and the operation frequency as an adjustment variable, calculating the marginal effect of each sensor data stream when the processing variable is changed through least square estimation, and aggregating the marginal effect average value of all samples to obtain the influence weight of each sensor data stream.
6. The control system of the intelligent trolley for tunnel lining construction according to claim 1, wherein the construction resistance dynamic index is obtained by extracting low-frequency components related to construction resistance from a characteristic representation, performing spectrum analysis on the low-frequency components, calculating distribution proportion of spectrum energy in a preset frequency band, and performing weighted summation on the distribution proportion by combining displacement change rate in the trolley movement track index to generate the construction resistance dynamic index.
7. The control system of the intelligent trolley for tunnel lining construction according to claim 6 is characterized in that the trolley motion trail index is obtained by separating trolley displacement trail data from quality difference measurement, calculating a first derivative and a second derivative of a displacement trail as motion characteristics, performing time warping processing on the motion characteristics, aligning trail segments in different construction stages, calculating covariance matrix characteristic values of the trail segments, and deriving the trolley motion trail index.
8. The control system of the intelligent trolley for tunnel lining construction according to claim 1, wherein the optimization influence weight selection condition comprises the steps of inputting a construction resistance dynamic index and a trolley motion track index into a pre-trained support vector machine model, outputting a threshold value for adjusting the influence weight by a decision function of the support vector machine model, comparing the influence weight of each sensor data stream according to the adjusted threshold value, and reserving a characteristic representation corresponding to the sensor data stream with the influence weight higher than the threshold value to form an optimization characteristic set.
9. The control system of the intelligent trolley for tunnel lining construction according to claim 1, wherein the partitioning of the optimized feature set according to the time sequence alignment relation of the construction waveform comprises the steps of extracting zero crossing points and extreme points from the construction waveform signals, dividing the waveform into a plurality of phase intervals, grouping the optimized feature set according to the time boundaries of the phase intervals, and performing time synchronization correction on feature representations in each group; the method for generating the driving instruction by adopting the attention distribution mechanism comprises the steps of calculating self-attention scores for the characteristic representations of each phase interval, carrying out weighted average on the characteristic representations according to the self-attention scores to obtain interval representations, inputting all the interval representations into a long-short-period memory network for sequence modeling, and mapping the final hidden state of the network into the driving instruction through a full-connection layer.
10. A tunnel lining construction intelligent trolley comprising a memory, a processor and a computer program stored in the memory and running on the processor, characterized in that the processor, when executing the computer program, realizes the functions of a control system of a tunnel lining construction intelligent trolley according to any one of the preceding claims 1 to 9.

Description

Tunnel lining construction intelligent trolley and control system thereof Technical Field The invention relates to the technical field of intelligent control of tunnel construction, in particular to an intelligent trolley for tunnel lining construction and a control system thereof. Background In tunnel lining construction, the prior art mainly relies on single-sensor monitoring or a simple multi-sensor data fusion method based on a fixed threshold value. Current systems typically process various types of sensor signals independently and set static safety thresholds for each parameter. When a certain parameter exceeds a threshold, the system triggers an alarm or executes a preset shutdown instruction. For multi-source data, the common practice is to use empirical weighted average or logic combination based on fixed rules to perform state judgment, and the data fusion model is not adaptive. The operator needs to set the thresholds and rules by means of personal experience, and the system itself cannot autonomously learn and identify key parameters having decisive influence on the construction quality from historical data or real-time operation. The processing mode regards all the sensor data as equal or static importance, and ignores the dynamic change characteristics of the influence of each parameter on the final quality in different operation modes in the construction process. In terms of control instruction generation, the prior art generally employs preprogrammed sequential control logic or transient state-based conditional decisions. The control system selects an execution action from a set of predefined instruction libraries based on whether the sensor reading at the current time satisfies a particular condition. The method cannot analyze dynamic interaction characteristics such as alignment, phase relation and the like of a pressure change curve and a vibration waveform on a time axis. The control strategy is reactive and fragmented and fails to understand a continuous piece of construction process as a complete context. The system can not simulate the operation wisdom of a skilled worker, namely, the control emphasis point is dynamically adjusted according to the staged progress of the process, so that the adaptability of the system to complex working conditions is poor, the generation of control instructions is difficult to achieve accurate and smooth transition, and the further improvement of the automation and intelligent level of lining construction is restricted. A drawback of the prior art is that its analysis framework is static and associative and the control strategy is instantaneous and isolated. This results in a system that has difficulty in accurately screening the true cause of the impact quality and the generated control instructions lack compliance with the dynamic evolution of the overall process of construction. Disclosure of Invention The invention aims to provide an intelligent trolley for tunnel lining construction and a control system thereof, which are used for solving the problems in the background technology. To achieve the above object, the present invention provides a control system of a tunnel lining construction intelligent trolley, the system comprising: The event monitoring module is used for monitoring occurrence signals of trolley operation events and identifying a reference time point of construction state change in a continuously acquired sensor data stream; the data acquisition and processing module is used for intercepting multi-source sensor data in a fixed time interval before and after a reference time point and carrying out transformation processing on the multi-source sensor data so as to derive a characteristic representation of a construction state; The pattern matching module is used for sending the characteristic representation into the pattern matching unit, evaluating the pattern consistency of the standard construction mode and the abnormal construction mode, and generating a quality difference measurement; The causal analysis module is used for using the trolley operating frequency as an adjusting variable, processing the quality difference measurement through a causal analysis framework and estimating the influence weight of each sensor data stream on the construction quality; The feature optimization module is used for integrating the construction resistance dynamic index and the trolley movement track index, optimizing the selection conditions of the influence weights, screening out feature representations corresponding to the high influence weights, and forming an optimized feature set; The instruction generation module is used for partitioning and combining the optimized feature set according to the time sequence alignment relation of the construction waveform, and generating a driving instruction by adopting an attention distribution mechanism. Preferably, the monitoring of the occurrence signal of the trolley operation event comprises continuou