CN-121413125-B - Crane performance detection method and system based on intelligent sensing and data analysis

CN121413125BCN 121413125 BCN121413125 BCN 121413125BCN-121413125-B

Abstract

The invention discloses a crane performance detection method and system based on intelligent sensing and data analysis, which relates to the technical field of crane detection and comprises the following steps, the multi-mode detection and preprocessing is used for synchronously acquiring and cleaning multi-dimensional signals such as strain, acceleration, load, temperature, ambient wind speed and the like, and the optimal detection excitation and sampling configuration is generated under the constraint of energy consumption budget. And (3) carrying out symmetrical group transformation by utilizing the structural symmetrical relation of the crane, reducing redundant channels, maintaining key modal characteristics, and simultaneously combining a physical constraint neural network and a sparse dynamics identification model to obtain structural parameters and health indexes which accord with a physical rule. And a constraint adjustment mechanism of energy consumption and precision is introduced in the detection process, so that stable detection under controlled energy consumption is realized. And further establishing a uniform geometric space through information geometric manifold learning, mapping detection results under different models and working conditions, and generating a resource-aware performance evolution trend index. When meeting the convergence condition, the health index and the trend index are output, so that the detection result has the advantages of accuracy, robustness and popularization.

Inventors

DU YIGUANG
CHEN YURU
YANG YALING
FANG ZHIWEN

Assignees

江西省检验检测认证总院特种设备检验检测研究院

Dates

Publication Date: 20260508
Application Date: 20250911

Claims (8)

1. The crane performance detection method based on intelligent sensing and data analysis is characterized by comprising the following steps of: In the running process of the crane, acquiring multi-mode detection data and preprocessing; based on the preprocessed multi-mode detection data and the energy consumption budget, generating detection excitation signals and sampling parameters with the aim of maximizing observability; Based on the structural symmetry relation of the crane, carrying out symmetrical group transformation on the multi-mode detection data, eliminating the arrangement redundancy of the sensors, and obtaining symmetrical and invariable compression characteristics; Modeling and inverting the compression characteristics based on a physical constraint neural network and a sparse dynamics identification model to obtain a structural parameter vector and a health index; The method comprises the steps of constructing a symmetrical group subset by utilizing crane structure symmetry, merging sensor data into a symmetrical orbit through displacement matrix operation to obtain a symmetrical invariable compression characteristic, introducing a dynamic equation residual error as constraint through a physical constraint neural network to ensure that a modeling result accords with a physical rule, extracting a structural parameter vector from a symmetrical allowed basis function library through sparse dynamics identification, and combining a reference baseline to obtain a health index; Calculating the detection energy consumption in the detection process, and carrying out constraint adjustment between the energy consumption budget and the detection precision to obtain an energy consumption residual error; based on information geometric manifold learning, mapping the structural parameter vector and the health index to a unified geometric space, calculating performance evolution trend indexes of resource perception, and outputting crane performance detection results comprising the health index and the performance evolution trend indexes when convergence conditions are met.
2. The crane performance detection method based on intelligent sensing and data analysis of claim 1, wherein the multi-modal detection data includes strain data, acceleration data, load data, temperature data, and ambient wind speed data; The generation of the detection excitation signal comprises a pseudo-random sequence signal, a short sweep frequency signal and a transient pulse signal, and the sampling parameters comprise sampling frequency, sampling duration and sampling channel priority.
3. The crane performance detection method based on intelligent sensing and data analysis according to claim 2, wherein the symmetrical group transformation of the multi-modal detection data comprises the steps of establishing a symmetrical group subset consistent with a geometric structure based on the relationship between the geometric structure of a crane and the arrangement of sensors, performing displacement operation on the multi-modal detection data by using a displacement matrix corresponding to the symmetrical group subset to obtain a symmetrical track, performing average processing on channel data in the same symmetrical track to obtain a symmetrical invariant statistical feature, extracting low-dimensional components conforming to structural modal symmetry by using projection operation corresponding to the symmetrical group subset, and splicing to form a compression feature.
4. The crane performance detection method based on intelligent sensing and data analysis according to claim 3, wherein modeling and inverting the compression characteristics comprises fitting the compression characteristics by using a physical constraint neural network, introducing a dynamic equation residual error as a constraint condition to ensure that a modeling result accords with a physical rule of a crane, adopting weight sharing and regularization constraint on the characteristics generated by a symmetrical orbit to ensure that modeling output accords with structural symmetry, performing sparse regression in a basis function library only comprising symmetric group allowable terms by using a sparse dynamics identification model to obtain a structural parameter vector, and calculating a health index by combining a reference baseline parameter vector.
5. The crane performance detection method based on intelligent sensing and data analysis according to claim 4, wherein the obtaining of the energy consumption residual error comprises calculating detection energy consumption in a detection process, comparing the detection energy consumption with the energy consumption budget, performing constraint adjustment on a detection excitation signal and sampling parameters according to a comparison result, keeping dynamic balance between the detection energy consumption and detection accuracy, and outputting the energy consumption residual error.
6. The crane performance detection method based on intelligent sensing and data analysis of claim 5, wherein the learning of the information-based geometric manifold comprises establishing a statistical model based on the compression characteristics and constructing a unified geometric space, mapping a structural parameter vector and a health index to the unified geometric space, mapping a reference baseline parameter vector to form a baseline manifold, and calculating a geometric distance between a current detection state and the baseline manifold in the unified geometric space.
7. The crane performance detection method based on intelligent sensing and data analysis of claim 6, wherein the computing resource-aware performance evolution trend index comprises performing normalization processing based on the change of the geometric distance within a set time window and combining the energy consumption budget and the energy consumption residual to obtain the performance evolution trend index, and outputting a crane performance detection result when a preset convergence condition is met.
8. The crane performance detection system based on intelligent sensing and data analysis is based on the crane performance detection method based on intelligent sensing and data analysis according to any one of claims 1 to 7, and is characterized by comprising the following steps: The multi-mode detection and preprocessing module is used for acquiring the multi-mode detection data and executing preprocessing in the crane operation process; The excitation generation and sampling configuration module is used for generating detection excitation signals and sampling parameters with the aim of maximizing observability based on the preprocessed multi-mode detection data and energy consumption budget; the symmetrical group transformation module is used for executing symmetrical group transformation on the multi-mode detection data based on the crane structure symmetrical relation, eliminating sensor arrangement redundancy and obtaining the compression characteristic; The modeling and inversion module is used for modeling and inverting the compression characteristic based on a physical constraint neural network and a sparse dynamics identification model to obtain the structural parameter vector and the health index; the energy consumption constraint adjustment module is used for calculating the detection energy consumption in the detection process and performing constraint adjustment between the energy consumption budget and the detection precision to obtain the energy consumption residual error; And the information geometric manifold learning module is used for mapping the structural parameter vector and the health index to a uniform geometric space, calculating the performance evolution trend index of the resource perception, and outputting the crane performance detection result when the convergence condition is met.

Description

Crane performance detection method and system based on intelligent sensing and data analysis Technical Field The invention relates to the technical field of crane detection, in particular to a crane performance detection method and system based on intelligent sensing and data analysis. Background The crane is used as key equipment in engineering construction, port loading and unloading and industrial production, and the operation safety and the structural performance of the crane are directly related to the operation efficiency and the personnel safety. The existing crane performance detection technology mainly depends on a limited number of strain gauges, accelerometers and load sensors, and monitors stress strain and vibration response of key parts. The method can find out abnormal structure to a certain extent, but has the defects that firstly, the multi-mode detection data is easy to generate redundancy and inconsistency in actual acquisition, the signal distribution difference of different channels is large, and the data management and the subsequent analysis are in high dimension and serious in redundancy. Secondly, most of traditional signal processing methods are based on single sensors or local features, lack of utilization of symmetrical relation of the whole structure of the crane often causes repetition of output results of symmetrically arranged sensors, and storage and calculation overhead is increased. In addition, the crane has a complex operating environment, and the detection process needs to be carried out under limited energy consumption. In the prior art, the application and sampling configuration of detection excitation is usually set based on experience, and dynamic balance between energy consumption and detection accuracy is difficult to achieve. Under the condition of large tonnage cranes or continuous operation, the excessive detection energy consumption can interfere with normal operation, and the too low energy consumption can lead to the deviation of detection results, so that the structural health state cannot be truly reflected. Meanwhile, with parallel application of cranes with multiple models and multiple structural forms, heterogeneous data compatibility becomes a great difficulty gradually. The difference of different cranes in sensor arrangement, structural parameters and working conditions makes the detection result lack of uniform comparison standards, and is difficult to form a common performance reference in mixed fleet or cross-model operation and maintenance. The existing method stays at the single machine analysis level, and lacks a unified evaluation mechanism for cross-model and cross-working conditions. Finally, in terms of performance trend determination, most of traditional detection methods adopt simple thresholds or statistical indexes, and structural mechanical constraint and information geometric measurement cannot be introduced to improve reliability of trend analysis. This results in detection results that are more sensitive to noise and energy consumption fluctuations, and are prone to false positives or false negatives. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the crane performance detection method based on intelligent sensing and data analysis solves the problems that the existing crane performance detection method is insufficient in detection precision, poor in adaptability and limited in popularization in the aspects that multi-mode detection data redundancy is difficult to eliminate, energy consumption and detection precision are difficult to consider, heterogeneous crane data lack of uniform mapping and performance evolution trend indexes lack of resource sensing capability. In order to solve the technical problems, the invention provides the following technical scheme: In a first aspect, the present invention provides a crane performance detection method based on intelligent sensing and data analysis, comprising: In the running process of the crane, acquiring multi-mode detection data and preprocessing; based on the preprocessed multi-mode detection data and the energy consumption budget, generating detection excitation signals and sampling parameters with the aim of maximizing observability; Based on the structural symmetry relation of the crane, carrying out symmetrical group transformation on the multi-mode detection data, eliminating the arrangement redundancy of the sensors, and obtaining symmetrical and invariable compression characteristics; Modeling and inverting the compression characteristics based on a physical constraint neural network and a sparse dynamics identification model to obtain a structural parameter vector and a health index; Calculating the detection energy consumption in the detection process, and carrying out constraint adjustment between the energy consumption budget and the detection precision to obtain an energy consumpt