CN-121561678-B - Elevator traction machine health assessment method and system based on multi-source data fusion

CN121561678BCN 121561678 BCN121561678 BCN 121561678BCN-121561678-B

Abstract

The invention relates to the technical field of elevator health and discloses an elevator traction machine health assessment method and system based on multi-source data fusion, wherein the method comprises the steps of extracting time domain features and frequency domain features of multi-source sensor data in an elevator traction machine to obtain a feature set of the elevator traction machine; the method comprises the steps of determining entropy values of all features in a feature set, distributing feature weights of the features according to information uncertainty of the features represented by the entropy values, carrying out weighted fusion on the features according to the feature weights to obtain fusion features of the elevator tractor, constructing a health assessment model of the elevator tractor based on historical elevator tractor data, inputting the fusion features into the health assessment model to generate health indexes of the elevator tractor, and carrying out maintenance assessment on the elevator tractor according to the health indexes.

Inventors

MAO XINJIAN
Zhu tongguang
ZHAN YIWEI

Assignees

杭州赛翔科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260116

Claims (7)

1. An elevator traction machine health assessment method based on multi-source data fusion, which is characterized by comprising the following steps: s1, extracting time domain features and frequency domain features of multi-source sensor data in an elevator traction machine to obtain a feature set of the elevator traction machine; s2, determining the entropy value of each feature in the feature set; S3, distributing feature weights of the features according to information uncertainty of the features represented by the entropy values, and carrying out weighted fusion on the features according to the feature weights to obtain fusion features of the elevator traction machine, wherein the method comprises the following steps: distributing the weight coefficient of the characteristic according to the information uncertainty of the entropy characterizing characteristic to obtain an initial weight distribution scheme of the characteristic; performing weight consistency verification on the initial weight distribution scheme to obtain verified feature weights of the features; smoothing the verified feature weights to obtain feature weight distribution of the features; The importance ranking is carried out on the features according to the feature weight distribution, so that feature weights of the features are obtained; the step of distributing the weight coefficient of the feature according to the uncertainty of the information of the entropy characterizing feature to obtain an initial weight distribution scheme of the feature, which comprises the following steps: Dividing the entropy value into a plurality of uncertainty grades and giving corresponding reference weight coefficients; According to the inter-feature correlation constraint condition, carrying out cooperative optimization on the reference weight coefficient to obtain a cooperative weight coefficient of the feature; dynamically adjusting the collaborative weight coefficient based on the accuracy of the historical evaluation result in the elevator traction machine to obtain an initial weight distribution scheme of the characteristics; The dynamic adjustment of the collaborative weight coefficient based on the accuracy of the historical evaluation result in the elevator traction machine is performed to obtain an initial weight distribution scheme of the feature, which comprises the following steps: Collecting health index output results and actual maintenance records in a historical evaluation period to obtain an evaluation accuracy sample set of the characteristics; Performing association analysis on the evaluation accuracy sample set and the corresponding cooperative weight coefficient, and identifying a weight configuration mode causing evaluation deviation; generating weight correction parameters of the weight configuration mode according to the mapping relation between the accuracy and the weight adjustment quantity; when the accuracy of a plurality of continuous evaluation periods reaches a stability threshold, locking the weight configuration of the current weight configuration mode to obtain an initial weight distribution scheme of the feature; S4, constructing a health evaluation model of the elevator traction machine based on historical elevator traction machine data; S5, inputting the fusion characteristics into the health assessment model to generate a health index of the elevator traction machine; and S6, carrying out maintenance and evaluation on the elevator traction machine according to the health index.
2. The elevator traction machine health assessment method based on multi-source data fusion according to claim 1, wherein the extracting the time domain features and the frequency domain features of the multi-source sensor data in the elevator traction machine to obtain the feature set of the elevator traction machine comprises: Performing signal integrity verification on multi-source sensor data of an elevator traction machine to obtain a sensor data sequence of the elevator traction machine; generating a time domain feature subset of the elevator traction machine according to the feature mean value, variance and peak factor counted in the sensor data sequence; Performing frequency domain transformation on the sensor data sequence to obtain a frequency domain feature subset of the elevator traction machine; and combining the time domain feature subset with the frequency domain feature subset to generate a feature set of the elevator traction machine.
3. The elevator traction machine health assessment method based on multi-source data fusion according to claim 1, wherein said determining the entropy value of each feature in the feature set comprises: performing sequence segmentation on the feature set to obtain a feature data subsequence of the feature set; Performing state division on the characteristic data subsequence to obtain a state sequence corresponding to the characteristic data subsequence; And calculating the approximate entropy of each feature in the feature set according to the state transition probability of the state sequence to obtain the entropy value of the elevator traction machine.
4. The elevator traction machine health assessment method based on multi-source data fusion according to claim 1, wherein the building of the elevator traction machine health assessment model based on historical elevator traction machine data comprises: Performing data cleaning and working condition division on historical elevator traction machine data to obtain a training sample set of the elevator traction machine; Performing multi-scale feature extraction on the training sample set to obtain a feature expression space of the elevator traction machine; the importance ranking is carried out on the features in the feature expression space, so that a feature subset of the elevator traction machine is obtained; Performing layer-by-layer feature abstraction on the feature subsets to obtain a health state mapping relation of the elevator traction machine; Constructing a layered evaluation network of the elevator traction machine according to the health state mapping relation; And training and optimizing the layered evaluation network, and the health evaluation model of the elevator traction machine.
5. The elevator traction machine health assessment method based on multi-source data fusion according to claim 4, wherein the training optimization is performed on the hierarchical assessment network, and the elevator traction machine health assessment model comprises: Performing countermeasure training on the layered evaluation network to obtain an initial training model of the layered evaluation network; identifying key feature layers in the initial training model, and adjusting a network connection structure of the initial training model to obtain a structure optimization model of the hierarchical evaluation network; According to the change trend of the training loss curve of the countermeasure training, the parameter updating step length of the structural optimization model is adaptively adjusted, and the parameter optimization model of the layered evaluation network is obtained; And migrating the evaluation capability of the expert models to the parameter optimization model to obtain the health evaluation model of the elevator traction machine.
6. The elevator traction machine health assessment method based on multi-source data fusion according to claim 1, wherein the maintenance assessment of the elevator traction machine according to the health index comprises: establishing a mapping rule base of historical health indexes and maintenance strategies; generating a preliminary maintenance suggestion of the elevator traction machine according to the numerical interval of the health index; Carrying out working condition suitability correction on the preliminary maintenance advice based on the operation environment parameters of the elevator traction machine to obtain an environment adaptation maintenance scheme of the elevator traction machine; based on the environment adaptation maintenance scheme, the execution priority and the resource allocation requirement of the maintenance procedure are defined; And maintaining the elevator traction machine according to the execution priority and the resource allocation requirement.
7. An elevator traction machine health evaluation system based on multi-source data fusion for realizing the elevator traction machine health evaluation method based on multi-source data fusion according to claim 1, the system comprising: The multi-source feature extraction module is used for extracting time domain features and frequency domain features of multi-source sensor data in the elevator traction machine to obtain a feature set of the elevator traction machine; The characteristic entropy value determining module is used for determining the entropy value of each characteristic in the characteristic set; the entropy weight fusion feature construction module is used for distributing feature weights of the features according to information uncertainty of the features represented by the entropy values, carrying out weighted fusion on the features according to the feature weights to obtain fusion features of the elevator traction machine, and is particularly used for: distributing the weight coefficient of the characteristic according to the information uncertainty of the entropy characterizing characteristic to obtain an initial weight distribution scheme of the characteristic; performing weight consistency verification on the initial weight distribution scheme to obtain verified feature weights of the features; smoothing the verified feature weights to obtain feature weight distribution of the features; The importance ranking is carried out on the features according to the feature weight distribution, so that feature weights of the features are obtained; the step of distributing the weight coefficient of the feature according to the uncertainty of the information of the entropy characterizing feature to obtain an initial weight distribution scheme of the feature, which comprises the following steps: Dividing the entropy value into a plurality of uncertainty grades and giving corresponding reference weight coefficients; According to the inter-feature correlation constraint condition, carrying out cooperative optimization on the reference weight coefficient to obtain a cooperative weight coefficient of the feature; dynamically adjusting the collaborative weight coefficient based on the accuracy of the historical evaluation result in the elevator traction machine to obtain an initial weight distribution scheme of the characteristics; The dynamic adjustment of the collaborative weight coefficient based on the accuracy of the historical evaluation result in the elevator traction machine is performed to obtain an initial weight distribution scheme of the feature, which comprises the following steps: Collecting health index output results and actual maintenance records in a historical evaluation period to obtain an evaluation accuracy sample set of the characteristics; Performing association analysis on the evaluation accuracy sample set and the corresponding cooperative weight coefficient, and identifying a weight configuration mode causing evaluation deviation; generating weight correction parameters of the weight configuration mode according to the mapping relation between the accuracy and the weight adjustment quantity; when the accuracy of a plurality of continuous evaluation periods reaches a stability threshold, locking the weight configuration of the current weight configuration mode to obtain an initial weight distribution scheme of the feature; The health evaluation model construction module is used for constructing a health evaluation model of the elevator traction machine based on historical elevator traction machine data; The health index generation module is used for inputting the fusion characteristics into the health evaluation model to generate the health index of the elevator traction machine; And the maintenance evaluation decision module is used for carrying out maintenance evaluation on the elevator traction machine according to the health index.

Description

Elevator traction machine health assessment method and system based on multi-source data fusion Technical Field The invention relates to the technical field of elevator health, in particular to an elevator traction machine health assessment method and system based on multi-source data fusion. Background In the prior art, a complete signal integrity verification process is lacking in processing of elevator traction machine multisource sensor data, residual noise and missing key operation information are easy to cause in the data, time domain features or frequency domain features are often extracted independently in a feature extraction link, the two features cannot be organically combined, a generated feature set cannot fully cover key dimensions of the traction machine operation state, so that accurate data support is lost in subsequent health assessment, potential fine anomalies of equipment are difficult to effectively identify, and reliability of an assessment result is affected. In the prior art, when feature weights are determined, feature information uncertainty reflected by an unbound entropy value is scientifically distributed, the weights are set by relying on manual experience, so that the effect of important features is weakened, the influence of secondary features is amplified, meanwhile, when a health evaluation model is built, the historical elevator traction machine data is not thoroughly cleaned, the working condition is not finely divided, the model training sample quality is poor, the mapping relation between the health state and the features cannot be accurately built, the output health index and the actual health state of equipment have larger deviation, the maintenance evaluation is lack of pertinence, and high-efficiency equipment maintenance is difficult to guide, so that the problem of how to improve the efficiency of elevator traction machine health evaluation is urgently solved. Disclosure of Invention The invention provides an elevator traction machine health assessment method and system based on multi-source data fusion, which are used for solving the problems in the background technology. In order to achieve the above purpose, the invention provides an elevator traction machine health assessment method based on multi-source data fusion, which comprises the following steps: s1, extracting time domain features and frequency domain features of multi-source sensor data in an elevator traction machine to obtain a feature set of the elevator traction machine; s2, determining the entropy value of each feature in the feature set; S3, distributing feature weights of the features according to information uncertainty of the features represented by the entropy values, and carrying out weighted fusion on the features according to the feature weights to obtain fusion features of the elevator traction machine; S4, constructing a health evaluation model of the elevator traction machine based on historical elevator traction machine data; S5, inputting the fusion characteristics into the health assessment model to generate a health index of the elevator traction machine; and S6, carrying out maintenance and evaluation on the elevator traction machine according to the health index. In a preferred embodiment, the extracting the time domain feature and the frequency domain feature of the multi-source sensor data in the elevator traction machine to obtain the feature set of the elevator traction machine includes: Performing signal integrity verification on multi-source sensor data of an elevator traction machine to obtain a sensor data sequence of the elevator traction machine; generating a time domain feature subset of the elevator traction machine according to the feature mean value, variance and peak factor counted in the sensor data sequence; Performing frequency domain transformation on the sensor data sequence to obtain a frequency domain feature subset of the elevator traction machine; and combining the time domain feature subset with the frequency domain feature subset to generate a feature set of the elevator traction machine. In a preferred embodiment, said determining the entropy value of each feature in said set of features comprises: performing sequence segmentation on the feature set to obtain a feature data subsequence of the feature set; Performing state division on the characteristic data subsequence to obtain a state sequence corresponding to the characteristic data subsequence; And calculating the approximate entropy of each feature in the feature set according to the state transition probability of the state sequence to obtain the entropy value of the elevator traction machine. In a preferred embodiment, the assigning feature weights of the features according to the information uncertainties of the features characterized by the entropy values comprises: distributing the weight coefficient of the characteristic according to the information uncertainty of the entropy characterizing