CN-122020365-A - Intelligent diagnosis and predictive maintenance method for heavy-load three-dimensional library equipment

CN122020365ACN 122020365 ACN122020365 ACN 122020365ACN-122020365-A

Abstract

The invention discloses an intelligent diagnosis and predictive maintenance method for heavy-load three-dimensional library equipment, which comprises the steps of synchronously collecting multi-source sensor data of three-dimensional library key equipment in real time, extracting multi-dimensional characteristics of the multi-source sensor data, constructing multi-dimensional characteristic vectors, carrying out self-adaptive data fusion on the multi-dimensional characteristic vectors through a weight distribution formula based on reliability coefficients of all sensors and real-time signal quality scores, calculating abnormal scores through an abnormal detection scoring function based on the fused characteristic data, comparing the abnormal scores with a threshold value to judge equipment states, and starting an intelligent diagnosis process when the abnormal scores exceed the threshold value, so as to finally generate decision suggestions containing maintenance priorities. The method and the system have the advantages that the accurate sensing, early abnormal early warning and intelligent fault diagnosis of the running state of the heavy-load stereo library equipment can be realized, and the predictive maintenance decision can be generated based on the equipment health decline prediction, so that the equipment reliability is remarkably improved, the service life is prolonged, and the operation and maintenance cost is reduced.

Inventors

WEI XIGUANG
ZHENG MINGHUA
ZHONG ZHIMIN
YAN DONGCHENG
WANG ZHIQING
ZHANG SHULONG
Li Minlei
YANG GANG
ZONG RUIQI
ZHAO YUN
GU WEIZHI

Assignees

科大智能物联技术股份有限公司
达力普石油专用管有限公司

Dates

Publication Date: 20260512
Application Date: 20260108

Claims (10)

1. The intelligent diagnosis and predictive maintenance method for the heavy-duty three-dimensional library equipment is characterized by comprising the following steps of: s1, synchronously acquiring multi-source sensor data of key equipment of a three-dimensional library in real time, wherein the key equipment comprises an ASRS stacker and a goods shelf structure, and the multi-source sensor data at least comprises vibration data, stress data, temperature data and acoustic data; s2, carrying out multidimensional feature extraction on the multi-source sensor data, and constructing multidimensional feature vectors comprising vibration feature vectors, stress feature vectors and temperature feature vectors; S3, performing self-adaptive data fusion on the multidimensional feature vectors through a weight distribution formula based on the reliability coefficient and the real-time signal quality score of each sensor; S4, calculating an abnormal score by using an abnormal detection scoring function based on the fused characteristic data, and comparing the abnormal score with a threshold value to judge the state of the equipment; And S5, when the abnormal score exceeds a threshold value, starting an intelligent diagnosis flow, wherein the intelligent diagnosis flow specifically comprises the steps of carrying out fault mode probability calculation based on a Bayesian theorem to identify fault types, evaluating equipment health score by integrating fault probability and severity degree, and finally generating decision advice containing maintenance priority according to the residual service life of the health degree degradation prediction equipment.
2. The intelligent diagnosis and predictive maintenance method for heavy-duty three-dimensional warehouse equipment according to claim 1, wherein in the step S1, the vibration data is collected by a vibration sensor deployed in a stacker transmission system, the stress data is collected by a stress sensor deployed in a shelf key stress member, the temperature data is collected by temperature sensors deployed in a motor and a bearing, and the acoustic data is collected by an acoustic sensor deployed near the equipment.
3. The intelligent diagnosis and predictive maintenance method for heavy-duty stereo garage equipment according to claim 1, wherein the specific steps in the step S2 include: the multidimensional feature vector is: ; Wherein, each dimension feature vector is: The vibration characteristic vector Including root mean square value, peak value, kurtosis, peak factor and dominant frequency of the vibration signal; The stress characteristic vector Including maximum stress, average stress, stress variance, and stress trend; The temperature characteristic vector The method comprises the steps of current temperature, temperature rise gradient, temperature change trend and maximum temperature difference; 。
4. The method for intelligent diagnosis and predictive maintenance of heavy-duty stereo garage equipment according to claim 1, wherein the specific steps in the step S3 include: the weight distribution formula is as follows: Wherein, the Is the first Fusion weights of the individual sensors; Is a sensor reliability coefficient; Scoring sensor signal quality; The sensor reliability coefficient Pre-calibrating according to historical stability data of a sensor under heavy load, dust and temperature change working conditions, wherein the signal quality of the sensor is scored And carrying out dynamic calculation according to the signal-to-noise ratio, the integrity and the effectiveness of the real-time acquired data.
5. The intelligent diagnosis and predictive maintenance method for heavy-duty stereo garage equipment according to claim 1, wherein the specific steps in the step S4 include: the anomaly detection scoring function is: Wherein, the Is the first The detection sensitivity of the individual sensors is such that, Is the first The dimension value of the characteristic is calculated, And The mean and standard deviation of the feature in the normal state are respectively.
6. The method for intelligent diagnosis and predictive maintenance of heavy-duty stereo library equipment according to claim 1, wherein in the step S5, the calculation of the probability of the failure mode based on bayesian theorem is specifically: Computing is obtaining a multidimensional feature vector The posterior probability of a device failure j is: 。
7. The intelligent diagnosis and predictive maintenance method for heavy-duty stereo garage equipment according to claim 6, wherein the specific steps in step S5 include: the device health score Hscore is calculated by the following formula: Wherein, the For reflecting faults, preset Severity coefficient of the degree of influence.
8. The intelligent diagnosis and predictive maintenance method for heavy-duty stereo garage equipment according to claim 1, wherein the specific steps in the step S5 include: The residual service life RUL is predicted by an exponential model based on the degradation degree of the reference service life and the health degree, and the calculation formula is as follows: Wherein, the For the baseline life of the device under normal operating conditions, As the attenuation coefficient associated with the type of device, Historical degradation amounts scoring device health; the maintenance priority Scoring by integrated device health Remaining service life Estimated cost impact of a shutdown And (3) performing multi-factor weighted calculation, wherein the calculation formula is as follows: Wherein, the Cost impact for a shutdown; , , Is the weight coefficient of each factor.
9. The intelligent diagnosis and predictive maintenance method for heavy-duty stereo garage equipment according to claim 1, further comprising the specific steps of: for the rated heavy load working condition of the key equipment of the three-dimensional library, collecting multi-source sensor data in the normal running state of the equipment to obtain the average value of the normal states of all characteristic dimensions required in the abnormal detection scoring function in a statistics mode And standard deviation 。
10. The intelligent diagnosis and predictive maintenance method for heavy-duty stereo garage equipment according to claim 1, further comprising the steps of generating fault diagnosis results, health degree assessment and residual service life And the maintenance decision suggestion is visually output, and when the maintenance priority exceeds a preset threshold, the maintenance work order is automatically triggered.

Description

Intelligent diagnosis and predictive maintenance method for heavy-load three-dimensional library equipment Technical Field The invention relates to the technical field of industrial Internet of things and intelligent diagnosis, in particular to an intelligent diagnosis and predictive maintenance method for heavy-load three-dimensional library equipment. Background Equipment health monitoring of traditional heavy-duty solid libraries (e.g., heavy-duty steel pipe libraries) mainly relies on a single type of sensor (e.g., vibration sensor) or a simple threshold alarm mechanism. Under severe working conditions such as heavy load, strong impact, multiple dust and the like, the single steel pipe weight can reach several tons, the traditional method faces significant bottlenecks that a single sensor signal cannot comprehensively and accurately reflect the real state of equipment under a composite load, an early fault signal is weak and is extremely easy to be covered by environmental noise and working condition fluctuation, early warning lag or false alarm is caused, meanwhile, the existing system is difficult to locate a fault source due to lack of fusion analysis on multi-source information and deep diagnosis based on the health state of the equipment, predictive maintenance is more difficult to realize, unplanned shutdown is frequent, maintenance cost is high, and great potential safety hazards exist. Disclosure of Invention The invention aims to overcome the defects in the prior art, and adopts an intelligent diagnosis and predictive maintenance method for heavy-load three-dimensional library equipment to solve the problems in the prior art. An intelligent diagnosis and predictive maintenance method for heavy-duty stereo library equipment comprises the following steps: s1, synchronously acquiring multi-source sensor data of key equipment of a three-dimensional library in real time, wherein the key equipment comprises an ASRS stacker and a goods shelf structure, and the multi-source sensor data at least comprises vibration data, stress data, temperature data and acoustic data; s2, carrying out multidimensional feature extraction on the multi-source sensor data, and constructing multidimensional feature vectors comprising vibration feature vectors, stress feature vectors and temperature feature vectors; S3, performing self-adaptive data fusion on the multidimensional feature vectors through a weight distribution formula based on the reliability coefficient and the real-time signal quality score of each sensor; S4, calculating an abnormal score by using an abnormal detection scoring function based on the fused characteristic data, and comparing the abnormal score with a threshold value to judge the state of the equipment; And S5, when the abnormal score exceeds a threshold value, starting an intelligent diagnosis flow, wherein the intelligent diagnosis flow specifically comprises the steps of carrying out fault mode probability calculation based on a Bayesian theorem to identify fault types, evaluating equipment health score by integrating fault probability and severity degree, and finally generating decision advice containing maintenance priority according to the residual service life of the health degree degradation prediction equipment. In the step S1, vibration data are collected by vibration sensors deployed on a stacker transmission system, stress data are collected by stress sensors deployed on key stress members of a shelf, temperature data are collected by temperature sensors deployed on a motor and a bearing, and acoustic data are collected by acoustic sensors deployed near equipment. As a further scheme of the invention, the specific steps in the step S2 comprise: the multidimensional feature vector is: ; Wherein, each dimension feature vector is: The vibration characteristic vector Including root mean square value, peak value, kurtosis, peak factor and dominant frequency of the vibration signal; The stress characteristic vector Including maximum stress, average stress, stress variance, and stress trend; The temperature characteristic vector The method comprises the steps of current temperature, temperature rise gradient, temperature change trend and maximum temperature difference; 。 as a further scheme of the invention, the specific steps in the step S3 comprise: the weight distribution formula is as follows: Wherein, the Is the firstFusion weights of the individual sensors; Is a sensor reliability coefficient; Scoring sensor signal quality; The sensor reliability coefficient Pre-calibrating according to historical stability data of a sensor under heavy load, dust and temperature change working conditions, wherein the signal quality of the sensor is scoredAnd carrying out dynamic calculation according to the signal-to-noise ratio, the integrity and the effectiveness of the real-time acquired data. As a further scheme of the invention, the specific steps in the step S4 comprise: the anomaly detection