PHM-based hoisting equipment state testing and verification method

Abstract

The present application relates to hoisting equipment detection technical field, disclose a kind of hoisting equipment state test and verification method based on PHM.The method includes extracting design parameters from hoisting equipment technical document, high-frequency operating characteristics are synchronously collected when equipment starts operating, whether to start multi-dimensional state evaluation process is judged by fusing two kinds of data.After starting, first calculate equipment design robustness index and detect structural vibration condition, combine both to complete modular segmentation and extract multiple evaluation units.Continuous monitoring vibration deviation and stress diffusion rate of each unit, evaluate its contribution to state deviation to obtain contribution factor, while tracking load duration interval.Based on load duration interval and contribution factor, unit rating is completed, interface stress and tolerance limit load are collected to derive rating weight, finally integrate each unit rating and weight, generate health index of hoisting equipment under current working condition, realize comprehensive and accurate evaluation of equipment state.

Description

Technical Field

[0001] This invention relates to the field of lifting equipment testing technology, specifically a lifting equipment condition testing and verification method based on PHM (Prognostics and Health Management). Background Technology

[0002] As key equipment in industrial production, logistics, and construction, the stability of lifting machinery's operational status directly affects production efficiency and operational safety. With the increasing level of industrial automation, the structure of lifting machinery is becoming increasingly complex, and its operating conditions are becoming more diverse and demanding, placing higher requirements on its condition monitoring and fault early warning. Currently, most lifting machinery condition testing methods rely on single sensor data acquisition, focusing only on local parameter changes during equipment operation, failing to fully consider the machinery's original design intent and inherent properties, resulting in insufficient comprehensiveness in condition assessment.

[0003] In practical applications, existing methods often overlook the potential impact of design parameters on equipment operating status, relying solely on real-time operational data for judgment, which can easily lead to assessment bias. For example, some methods only use vibration sensor data to determine if there are structural anomalies, without considering robustness indicators from the equipment design phase, making it difficult to effectively identify latent faults caused by design defects. Furthermore, traditional assessment methods lack scientific modular segmentation strategies, treating the lifting equipment as a whole for status assessment, failing to accurately pinpoint the specific unit where the fault source is located. When slight deviations in equipment status occur, it is difficult to distinguish the degree of impact of each component, resulting in low fault diagnosis efficiency.

[0004] Current health assessment systems often employ single-dimensional rating standards, lacking a weighting mechanism that matches load characteristics and stress states. This results in a lack of specificity and scientific rigor in the final health status assessment. In high-intensity operational scenarios, this broad-based assessment method fails to accurately reflect the true operating status of the equipment, potentially leading to over-maintenance or delayed maintenance. This not only increases operating costs but also introduces safety hazards. Therefore, a method for testing and verifying the condition of lifting equipment that integrates design parameters and operational characteristics to achieve refined assessment is needed to address the numerous shortcomings of existing technologies. Summary of the Invention

[0005] The purpose of this invention is to provide a PHM-based method for testing and verifying the condition of lifting equipment, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a PHM-based method for testing and verifying the condition of lifting equipment, the method comprising:

[0007] Design parameters are extracted from the lifting equipment's technical documents. High-frequency operating characteristics are collected during the lifting equipment's startup operation. The design parameters and high-frequency operating characteristics are integrated to determine whether to initiate a multi-dimensional condition assessment process. If the multi-dimensional condition assessment process is initiated, the lifting equipment's design robustness index is calculated, the structural vibration status of the lifting equipment is detected, and the lifting equipment is modularly segmented and multiple assessment units are extracted based on the structural vibration status and design robustness index. The vibration deviation and stress diffusion rate of each assessment unit are continuously monitored, the contribution of each assessment unit to the condition deviation is evaluated and the contribution factor is calculated, the load duration interval of each assessment unit is tracked, and low-contribution units are selected as identified assessment units. The unit rating is performed by associating the load duration interval of each identified assessment unit with the corresponding unit's contribution factor. The interface stress and tolerance limit load of each identified assessment unit are collected, the rating weight of each identified assessment unit is derived, and the rating and weight of each identified assessment unit are integrated to obtain the lifting equipment's health index under the current operating conditions.

[0008] Preferably, the design parameters include structural material durability limits. By reading the material density, elastic modulus, and material thickness, a stress superposition model is constructed to determine the surface stress of the structure and aggregate it with the baseline stress to obtain the structural material durability limits. Using a standard operating test sequence and operating frequency setting, the crane under test is made to perform cyclic load operation within a specified period, and the total number of operations is counted to generate high-frequency operating characteristics. The structural material durability limits and high-frequency operating characteristics are compared with preset thresholds in turn to generate a conclusion on the rationality of the current crane design. If the conclusions all show that the current crane design is invalid, the health index of the crane under the current state is directly determined to be low. Otherwise, a multi-dimensional state assessment process is performed on the crane under the current state.

[0009] Preferably, the durability limits and high-frequency operating characteristics of the structural materials are normalized, and a weighted average algorithm is applied to output the design robustness index. With the help of miniature accelerometers arranged on the structural channel of the lifting equipment, the vibration amplitude and frequency variation are measured to detect the vibration status of the lifting equipment structure. An effective vibration assessment model based on vibration propagation theory is used to output the vibration intensity per unit time.

[0010] Preferably, the design robustness index and vibration intensity per unit time are standardized and converted, and then input into the decision tree model to generate unit screening index values; the lifting equipment is modularly segmented to obtain each unit of the lifting equipment, and a comprehensive calculation is performed based on the unit vibration demand metric and the unit historical failure probability; according to the material stress properties of the unit, the expected stress load of the unit per unit time is calculated, and the minimum attenuation energy required is estimated with reference to the structural damping model to obtain the unit vibration demand metric; by searching the historical records of the lifting equipment, failure instances of each structural unit are extracted, and they are categorized and summarized by unit. The cumulative number of failures of the unit within the reference period is calculated as a ratio with the number of running segments to obtain the unit historical failure probability.

[0011] Preferably, the unit vibration demand measurement and the unit historical failure probability are normalized and fed into the comprehensive risk analysis model to determine the unit activity level of each unit of the lifting equipment. The unit activity levels are arranged in ascending order of value. Through multiple predefined classification thresholds, the unit screening index value is compared with the multiple classification thresholds. Evaluation unit groups whose classification thresholds do not exceed the current index value are selected. Combined with the activity ranking results, all evaluation units whose classification thresholds are located before the current index value in the ranking are selected simultaneously.

[0012] Preferably, an embedded accelerometer network and a thermal imager or strain gauge array are used to continuously measure the vibration deviation and stress diffusion rate of each evaluation unit. By recording the vibration change sequence per unit time, setting the sampling interval, continuous vibration readings are obtained, and the vibration deviation is generated based on the differential and wave analysis model. By monitoring the stress change of each evaluation unit per unit time, setting the observation period, continuous stress data are obtained. The stress at the end of the stress observation is subtracted from the stress at the beginning of the stress observation, and the quotient is calculated with the corresponding observation duration to obtain the stress diffusion rate. The vibration deviation and stress diffusion rate are standardized and input into a linear correlation model to generate the contribution factor of the state deviation.

[0013] Preferably, the contribution factor of the state deviation in each evaluation unit is compared with a predefined influence threshold. If the contribution factor reaches or exceeds the influence threshold, the contribution of the state deviation is considered to be high, and the corresponding evaluation unit is marked as low-level and excluded. Otherwise, if the contribution factor is lower than the influence threshold, the contribution of the state deviation is considered to be low, and the corresponding evaluation unit is retained and marked. When the lifting equipment starts operation, the load duration interval of the marked evaluation units is monitored, the start and end times of the load of each marked evaluation unit are captured at the moment the lifting equipment is triggered, and the load duration interval is obtained based on the time difference calculation.

[0014] Preferably, the contribution factors of the corresponding units are normalized by associating the load duration interval of each identification evaluation unit with the corresponding unit, and then input into the S-curve fusion model to output the rating of each identification evaluation unit.

[0015] Preferably, by applying a constant low load to each identification assessment unit and measuring its deformation, the interfacial stress of each identification assessment unit is calculated based on elasticity theory; combined with the structural material properties, a fatigue response model is established to set the maximum allowable deformation, and the ultimate load that each identification assessment unit can withstand is calculated; the interfacial stress and ultimate load that each identification assessment unit can withstand are standardized and input into the entropy weight allocation model to obtain the rating weight of each identification assessment unit.

[0016] Preferably, the health index of the lifting equipment under the current working conditions is output by combining the ratings and rating weights of each identification evaluation unit through linear combination calculation.

[0017] Compared with the prior art, the beneficial effects of the present invention are:

[0018] By integrating the design parameters and high-frequency operating characteristics of lifting equipment to determine whether to initiate a multi-dimensional condition assessment process, this approach breaks away from the limitations of traditional assessment methods that rely solely on single operational data. This makes the initiation of the assessment process more targeted and rational. Design parameters, as inherent attributes of lifting equipment, directly determine the theoretical operating limits and structural load-bearing capacity of the equipment, while high-frequency operating characteristics can reflect the actual working status of the equipment in real time. The organic combination of the two allows for a comprehensive consideration of the equipment's design potential and actual operational performance, avoiding the one-sidedness of assessments caused by focusing on only a single dimension. This ensures that the initiation judgment of the assessment process better reflects the actual condition requirements of the equipment.

[0019] In the multidimensional condition assessment process, modular segmentation and extraction of assessment units are achieved by combining design robustness indicators and structural vibration conditions, thus realizing the scientific and accurate division of assessment units. Design robustness indicators reflect the equipment's ability to resist various disturbances during the design phase, while structural vibration conditions reflect the actual structural response of the equipment during operation. Modular segmentation based on these two key factors can divide the equipment into units with clear assessment significance according to functional attributes and structural characteristics, avoiding the ambiguity caused by traditional overall assessment or arbitrary segmentation, and laying the foundation for subsequent accurate assessment.

[0020] By continuously monitoring and evaluating the vibration deviation and stress diffusion rate of the evaluation units, calculating contribution factors, and tracking load duration intervals, the impact of each unit on the overall equipment condition deviation can be accurately located. Vibration deviation and stress diffusion rate are core parameters reflecting the structural integrity of the equipment. Real-time monitoring can promptly capture abnormal changes in units, while the calculation of contribution factors quantifies the influence weight of each unit, and tracking load duration intervals can determine the wear status of units in conjunction with operational intensity. This multi-parameter collaborative monitoring and quantitative analysis method makes fault tracing more directional, facilitating staff to quickly identify key problem units.

[0021] Element rating is performed by associating the load duration interval with the contribution factor, and the rating weights are derived by combining interface stress and ultimate load tolerance, making the element rating results more objective and targeted. The load duration interval is directly related to the fatigue wear degree of the element, and the contribution factor reflects the element's impact on the overall state. The correlation between the two can comprehensively reflect the actual operating state of the element. Interface stress and ultimate load tolerance are key performance parameters of the element. Using these as the basis to derive the rating weights can match the weight allocation with the structural characteristics and load-bearing capacity of the element, avoiding the rating distortion problem caused by uniform weights.

[0022] By integrating the ratings and weights of various assessment units to obtain a health index, a comprehensive and accurate representation of the overall status of the lifting equipment under its current operating conditions is achieved. This health index is not a simple summation of single parameters, but a comprehensive evaluation result integrating multi-dimensional information such as design attributes, operational characteristics, structural status, and load characteristics. It can truly reflect the operating status and health level of the equipment, providing a scientific reference for equipment maintenance and work scheduling, helping to optimize equipment management strategies, extend equipment lifespan, and ensure operational safety and efficiency. Furthermore, the entire methodology is logically rigorous, highly operable, and applicable to different types of lifting equipment under various operating conditions, possessing broad application scenarios and practical value. Attached Figure Description

[0023] Figure 1 This is a schematic diagram illustrating the working principle of the PHM-based lifting equipment condition testing and verification method described in this invention.

[0024] Figure 2 To design a flowchart for robustness index and vibration condition assessment;

[0025] Figure 3 A flowchart for unit activity and evaluation unit selection;

[0026] Figure 4 A comparison chart of vibration deviation and stress diffusion velocity monitoring for each evaluation unit of the lifting equipment;

[0027] Figure 5This is a graph showing the trend of the health index of lifting equipment over operating time. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Please see Figure 1 This invention provides a PHM-based method for testing and verifying the condition of lifting equipment. The method includes: systematically extracting design parameters from the lifting equipment's technical documentation and collecting high-frequency operating characteristics in real time during the lifting equipment's startup operation; processing the design parameters and high-frequency operating characteristics through a data fusion module to logically determine whether to initiate a multi-dimensional condition assessment process; if the fusion result triggers the startup condition, calculating the lifting equipment's design robustness index and simultaneously detecting the lifting equipment's structural vibration status; using the structural vibration status and design robustness index as inputs, modularly segmenting the lifting equipment and extracting multiple sets of assessment units; continuously monitoring the multi-dimensional condition data of the assessment units, including vibration deviation and stress diffusion rate, and evaluating the contribution of each assessment unit to the condition deviation based on these data to calculate the contribution factor, while tracking the load duration interval of each assessment unit and selecting low-contribution units as identified assessment units; correlating the load duration interval with the contribution factor of the corresponding unit to perform unit rating, and collecting the interface stress and tolerance limit load of each identified assessment unit to derive the rating weight; finally, by integrating the ratings and weights of each identified assessment unit, outputting the lifting equipment's health index under the current operating condition.

[0030] Example 1: See Figure 2 , Figure 2This document outlines a flowchart for assessing robustness indicators and vibration conditions. Design parameters in the lifting equipment's technical documentation include structural material durability limits. These limits are obtained by reading material density, elastic modulus, and thickness values, which are derived from the lifting equipment's design specifications or material certification. Constructing a stress superposition model requires integrating principles of materials mechanics. This model aggregates and calculates multi-source stress data, including static load stress and dynamic fluctuation stress. It determines the stress distribution on the structural surface and compares it with baseline stress, derived from historical standard data or rated values ​​in design specifications. The final result is a quantitative output of the structural material durability limits. The generation of high-frequency operating characteristics employs a standard operating test sequence and operating frequency setting. The standard operating test sequence consists of a predefined load cycle pattern, which simulates the lifting, translating, and lowering actions of the lifting equipment in actual operation. The operating frequency setting is adjusted according to the type of lifting equipment; bridge cranes and gantry cranes use different frequency parameters. The crane under test is instructed to perform cyclic load operation within a specified time period. The length of the specified time period is set according to the test accuracy requirements. During the cyclic load operation, the encoder and torque sensor installed on the crane's drive component record the number of operations and real-time parameters. The total number of operations is counted, and a high-frequency operating characteristic dataset is generated. The high-frequency operating characteristic dataset includes peak load, operating cycle, and acceleration waveform.

[0031] The structural material durability limits and high-frequency operating characteristics are then input into a comparison module, which uses digital logic circuits or software algorithms to compare them with preset thresholds. These preset thresholds are set based on industry standards or experimental data, including GB / T3811-2008 Crane Design Specification. The preset thresholds are divided into upper and lower thresholds. The comparison results generate a conclusion on the rationality of the current crane design. This conclusion is a binary logic output. If either the structural material durability limit or the high-frequency operating characteristics exceed the preset threshold, the conclusion indicates that the current crane design is invalid. The crane's health index is then directly determined to be low, triggering an alarm signal and halting subsequent processes. Conversely, if the conclusion indicates that the design is valid, a multi-dimensional state assessment process is performed on the crane in its current state, initiating full data acquisition from the sensor network. The robustness index is calculated through normalization. The durability limits of structural materials and high-frequency operating characteristics are converted into a unified dimension. The conversion method adopts the minimum-maximum scaling algorithm to linearly map the original data to the [0,1] interval. The weighted average algorithm is applied, and the weighting coefficients are allocated according to the importance of the parameters. The weighting coefficients are determined by expert scoring or analytic hierarchy process. The scalar value of the robustness index is output. The higher the value of the robustness index, the more robust the crane design. The vibration status of the lifting machinery structure is detected using a network of miniature accelerometers deployed along the structural channels. These miniature accelerometers, manufactured using MEMS technology, are positioned to cover key nodes of the main beam, outriggers, and hoisting mechanism of the lifting machinery. The miniature accelerometers measure vibration amplitude and frequency variation data. Vibration amplitude reflects instantaneous vibration intensity, while frequency variation reveals changes in vibration modes. The measurement data is transmitted to a signal processor via a CAN bus. The measurement data is then input into an effective vibration assessment model based on vibration propagation theory. This model integrates wave equations and damping factors to calculate the vibration intensity per unit time. This vibration intensity per unit time is used as a quantitative indicator for subsequent decision-making; the unit of vibration intensity per unit time is millimeters per second².

[0032] In the design parameter extraction phase, automated software tools are used to parse technical documents. These tools integrate OCR technology and natural language processing algorithms to read text and numerical information from CAD drawings or specification sheets. Key attributes such as material density, elastic modulus, and material thickness are extracted. The unit for material density is kg / m³, the unit for elastic modulus is Pascal, and the unit for material thickness is millimeters. When constructing the stress superposition model, the load superposition effect is considered, which includes the interaction between static and dynamic loads. The model outputs a stress distribution map, which is visualized as a cloud map. The aggregation with the baseline stress is completed through weighted averaging. The baseline stress is derived from standard test data, which is stored in a relational database. The weights are dynamically adjusted according to the stress type. During high-frequency operation characteristic acquisition, a standard operation test sequence simulates actual working conditions. The number of cycles in the standard operation test sequence is set to over 1000, and the operating frequency is set to ensure coverage of typical operating ranges, including no-load, half-load, and full-load states. The specified time period for cyclic load operation is adjusted according to the size of the equipment; for large lifting equipment, the test period can be as long as 72 hours. The total number of operations is counted, including start and stop events, and the timestamps of start and stop events are recorded by the PLC controller. The preset threshold comparison module uses a logic comparator integrated into the FPGA chip. The threshold value is dynamically updated to adapt to environmental changes, including temperature fluctuations and humidity effects. After the design rationality conclusion is generated, if it is invalid, the health index is directly set to a low level, corresponding to a value below 0.3, to avoid unnecessary consumption of computing resources. Normalization is achieved using a minimum-maximum scaling method, with the formula (x-min) / (max-min), mapping structural material durability limits and high-frequency operating characteristics to a 0-1 range. The weights of the weighted average algorithm are set based on expert knowledge derived from the experience database of domain engineers. Robustness indicators are designed as comprehensive feedback indicators for system status, retaining four decimal places of precision. The arrangement of miniature accelerometers follows structural dynamics principles, requiring sensors to be installed at the peak points of modal vibrations, with the sensor network covering key nodes, determined through finite element analysis. Vibration amplitude and frequency variation data are denoised using Butterworth low-pass filters with a cutoff frequency of 100Hz. The effective vibration assessment model integrates wave equations, which are discretized, and the vibration intensity per unit time is calculated as the average energy value, taken as the root mean square value within the moving window. The entire process ensures data accuracy and real-time performance. Data accuracy is maintained through sensor calibration, while real-time performance is achieved through 5G transmission protocols, providing reliable input for multi-dimensional state assessment.

[0033] The construction of the stress superposition model involves the mathematical expression of the material constitutive relation, which adopts the generalized form of Hooke's law. Input parameters include Poisson's ratio and yield strength. After the stress distribution map is generated, matrix addition is performed when it is aggregated with the baseline stress. The dimensions of the baseline stress matrix are consistent with the measured data. The cyclic load operation of the high-frequency running feature acquisition is controlled by a test bench equipped with a hydraulic servo system, with load accuracy controlled within ±1%. The running frequency setting is written to the PLC register, the real-time parameter sampling rate is set to 1kHz, and the high-frequency running feature dataset is stored in HDF5 format. The preset threshold update mechanism of the comparison module adopts a sliding window algorithm with a sliding window length of 30 days. Binary logic output drives the status indicator light. The weighting coefficients in the robustness index calculation are updated quarterly, and the weighting coefficient table is stored on a cloud server. Data acquisition synchronization of the miniature accelerometer uses the PTP clock protocol, achieving microsecond-level time alignment accuracy for vibration data. The effective vibration assessment model outputs the vibration intensity value per unit time per second. The technical document parsing tool supports multiple languages. CAD drawing parsing uses the DXF parsing library, and material properties are extracted and written to XML files. The stress superposition model runs on edge computing devices, and the baseline stress database is synchronized every 24 hours. Specified time periods for cyclic load testing can be remotely configured, and test progress is displayed in real-time on a monitoring screen. A report is automatically generated when the total number of runs exceeds a threshold. The preset threshold comparison module integrates a self-learning function; threshold values ​​are automatically corrected based on historical deviations, and a maintenance work order is triggered when the health index is low. The parameter boundaries for normalization processing are adjusted weekly, weighting coefficients are optimized using the AHP algorithm, and robust indicators are designed to participate in the health index fusion calculation.

[0034] Example 2: See Figure 3 , Figure 3This is a flowchart illustrating the unit activity and evaluation unit selection process. The modular segmentation and evaluation unit extraction process for lifting equipment involves designing robustness indices and vibration intensity per unit time for standardization transformation. The standardization transformation uses the Z-score method to convert the original data into a distribution with a mean of zero and a variance of one. The mean and standard deviation in the Z-score transformation formula are calculated from historical datasets containing samples of design robustness indices and vibration intensity per unit time under different operating conditions. The transformed data is input into a decision tree model, which is built based on the CART algorithm. The training data comes from the historical operation records of the lifting equipment. The maximum depth of the decision tree model is set to 10 layers, and the minimum number of leaf node samples is set to 5. The decision tree model generates unit selection index values, which are continuous numerical values ​​ranging from 0 to 1. These unit selection index values ​​serve as the threshold for modular segmentation. The lifting equipment is modularly divided based on the functional independence of the mechanical structure, resulting in various units of the lifting equipment. Each unit includes the main beam structure, outrigger system, hoisting mechanism, traveling device, and electrical control system. The unit vibration demand measurement and the unit historical failure probability are comprehensively calculated. The unit vibration demand measurement is based on the material stress properties, including yield strength and fatigue limit. The expected stress load per unit time is calculated, and the expected stress load is derived from load spectrum statistical analysis.

[0035] The unit vibration demand measurement is calculated by referencing the structural damping model to estimate the minimum attenuation energy required. The structural damping model adopts viscous damping theory, and the damping coefficient is obtained through modal testing. The unit historical failure probability is obtained by retrieving the historical records of the lifting equipment, which are stored in a time-series database. Fault instances of each structural unit are extracted, including crack formation, bearing damage, and electrical faults. Fault data are summarized by unit category, and the ratio of the cumulative number of failures of the unit within a reference time period to the number of operating segments is calculated. The number of operating segments is based on an eight-hour period to obtain the unit historical failure probability value, which is retained to four decimal places. The unit vibration demand measurement and unit historical failure probability are normalized. The normalization uses a maximum-minimum scaling method to map the data to the [0,1] interval and feeds it into the comprehensive risk analysis model. The comprehensive risk analysis model integrates fuzzy logic and grey system theory to determine the unit activity level of each unit of the lifting equipment. The larger the unit activity level value, the more frequently the unit needs to be monitored. Unit activity is sorted in ascending order of value, and the sorting results generate an ordered list. Multiple predefined classification thresholds are used, which are set based on percentiles and are divided into high-risk, medium-risk, and low-risk thresholds. The unit screening index value is compared with the multiple classification thresholds. The comparison operation uses a binary search algorithm to select evaluation unit groups whose classification thresholds do not exceed the current index value. At the same time, combined with the activity sorting results, all evaluation units whose classification thresholds are located before the current index value in the sorting are selected simultaneously. The final set of evaluation units is used for subsequent status monitoring.

[0036] The comprehensive risk analysis model is a multi-dimensional risk assessment model built on the PHM (Prognostics and Health Management) technical framework. Its core is to integrate the status data and full lifecycle information of each identification assessment unit of the lifting equipment to conduct risk assessment. The model first identifies potential risk sources in mechanical structures, electrical systems, and other aspects through methods such as Failure Mode and Effects Analysis (FMEA), covering key risk factors such as component fatigue damage, material aging, and controller failure. Then, it combines real-time monitoring data such as vibration deviation, stress diffusion rate, and load duration intervals with historical data from maintenance records to comprehensively analyze the probability and impact of each risk factor.

[0037] During the standardization process, robustness indicators and vibration intensity per unit time are first used for outlier detection. Outlier detection employs the 3σ principle, removing data points that deviate from the mean by more than three standard deviations. Z-score transformation ensures comparability of data across different dimensions, and the transformed data is stored in a new data table. The decision tree model training process uses 10-fold cross-validation. Feature importance ranking shows that the influence weight of vibration intensity per unit time on the unit screening index value is 0.6, and the influence weight of the robustness indicator is 0.4. After the unit screening index value is output, it is compared with a preset benchmark threshold of 0.7; units exceeding the benchmark threshold are placed in a priority processing queue. Modular segmentation is implemented based on the crane assembly drawings. The geographical coordinates of each unit are obtained through 3D modeling software, and the connection relationships between units are represented by an adjacency matrix. When calculating unit vibration demand, material stress properties are retrieved from a material library, the expected stress load considers the dynamic load coefficient, and the structural damping model solution requires iterative calculation of the minimum attenuation energy. Historical record retrieval uses a time range query, fault instances are graded by severity, unit classifications are summarized to generate a heatmap for visualization, and ratio calculation results are stored in a probability distribution table. Normalization is performed using a linear transformation formula. The comprehensive risk analysis model includes two dimensions: risk probability and risk impact. Unit activity calculation results are updated in real-time to the status monitoring dashboard. Unit activity ranking uses a quicksort algorithm, and the ranked list is stored in a priority queue. Classification threshold values ​​are dynamically adjusted based on quarterly assessments. A bitmap index is generated by comparing the unit screening index value with multiple classification thresholds. The final determination of the evaluation unit set requires consistency verification. Each node in the decision tree model includes feature selection rules, and the information gain ratio is used to determine split points. The calculation accuracy of the unit screening index value reaches 0.001. Boundary conditions for modular segmentation consider stress concentration phenomena, and the monitoring point layout scheme for each unit is optimized through finite element analysis. A safety factor is introduced into the calculation of unit vibration demand measurement. The parameters of the structural damping model are calibrated through experimental modal analysis, and a fault coding system is established through historical record retrieval. The risk matrix of the comprehensive risk analysis model is 5×5, and the unit activity output value is updated every ten minutes. Classification threshold comparison results trigger different colored warning signals.

[0038] The formation process of the evaluation unit set undergoes redundancy checks to ensure that each physical unit is classified only once. The stability of the unit screening index value is enhanced using a sliding window averaging method, and the modular segmentation scheme has been reviewed and certified by experts. Unit vibration demand measurement calculations consider temperature compensation, historical record retrieval establishes data lineage tracking, and the comprehensive risk analysis model integrates Monte Carlo simulation. Unit activity ranking is updated in real time, and the time complexity of the classification threshold comparison algorithm is optimized to O(logn). The output format of the evaluation unit set adopts the JSON standard. A quality traceability log is established for the entire modular segmentation process, recording timestamps and operator information for each operation step. Changes to the evaluation unit set require change control approval. Strain gauge calibration coefficients are introduced in the unit vibration demand measurement, historical record retrieval establishes a data quality scoring mechanism, and the comprehensive risk analysis model outputs a risk heatmap. Unit activity ranking is visualized, classification threshold comparison results generate decision reports, and version management of the evaluation unit set adopts the Git protocol. The integrity of modular segmentation is measured by coverage metrics, an exception handling mechanism is established for the calculation of unit screening index values, and the export of the evaluation unit set supports multiple file formats.

[0039] Modular segmentation of lifting equipment needs to consider on-site installation conditions, and the weight distribution of each unit affects the sensor placement scheme. Dynamic load spectra are introduced in the calculation of unit vibration demand, and a fault tree model is established through historical record retrieval. A comprehensive risk analysis model couples multiple physical field data, and the unit activity ranking achieves adaptive adjustment. A classification threshold comparison algorithm is embedded in edge computing devices. A configuration management database is established for maintaining the evaluation unit set, and the optimization of modular segmentation is achieved through machine learning algorithms. Cross-validation is used to verify the unit selection index values. A continuous improvement mechanism is established throughout the process, with the modular segmentation scheme revised periodically based on actual operating data, and the update frequency of the evaluation unit set consistent with the equipment overhaul cycle. Non-destructive testing data is incorporated into the calculation of unit vibration demand, a knowledge graph is established through historical record retrieval, and digital twin technology is integrated into the comprehensive risk analysis model.

[0040] Example 3: Continuous monitoring of vibration deviation and stress diffusion velocity in each evaluation unit is performed using an embedded accelerometer network and a thermal imager or strain gauge array. The embedded accelerometer network is deployed on the surface of the evaluation unit, with sensor node spacing arranged according to the Nyquist sampling theorem. The thermal imager uses an infrared focal plane array detector, and the strain gauge array uses foil resistance strain gauges. During continuous measurement, the vibration change sequence per unit time is recorded, with a sampling interval of 100 microseconds, to obtain continuous vibration readings. The vibration readings are digitized using a 24-bit analog-to-digital converter. The vibration deviation is generated based on differential and wave analysis models. The differential model calculates the first-order forward difference between adjacent sampling points, and the wave analysis model uses an empirical mode decomposition algorithm to identify intrinsic mode functions. The quantized value of the vibration deviation is calculated using the following formula:

[0041]

[0042] in: This represents the vibration deviation (unit: m / s²). Indicates the total number of sampling points. This represents the vibration acceleration reading at the i-th sampling point (unit: m / s²). This represents the vibration acceleration reading at the (i-1)th sampling point (unit: m / s²). This represents the arithmetic mean of the differences between adjacent sampling points (unit: m / s²).

[0043] The stress diffusion rate is monitored by measuring the stress change per unit time in each evaluation unit. The observation period is set to 1 second, and the observation period is synchronized with vibration sampling to acquire continuous stress data. The stress difference is obtained by subtracting the stress value at the start of the observation from the stress value at the end of the observation. Dividing the stress difference by the corresponding observation duration yields the stress diffusion rate scalar, with units of MPa / s. Vibration deviation and stress diffusion rate are standardized using a minimum-maximum normalization method to map the data to the [0,1] interval and input into a linear correlation model. This model is a multiple linear regression model, generating a contribution factor for the state deviation. This contribution factor represents the degree of influence of each evaluation unit on the overall state deviation. The embedded accelerometer network is configured considering spatial coverage density. Each evaluation unit surface has at least 8 sensor nodes. The sensor nodes use industrial-grade MEMS accelerometers with a range of ±50g and a frequency response of 0-5kHz. The thermal imager has a spatial resolution of 0.5mrad, a thermal sensitivity of 0.05℃, and a strain gauge array with a grid length of 3mm and a resistance of 120Ω. Vibration change sequences are recorded using a circular buffer with a depth of 10,000 sampling points. Continuous vibration readings are preprocessed using a digital filter, specifically a fourth-order Butterworth low-pass filter with a cutoff frequency of 1 kHz. The differential model calculation employs a real-time sliding window algorithm with a window width of 100 sampling points. The eigenmode function decomposition layer of the wave analysis model is set to 10 layers, and the vibration deviation calculation results are updated every second.

[0044] In stress diffusion rate monitoring, GPS clock sources are used for time synchronization at the start and end of the observation period. Continuous stress data is acquired through a full-bridge circuit, and temperature compensation coefficients are considered in the stress difference calculation. The stress diffusion rate calculation results are smoothed using a moving average, with a moving average window width of 10 data points. During standardization, the boundary values ​​for minimum-maximum normalization are derived from historical data statistics. The training data for the linear correlation model contains 1000 samples, and the output values ​​of the contribution factors are retained to three decimal places. Data transmission in the embedded sensor network adopts the industrial Ethernet protocol, with a transmission latency of less than 1 millisecond. The temperature measurement accuracy of the thermal imager reaches ±1℃, and the micro-strain measurement resolution of the strain gauge array reaches 1. The sampling interval of the vibration change sequence is dynamically adjusted according to the dominant vibration frequency. The Nyquist frequency is set to be more than twice the highest frequency of the signal. The storage format of continuous vibration readings adopts the IEEE floating-point standard. The first-order forward difference calculation of the differential model adopts a parallel processing architecture, and the Hilbert transform operation of the wave analysis model is accelerated using FPGA. The real-time display of vibration deviation is in the form of waveform charts. The observation period for stress diffusion rate monitoring can be configured from 0.1 seconds to 10 seconds. The sampling rate of continuous stress data is set to 10 kSPS. The stress difference calculation incorporates a material creep correction coefficient. The alarm threshold for stress diffusion rate is set according to the material yield strength, and exceeding the threshold triggers an audible and visual alarm. The parameters of the standardized processing are automatically calibrated every 24 hours. The regression coefficients of the linear correlation model are solved using the least squares method, and the effective range of the contribution factor is between 0 and 1. The node layout scheme of the embedded acceleration sensor network has been verified by modal testing. The sensors are installed using magnetic fixation, and the network topology adopts a star connection structure. The infrared band of the thermal imager was selected as 8-14 μm, the image acquisition frequency was 30 frames / second, and the bridge power supply voltage of the strain gauge array was 10V DC. Lossy compression algorithm was used for data compression of the vibration change sequence, with a compression ratio set to 4:1. Time-domain feature extraction of continuous vibration readings included peak factor and kurtosis index. The numerical stability of the difference model was guaranteed by the fourth-order Runge-Kutta method, the boundary treatment of the wave analysis model used the mirror continuation method, and the long-term trend analysis of vibration deviation used the ARIMA model.

[0045] The influence of ambient temperature on stress diffusion rate monitoring is compensated for by thermocouples. Noise suppression of continuous stress data employs digital lock-in amplification technology, and the real-time performance of stress difference calculation meets millisecond-level response requirements. Outlier detection in standardized processing utilizes the box plot method, the significance test of the linear correlation model uses the F-test, and the confidence interval calculation of contribution factors uses the Bootstrap method. The entire monitoring system establishes a self-diagnostic mechanism, automatically detecting sensor faults and switching to backup channels. The integrity of data acquisition is guaranteed by CRC verification. Vibration deviation calculation incorporates a speed compensation factor to eliminate the impact of equipment speed fluctuations. A material database is established for stress diffusion rate monitoring, with the elastic modulus and Poisson's ratio parameters of different materials pre-entered. The signal conditioning circuit of the embedded accelerometer network uses an instrumentation amplifier design, achieving a common-mode rejection ratio of 100dB. Non-uniformity correction of the thermal imager is automatically performed every 8 hours. Spectral analysis of vibration change sequences uses Fast Fourier Transform with a frequency resolution of 0.5Hz. Continuous vibration readings are stored using a ring buffer structure to prevent data loss. The real-time implementation of the difference model uses a recursive algorithm to reduce computational memory usage, and the endpoint effects of the fluctuation analysis model are improved by symmetric extension of extreme points.

[0046] See Figure 4 , Figure 4 This chart presents a comparative analysis of vibration deviation and stress diffusion rate monitoring across various assessment units of the lifting equipment. As a grouped bar chart, it focuses on six assessment units: the main beam unit, wire rope unit, pulley block unit, outrigger system, hoisting mechanism, and traveling device. It simultaneously displays two core monitoring indicators: vibration deviation and stress diffusion rate, visually depicting the distribution differences of multi-dimensional state data. From a data perspective, vibration deviation reflects the degree of vibration fluctuation in a unit, while stress diffusion rate reflects the rate of stress change per unit time. Together, they constitute a key quantitative dimension of the structural health of the assessment unit. This chart provides intuitive data support for the technical process of continuously monitoring vibration deviation and stress diffusion rate in each assessment unit, assessing the contribution of state deviation, and calculating contribution factors. Through the visual comparison of these two indicators, it helps technicians quickly identify differences in the state characteristics of each unit, laying a foundation for accurate data visualization analysis in subsequent contribution factor calculations and unit rating.

[0047] Example 4: Evaluation of the contribution of state deviations in each assessment unit and continuous load interval tracking. The contribution factor of the state deviation in each assessment unit is compared with a predefined impact threshold. The impact threshold is set based on statistical analysis of historical fault data, and units with low contributions are selected as identified assessment units. The contribution factor comes from the standardized value output by the linear correlation model, and the impact threshold value is dynamically adjusted according to the equipment type. The comparison operation is implemented using a digital comparator circuit. If the contribution factor reaches or exceeds the impact threshold, the contribution of the state deviation is considered high, and the corresponding assessment unit is marked as low-level and excluded. The exclusion operation includes removing the unit from the assessment list and generating a maintenance work order. If the contribution factor is lower than the impact threshold, the contribution of the state deviation is considered low, and the corresponding assessment unit is retained and identified. The identification information includes the unit number, contribution factor value, and monitoring timestamp. When the lifting equipment starts operation, the load duration interval of the monitoring and evaluation unit is captured by a high-precision timing module. The load start time and load end time of each evaluation unit at the moment of triggering of the lifting equipment are recorded by pressure sensors. The load duration interval value is obtained based on time difference calculation, and the time difference calculation takes into account signal transmission delay compensation.

[0048] Table 1: Comparison Rules between Contribution Factors and Impact Thresholds

[0049]

[0050] Referring to Table 1, the load duration interval of each identification evaluation unit is associated with the corresponding unit's contribution factor and normalized. The normalization process uses a minimum-maximum scaling algorithm to map the data to the [0,1] interval, which is then input into the S-curve fusion model. The S-curve fusion model uses a logic function to output the rating of each identification evaluation unit, with rating results divided into four levels: A / B / C / D. The contribution factor and influence threshold comparison module is integrated in the programmable logic controller, and the influence threshold value is stored in non-volatile memory. The comparison frequency is set to once per second. The marker of a high-contribution unit triggers an audible and visual alarm device, and the identification information of a low-contribution unit is written to the database. The load duration interval monitoring uses a 32-bit timer chip, and the timestamp synchronization uses the IEEE 1588 precision clock protocol. The load start time detection is based on a threshold triggering mechanism, and the load termination time determination considers a signal de-jitter algorithm. Temperature drift compensation is introduced in the time difference calculation, and the load duration interval value is stored in milliseconds. The parameter boundaries of the normalization process are updated weekly, the slope parameter of the S-curve fusion model is configurable, and the unit rating results are displayed in real time on the monitoring interface. Historical data on contribution factors is stored for more than three years, and the impact threshold is dynamically adjusted according to seasonal changes. Maintenance reports are generated for the exclusion records of high-contribution units. Load duration interval data for the identified evaluation units is stored cyclically, retaining the most recent 1000 operation records. The sampling frequency of the pressure sensor is set to 1kHz. The accuracy of time difference calculation reaches the microsecond level, signal transmission delay is determined through calibration experiments, and outliers in the load duration interval are eliminated using the 3σ criterion. The output of the S-curve fusion model is rounded, and the unit rating results are linked to the equipment maintenance plan. Grade A units have extended inspection cycles, and Grade D units initiate emergency maintenance. The contribution factor and impact threshold comparison system establishes an automatic audit log, recording the operator's identity and time information for each comparison operation. Exclusion of high-contribution units requires secondary confirmation permissions to prevent monitoring interruptions due to misoperation. The load duration interval monitoring module is equipped with a backup battery power supply to ensure no data loss during power failures, and the constant of the load start-up detection filter is set to 0.1 seconds. The maximum and minimum values ​​of the normalization processing are derived from statistics from the rolling time window. The parameters of the S-curve fusion model are optimized through machine learning algorithms, and the unit rating results are pushed to mobile terminals. The geographical location information of the assessment units is linked to the rating results to generate a three-dimensional visualization map, with different colors indicating the distribution of units at different rating levels.

[0051] Load duration interval data is cross-validated with equipment operation logs to ensure the accuracy of time recording; abnormal fluctuations in contribution factors trigger a data review process to eliminate misjudgments caused by sensor malfunctions. An S-curve fusion model establishes a temperature compensation coefficient to eliminate the influence of ambient temperature on rating results; historical trend analysis of unit ratings uses a moving average algorithm to identify deteriorating rating trends. A tiered early warning mechanism is set for contribution factor comparison thresholds: yellow levels trigger attention alerts, and red levels trigger emergency responses; a baseline model is established for load duration interval monitoring to identify abnormal load patterns. The normalization algorithm undergoes numerical stability testing to prevent calculation errors caused by data overflow; the S-curve fusion model outputs confidence intervals, and the rating results include reliability indicators. Histogram analysis is used to statistically analyze the load duration interval distribution of the evaluation units to identify typical load patterns; an autocorrelation model is established for contribution factor time series to detect periodic changes in contribution factors. S-curve fusion model parameters are periodically validated, and the model output is calibrated using standard test signals; unit rating results are coupled with an equipment life prediction model to generate an estimate of remaining service life. Load duration interval monitoring data is compressed and stored using a lossy compression algorithm to reduce storage space; contribution factor comparison results generate a quality report, statistically analyzing the proportional distribution of various contribution units. Normalization is used to establish an abnormal data processing flow, and missing data is supplemented using interpolation methods; the S-curve fusion model integrates fault diagnosis rules to distinguish between occasional faults and trend faults.

[0052] The system analyzes the correlation between load duration interval and work cycle count for each evaluation unit to establish a load intensity index; a spatial distribution map of contribution factors shows the interrelationships between units. An S-curve fusion model establishes an adaptive adjustment mechanism to adjust rating standards based on equipment aging; unit rating results drive maintenance resource allocation and optimize maintenance strategies. The load duration interval monitoring system incorporates electromagnetic compatibility design to suppress electromagnetic interference in the industrial environment; a data quality assessment mechanism is established for contribution factor calculation, marking low-quality data as unusable. Normalized parameters are remotely configured, supporting online modification of boundary values; the S-curve fusion model output is smoothed to avoid frequent rating level jumps. A contribution factor comparison system monitors performance indicators and statistically compares operation response times; load duration interval data is encrypted and stored to prevent unauthorized modifications. The load duration interval of each evaluation unit is compared with the design specifications to verify equipment operating status; the S-curve fusion model establishes version control, recording parameter modification history. Unit rating results generate standardized reports conforming to industry-standard formats; load duration interval analysis identifies overload operating modes, and contribution factor trends predict equipment health status. The normalization algorithm is used to perform numerical accuracy testing to ensure that the calculation error is within the allowable range; the S-curve fusion model integrates expert knowledge to improve the interpretability of the rating results.

[0053] Load duration monitoring data is backed up to a cloud server to achieve data redundancy across multiple locations; contribution factor comparison results generate decision support information to assist maintenance personnel in formulating maintenance plans. Historical rating data for identified assessment units is indexed for rapid querying and analysis; genetic algorithms are used to optimize the parameters of the S-curve fusion model to find the optimal parameter combination. Correlation analysis between unit rating results and equipment operating efficiency assesses the effectiveness of maintenance measures; statistical features of load duration intervals are extracted to establish an equipment operating fingerprint database. Spatial correlation analysis of contribution factors identifies fault propagation paths between units; normalization processing establishes data quality standards, ensuring only qualified data enters the fusion process. The S-curve fusion model outputs uncertainty quantification, and rating results include probability distribution information; anomaly detection of load duration intervals uses an isolated forest algorithm to identify abnormal operating patterns. The contribution factor comparison system performs sensitivity analysis to assess the impact of threshold changes on results; correlation analysis is performed between the load duration interval of identified assessment units and environmental impact factors. Unit rating results are visualized, supporting multi-dimensional data drill-down; normalization processing establishes data lineage tracking, recording data transformation history. The S-curve fusion model integrates real-time learning capabilities, dynamically adjusting model parameters based on new data. Load duration interval prediction employs time series analysis to identify potential failure risks in advance. The contribution factor comparison system establishes a verification test case library and conducts regular functional verification. It performs a goodness-of-fit test on the load duration interval distribution of the evaluation units to verify the distribution hypothesis. Unit rating results are correlated with equipment energy consumption data to assess operational economy. The S-curve fusion model generates explanatory reports explaining the rating basis. A completeness verification mechanism is established for load duration interval monitoring data to prevent data loss. The contribution factor calculation process is traceable, supporting result reproducibility. The normalization algorithm undergoes boundary value testing to ensure correct handling in extreme cases. The output of the S-curve fusion model is compared with manual evaluation results to verify model accuracy.

[0054] The system employs a comprehensive evaluation method, including load duration interval clustering analysis to identify typical operating modes and contribution factor change rate monitoring to detect abrupt state changes. Unit rating results generate maintenance suggestions to guide on-site operators; an S-curve fusion model establishes a reliability growth model to continuously improve rating accuracy. Load duration interval data quality monitoring detects sensor drift; contribution factor spatial interpolation analysis generates a full-field state distribution map. Normalization parameters are automatically optimized, adjusting the mapping range based on data characteristics; the S-curve fusion model integrates multi-source information to improve rating robustness. Statistical process control of contribution factor comparison results detects system performance changes; load duration interval periodic analysis identifies equipment operating cycles. A comprehensive health status assessment of the identified evaluation units integrates multi-parameter information; the S-curve fusion model outputs confidence assessments to support decision-making. Correlation analysis between load duration intervals and equipment operation records reconstructs operational history; contribution factor frequency characteristic analysis identifies characteristic frequency components. Trend prediction of unit rating results provides early warning of state degradation; normalization processing establishes data governance standards to ensure data consistency. S-curve fusion model parameter sensitivity analysis identifies key parameters; abnormal load duration pattern identification establishes a fault feature library. Contribution factor comparison system undergoes stress testing to verify system stability; a comprehensive status assessment report for the identified evaluation units is generated, producing standardized output documents.

[0055] Example 5: Taking the main beam evaluation unit of a lifting machine as an example, the main beam evaluation unit is numbered GJ-2024-MB-01, made of Q345 low alloy steel, with a constant low load applied at 50 kN. Laser displacement sensor measurement points are located at the mid-span and support points. After filtering, the elastic deformation measured at the mid-span of the main beam is 2.5 mm, and at the support point, it is 0.8 mm. Interface stress calculation is performed using finite element software, discretizing the main beam evaluation unit into 10,000 hexahedral elements. Boundary conditions simulate the actual support state, and the calculated maximum interface stress is 185 MPa, with a stress concentration factor of 1.8. Fatigue response model parameters include a stress ratio R=0.1, a load frequency of 2 Hz, and a maximum allowable deformation set to 1 / 400 of the span according to design specifications. The calculated ultimate load tolerance of the main beam evaluation unit is 280 kN. The interface stress and ultimate load tolerance of each identification assessment unit were standardized using the Z-score method, converting the raw data into a distribution with a mean of zero and a standard deviation of one. The data were then input into an entropy weight allocation model, which calculates the dispersion of the indicators based on information entropy theory to obtain the rating weights for each identification assessment unit. The rating weights range from 0 to 1, and the sum of all weights is 1. The ratings and weights of each identification assessment unit were combined using a linear combination calculation employing a weighted arithmetic mean formula to output the health index of the lifting equipment under the current operating conditions. The health index value is a scalar value between 0 and 100. The hydraulic servo control system achieves a pressure accuracy of ±0.5%FS, a loading rate controlled at 10 mm / min, and the laser displacement sensor installation position is precisely located using a total station. Material parameters in the elasticity theory calculations were obtained from material certification certificates, with a Poisson's ratio of 0.3 and an elastic modulus of 206 GPa. The interface stress calculation results were verified for mesh independence. The load spectrum of the fatigue response model adopts a normal distribution assumption, with the number of cycles set to 2 million, and the maximum allowable deformation value confirmed by expert review. The mean and standard deviation of the standardized processing are obtained from historical database statistics. The calculation process of the entropy weight allocation model includes constructing a judgment matrix and a consistency test.

[0056] Taking the wire rope evaluation unit as an example, the unit number is GJ-2024-WR-02, with a structural specification of 6×36WS+IWR. The constant low load applied is 25 kN, and the deformation measurement uses a binocular vision measurement system. The elongation measurement of the wire rope evaluation unit under load is 0.15%. The interface stress calculation considers the contact stress between strands, and the maximum interface stress is 315 MPa. The tolerance limit load calculation incorporates the damage tolerance design concept, with a safety factor of 3.0, resulting in a tolerance limit load of 75 kN for the wire rope evaluation unit. In the entropy weight allocation model calculation, the information entropy value of the interface stress is 0.86, and the information entropy value of the tolerance limit load is 0.92, resulting in a rating weight of 0.15 for the wire rope evaluation unit. The weighting coefficients of the linear combination operation are displayed in real time on the monitoring interface. The health index calculation results are updated every 5 minutes. A yellow warning is triggered when the health index is below 60, and a red alarm is triggered when it is below 30. The pressure sensor of the hydraulic servo control system is calibrated periodically, with a calibration cycle of three months. The temperature compensation coefficient of the laser displacement sensor is preset. Elastic theory calculations establish a constitutive relation database for materials, with elastic modulus values ​​at different temperatures pre-entered, and interface stress distribution cloud maps displayed in real time. Load history data for the fatigue response model is extracted from actual operating records, and the maximum allowable deformation value is adjusted according to the equipment's service life. A standardization process establishes a dynamic adjustment mechanism, with the mean and standard deviation recalculated every 24 hours.

[0057] Taking the pulley block evaluation unit as an example, the unit number is GJ-2024-PU-03, made of ZG310-570 cast steel, with a constant low load applied at 15 kN. Deformation is measured using resistance strain gauges. The maximum strain measured at the rope groove of the pulley block evaluation unit is 520 με. The interface stress calculation considers the coupling of bending stress and contact stress, with a maximum interface stress value of 180 MPa. A stress concentration factor of 1.5 is introduced into the ultimate load tolerance calculation, resulting in an ultimate load tolerance of 45 kN for the pulley block evaluation unit. In the entropy weight allocation model calculation, the entropy weight for interface stress is 0.32, the entropy weight for ultimate load tolerance is 0.68, and the rating weight for the pulley block evaluation unit is 0.08. A historical data comparison mechanism is established for the health index calculation. An analysis report is generated when the current health index is compared with the data from the same period last week, and the change rate exceeds 10%. The safety interlock device of the hydraulic servo control system ensures safety during the loading process, and the laser displacement sensor has an IP67 protection rating. A verification model was established using elasticity theory calculations, and photoelastic experiments were used to verify the stress distribution. The locations of peak interface stresses were marked as key monitoring points. The damage accumulation calculation of the fatigue response model considered the influence of overload conditions, and a 20% margin was set for the maximum allowable deformation. Standardization processing was used to establish data quality inspection rules, automatically removing abnormal data and re-collecting it.

[0058] See Figure 5 , Figure 5This document presents a trend chart of the health index of lifting equipment over operating time. A multi-level review process is established for health index output, requiring confirmation from the equipment administrator for the health index report to be effective. The pressure curve of the hydraulic servo control system is recorded in real time, and the measurement data from the laser displacement sensor is time-stamped synchronously. An elasticity theory calculation establishes a material damage database, recording calculation parameters for each iteration, and performing uncertainty analysis on the interface stress results. The damage threshold of the fatigue response model is adjusted according to material batches, and a safety warning line is set for the maximum allowable deformation. A standardized data quality assessment system is established, with a quality score attached to each data point. The final health index calculation results generate a standardized report, including detailed data for each assessment unit, the weight allocation process, and the health index calculation steps. Health index values ​​are pushed to the enterprise equipment management system and automatically integrated with the maintenance work order system. The health index trend is incorporated into the equipment's full lifecycle management. Historical health data from each assessment unit is used to establish predictive models, and the effectiveness of the health index calculation method is continuously verified and optimized through actual operating data.

[0059] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for testing and verifying the condition of lifting equipment based on PHM, characterized in that, The method includes: Design parameters are extracted from the technical documents of the lifting equipment, and high-frequency operating characteristics are collected when the lifting equipment is started. The design parameters and high-frequency operating characteristics are combined to determine whether to start the multi-dimensional status assessment process. The design parameters include structural material durability limits. By reading material density, elastic modulus, and material thickness, a stress superposition model is constructed to determine the surface stress of the structure and aggregate it with the baseline stress to obtain the structural material durability limits. Using a standard operating test sequence and operating frequency setting, the crane under test is subjected to cyclic load operation within a specified period, and the total number of operations is counted to generate high-frequency operating characteristics. The structural material durability limits and high-frequency operating characteristics are compared with preset thresholds in turn to generate a conclusion on the rationality of the current crane design. If the conclusions all indicate that the current crane design is invalid, the crane health index is directly determined to be low in the current state. Otherwise, a multi-dimensional state assessment process is performed on the crane in the current state. If the multidimensional state assessment process is initiated, the design robustness index of the lifting equipment is calculated, the structural vibration status of the lifting equipment is detected, and the lifting equipment is modularly segmented and multiple sets of assessment units are extracted by combining the structural vibration status and the design robustness index. Continuously monitor the vibration deviation and stress diffusion rate of each evaluation unit, evaluate the contribution of each evaluation unit in the lifting equipment to the state deviation and calculate the contribution factor, track the load duration interval of each evaluation unit, and screen out low contribution units as identification evaluation units. The contribution factors of each unit are associated with the load duration interval of each identification assessment unit to perform unit rating. The interface stress and tolerance limit load of each identification assessment unit are collected, the rating weight of each assessment unit is derived, and the rating and weight of each identification assessment unit are integrated to obtain the health index of the lifting equipment under the current working conditions.

2. The method for testing and verifying the condition of lifting equipment based on PHM according to claim 1, characterized in that, The durability limits and high-frequency operating characteristics of the structural materials are normalized, and a weighted average algorithm is applied to output the design robustness index. With the help of miniature accelerometers arranged on the structural channel of the lifting machinery, the vibration amplitude and frequency variation are measured to detect the vibration status of the lifting machinery structure. An effective vibration assessment model based on vibration propagation theory is used to output the vibration intensity per unit time.

3. The method for testing and verifying the condition of lifting equipment based on PHM according to claim 2, characterized in that, The design robustness index and vibration intensity per unit time are standardized and transformed, and then input into the decision tree model to generate unit screening index values; the lifting equipment is modularly segmented to obtain each unit of the lifting equipment, and a comprehensive calculation is performed based on the unit vibration demand metric and the unit's historical failure probability; according to the material stress properties of the unit, the expected stress load of the unit per unit time is calculated, and the minimum attenuation energy required is estimated with reference to the structural damping model to obtain the unit vibration demand metric. By retrieving the historical records of lifting equipment, fault instances of each structural unit are extracted and summarized by unit. The cumulative number of faults of the unit within the reference period is compared with the number of running segments to obtain the historical failure probability of the unit.

4. The method for testing and verifying the condition of lifting equipment based on PHM according to claim 3, characterized in that, The unit vibration demand measurement and the unit historical failure probability are normalized and fed into the comprehensive risk analysis model to determine the unit activity level of each unit of the lifting equipment. The unit activity levels are sorted in ascending order of numerical value. The unit screening index value is compared with the multiple sets of classification thresholds through predefined groups of classification thresholds. Evaluation unit groups in which all classification thresholds do not exceed the current index value are selected. Combined with the activity level sorting results, all evaluation units in which these classification thresholds are located before the current index value in the sorting are selected simultaneously.

5. The method for testing and verifying the condition of lifting equipment based on PHM according to claim 1, characterized in that, An embedded accelerometer network and a thermal imager or strain gauge array are used to continuously measure the vibration deviation and stress diffusion rate of each evaluation unit. By recording the vibration change sequence per unit time and setting the sampling interval, continuous vibration readings are obtained and processed according to the differential and wave analysis model to generate the vibration deviation. By monitoring the stress change of each evaluation unit per unit time and setting the observation period, continuous stress data are obtained. The stress diffusion rate is obtained by subtracting the stress at the beginning of the stress observation from the stress at the end of the stress observation and quotienting it with the corresponding observation duration. The vibration deviation and stress diffusion rate are standardized and input into a linear correlation model to generate the contribution factor of the state deviation.

6. The method for testing and verifying the condition of lifting equipment based on PHM according to claim 5, characterized in that, The contribution factor of the state deviation in each evaluation unit is compared with the predefined influence threshold. If the contribution factor reaches or exceeds the influence threshold, the contribution of the state deviation is considered to be high, and the corresponding evaluation unit is marked as low and excluded. Otherwise, if the contribution factor is lower than the influence threshold, the contribution of the state deviation is considered to be low, and the corresponding evaluation unit is retained and marked. When the lifting equipment starts operation, the load duration interval of the marked evaluation units is monitored, the start and end times of the load of each marked evaluation unit are captured at the moment the lifting equipment is triggered, and the load duration interval is obtained based on the time difference calculation.

7. The method for testing and verifying the condition of lifting equipment based on PHM according to claim 6, characterized in that, After normalizing the contribution factors of the corresponding units associated with the load duration interval of each identification evaluation unit, the data is input into the S-curve fusion model to output the rating of each identification evaluation unit.

8. The method for testing and verifying the condition of lifting equipment based on PHM according to claim 7, characterized in that, By applying a constant low load to each identification assessment unit and measuring its deformation, the interfacial stress of each identification assessment unit is calculated based on elasticity theory; combined with the structural material properties, a fatigue response model is established to set the maximum allowable deformation, and the ultimate load that each identification assessment unit can withstand is calculated; the interfacial stress and ultimate load of each identification assessment unit are standardized and input into the entropy weight allocation model to obtain the rating weight of each identification assessment unit.

9. The method for testing and verifying the condition of lifting equipment based on PHM according to claim 8, characterized in that, The health index of the lifting equipment under the current working conditions is output by combining the ratings and rating weights of each identification assessment unit through linear combination calculation.

Images