Methods, systems and related devices for predicting fatigue life of high-parameter elevator traction machine spindles

Abstract

This disclosure provides a method, system, and related apparatus for predicting the fatigue life of a high-parameter elevator traction machine spindle. The method includes: constructing a digital model of the spindle based on its structural and operational parameters; using a laser displacement sensor to acquire the actual bending deformation of the spindle under at least one load condition, comparing it with the simulated bending deformation obtained from the digital model to verify and correct the digital model; using the corrected digital model to simulate and acquire fatigue life data under different bending deformations; training a fatigue life prediction model based on a random forest regression algorithm using the simulated fatigue life data; acquiring the actual bending deformation and current load condition of the traction machine spindle in real time; inputting the actual bending deformation and current load condition into the trained fatigue life prediction model, which then outputs a predicted value for the remaining fatigue life of the traction machine spindle.

Description

Technical Field

[0001] This invention relates to the field of high-parameter elevator manufacturing and maintenance technology, and more specifically, to a method, system and related device for predicting the fatigue life of the main shaft of a high-parameter elevator traction machine. Background Technology

[0002] High-parameter elevators refer to two categories: high-speed elevators and heavy-load elevators. The complex dynamic characteristics they exhibit during operation cause their traction machine main shaft to operate under high-speed, variable-load conditions for extended periods, leading to wear or bending deformation. This significantly impacts the fatigue life of the traction machine main shaft and jeopardizes the safe operation of the elevator.

[0003] Effective prediction of spindle fatigue life is an important means to ensure the safe and reliable operation of high-parameter elevators. Existing analysis and prediction of traction machine spindle fatigue life includes: (1) disassembling the traction machine spindle on site, then inspecting each major component, analyzing the functional status of the traction machine spindle, and evaluating the fatigue life of the spindle. The whole process is quite complicated; (2) analyzing and judging the fatigue life of the spindle by monitoring the vibration, frictional heat and other relevant operating data of the spindle accessories such as bearings. It is necessary to ensure that the bearings and other accessories themselves are operating normally and within a reliable operating life.

[0004] As can be seen from the above, the existing detection technology for predicting the fatigue life of elevator traction machine main shaft is quite complicated. It generally requires shutting down the elevator and disassembling the main shaft for detection, or it requires analyzing the operating status of its auxiliary parts. This reduces the detection efficiency and accuracy of elevator main shaft fatigue life. Summary of the Invention

[0005] This invention provides a method, system, and related device for predicting the fatigue life of a high-parameter elevator traction machine spindle. It uses a fatigue life prediction model based on a random forest regression algorithm to predict the fatigue life of the spindle directly, thus solving the technical problems of low detection efficiency and accuracy of existing elevator spindle fatigue life.

[0006] On one hand, this invention provides a method for predicting the fatigue life of a high-parameter elevator traction machine spindle, comprising: constructing a digital model of the spindle based on its structural and operational parameters; acquiring the actual bending deformation of the spindle under at least one load condition using a laser displacement sensor installed on the upper end cover of the traction machine, and comparing the actual bending deformation with the simulated bending deformation obtained from the simulation of the spindle digital model to verify and correct the spindle digital model; using the corrected spindle digital model to simulate and obtain fatigue life data of the traction machine spindle under different bending deformations; training a fatigue life prediction model based on a random forest regression algorithm using the simulated fatigue life data; acquiring the actual bending deformation and current load condition of the traction machine spindle in real time using the laser displacement sensor; inputting the real-time acquired actual bending deformation and current load condition into the trained fatigue life prediction model, and having the fatigue life prediction model output a predicted value for the remaining fatigue life of the traction machine spindle.

[0007] In some embodiments, the present invention distinguishes between the elastic deformation stage and the plastic deformation stage of the traction machine spindle when constructing the digital model of the spindle, wherein:

[0008] For the elastic deformation stage, the fatigue life of the principal shaft is represented by the Basquin equation, as follows:

[0009]

[0010] N f Indicates the spindle fatigue life. δ is the material fatigue strength coefficient, b is the Basquin exponent, L represents the main shaft span, d represents the main shaft diameter, δ represents the bending deformation of the main shaft (also known as "deflection"), and E is the elastic modulus.

[0011] For the plastic deformation stage, the fatigue life of the spindle is represented by the Manson-Coffin equation, as follows:

[0012]

[0013] N f Indicates the spindle fatigue life, δ p This indicates the amount of permanent bending deformation of the spindle. is the fatigue ductility coefficient, c is the fatigue ductility index, L represents the spindle span, and d represents the spindle diameter.

[0014] In some embodiments, the laser displacement sensor includes: a first laser displacement sensor and a second laser displacement sensor respectively installed at the two bearing supports of the main shaft near the upper end cover of the traction machine, and a third laser displacement sensor installed at the center of the upper end cover of the traction machine. The actual bending deformation of the main shaft is obtained using a three-point testing method, which includes:

[0015] The first laser displacement sensor, the second laser displacement sensor, and the third laser displacement sensor are used to measure the initial displacement value when the elevator is unloaded during its first use and the displacement value detected subsequently.

[0016] The bending deformation of the spindle is calculated using the following formula:

[0017] δ = Δb - 0.5 × (Δa + Δc)

[0018] Wherein, δ is the bending deformation of the main shaft, used to represent the bending deformation of the main shaft, Δa represents the difference between the displacement value currently detected by the first laser displacement sensor and the corresponding initial displacement value, Δb represents the difference between the displacement value currently detected by the second laser displacement sensor and the corresponding initial displacement value, and Δc represents the difference between the displacement value currently detected by the third laser displacement sensor and the corresponding initial displacement value.

[0019] In some embodiments, the step of using the modified spindle digital model to simulate and obtain fatigue life data of the traction machine spindle under different bending deformation amounts includes: taking values ​​of different bending deformation amounts within a predetermined range with preset increments, and obtaining fatigue life data covering the elastic deformation and plastic deformation stages through the spindle digital model.

[0020] In some implementations, training the fatigue life prediction model based on the random forest regression algorithm using the fatigue life data obtained from simulation includes: analyzing the fatigue life of the traction machine main shaft under elevator no-load and full-load conditions using the fatigue life data obtained from simulation, and corresponding the obtained fatigue life values ​​with each bending deformation to form a dataset; using the dataset, with bending deformation and load state as input features and fatigue life as output, to train the fatigue life prediction model.

[0021] In some implementations, the step of correlating the analyzed fatigue life values ​​with each bending deformation to form a dataset includes: classifying the fatigue life values ​​into initial, intermediate, and final stages, setting labels and corresponding bending deformation and load states to form a dataset.

[0022] In some implementations, the method further includes: dividing the dataset into a training set and a test set according to a predetermined ratio, and using stratified sampling to maintain data distribution consistency.

[0023] In some implementations, training the fatigue life prediction model using the dataset includes: drawing multiple training subsets with replacement from the training set using a bootstrap method (a statistical method based on repeated sampling), with each decision tree generated based on one training subset; randomly selecting a predetermined number of candidate features from all input features when splitting at each node of a single decision tree, and selecting the best split from the predetermined number of candidate features to minimize the squared error; and evaluating the fatigue life prediction model based on the data from the test set using the out-of-bag coefficient of determination and the root mean square error to determine the effectiveness of the prediction model.

[0024] In some implementations, the method further includes: performing hyperparameter optimization using a grid search method during training, wherein the optimized hyperparameters include at least: the number of decision trees (n_estimators), the maximum number of features (max_feature), the maximum depth of the tree (max_depth), the minimum number of samples required for node splitting (min_samples_split), and the minimum number of samples required for leaf nodes (min_samoles_leaf).

[0025] On the other hand, embodiments of the present invention provide a high-parameter elevator traction machine spindle fatigue life prediction system, used to implement the method described in any embodiment of the present invention, comprising: a digital modeling module, used to construct a digital model of the spindle based on the structural and operating parameters of the traction machine spindle; a sensor detection module, including a laser displacement sensor installed on the upper end cover of the traction machine, used to detect the actual bending deformation of the spindle under at least one load state; a digital model verification module, used to compare the actual bending deformation with the simulated bending deformation obtained by simulating the spindle digital model to verify and correct the spindle digital model; a data analysis module, used to use the corrected spindle digital model to simulate and obtain fatigue life data of the traction machine spindle under different bending deformations; a prediction model training module, used to train a fatigue life prediction model constructed based on a random forest regression algorithm using the fatigue life data obtained by simulation; and a real-time prediction module, used to input the actual bending deformation and current load state obtained in real time by the sensor detection module into the trained fatigue life prediction model, and the fatigue life prediction model outputs a predicted value of the remaining fatigue life of the traction machine spindle.

[0026] Furthermore, embodiments of the present invention provide a computer program product comprising a computer program that, when executed by a processor, implements the steps of the method described in any embodiment of the present invention. Embodiments of the present invention also provide a computer-readable storage medium storing a computer program thereon that, when executed by a processor, implements the steps of the method described in any embodiment of the present invention.

[0027] In addition, embodiments of the present invention also provide a computer device, comprising: a memory storing a computer program thereon; and a processor that executes the computer program to implement the steps of the method described in any embodiment of the present invention.

[0028] According to the above embodiments, the present invention has the following technical effects:

[0029] This invention addresses the impact of bending deformation on the fatigue life of a high-parameter elevator traction machine's main shaft. A digital model of the main shaft is constructed based on its structural and operational parameters. The accuracy of the model is assessed using actual bending data obtained from the main shaft via a laser displacement sensor and a three-point testing method. If the model is accurate, the constructed digital model is used to analyze the relationship between the shaft deformation and the fatigue life of the traction machine's main shaft. This determines the intrinsic connection between the operating state of the main shaft bearings and the deformation state of the traction machine's main shaft, acquiring fatigue life data under different bending deformation amounts. This data is then used to train a main shaft fatigue life prediction model based on a random forest regression algorithm. In real-time monitoring, the actual bending deformation of the traction machine's main shaft is obtained in real time using a laser displacement sensor and the three-point testing method. This data is then input into the prediction model to analyze and evaluate the fatigue life of the main shaft, achieving efficient and accurate prediction of the main shaft's fatigue life. This ensures the stable and reliable operation of the traction machine's main shaft, thereby guaranteeing the safe operation of the high-parameter elevator.

[0030] Various aspects, features, and advantages of the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Attached Figure Description

[0031] Figure 1 This is a schematic diagram illustrating the main process of a high-parameter elevator traction machine spindle fatigue life prediction method according to an embodiment of the present invention.

[0032] Figure 2 This is a schematic diagram illustrating the specific process of a high-parameter elevator traction machine spindle fatigue life prediction method according to an embodiment of the present invention.

[0033] Figure 3 This is a schematic diagram illustrating the flow of the random forest regression algorithm according to an embodiment of the present invention.

[0034] Figure 4This is a schematic diagram illustrating the hardware configuration upon which the three-point test method according to an embodiment of the present invention is based.

[0035] Figure 5 This is a block diagram illustrating a high-parameter elevator traction machine spindle fatigue life prediction system according to an embodiment of the present invention.

[0036] Figure 6 This is a structural block diagram illustrating a computer device according to an embodiment of the present invention. Detailed Implementation

[0037] Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the present invention can be implemented in various forms and is not limited to the specific embodiments described herein or shown in the accompanying drawings.

[0038] In this document, the terms “comprising,” “including,” and “having” are used to indicate the presence of certain features, steps, operations, elements, and / or components, but do not exclude the presence or addition of other features, steps, operations, elements, components, or combinations thereof. Unless explicitly stated in the context, the terms “first,” “second,” and similar terms do not indicate any priority or order, but are used only to distinguish different elements in the description.

[0039] Figure 1 The main flow of a high-parameter elevator traction machine spindle fatigue life prediction method according to an embodiment of the present invention is shown. In general, the high-parameter elevator traction machine spindle fatigue life prediction method of the present invention includes: S10, constructing a spindle digital model based on the structural and operating parameters of the traction machine spindle; S20, using a laser displacement sensor installed on the upper end cover of the traction machine to obtain the actual bending deformation of the spindle under at least one load state; S30, comparing the actual bending deformation with the simulated bending deformation obtained by the spindle digital model simulation to verify and correct the spindle digital model; S40, using the corrected spindle digital model to simulate and obtain fatigue life data of the traction machine spindle under different bending deformations; S50, using the fatigue life data obtained by simulation to train a fatigue life prediction model constructed based on a random forest regression algorithm; S60, using the laser displacement sensor to obtain the actual bending deformation and current load state of the traction machine spindle in real time; S70, inputting the real-time obtained actual bending deformation and current load state into the trained fatigue life prediction model, and having the fatigue life prediction model output a predicted value for the remaining fatigue life of the traction machine spindle.

[0040] For ease of explanation, the traction machine spindle will be simplified as a simply supported beam subjected to a concentrated load at its midpoint. For example... Figure 4As shown, the high-parameter elevator traction machine mainly includes a main shaft 401 and a traction sheave 402 mounted on the main shaft 401. The main shaft 401 is rotatably supported on a fixed base 405 and an upper end cover 406 via a first bearing 403 and a second bearing 404.

[0041] The main axis is set as a linearly elastic, homogeneous, isotropic circular cross-section beam with a diameter of d (m) and a span (the length between the first bearing 403 and the second bearing 404) of L (m). Analysis shows that the load F (N) mainly acts on the mid-span, producing a mid-point bending deformation δ (m); the load cycle is symmetrical, R = –1 (rotational bending).

[0042] When the spindle undergoes elastic deformation, the amount of bending deformation has the following relationship with the load:

[0043] (1)

[0044] (2)

[0045] In formulas (1) and (2), δ is the bending deformation of the traction machine spindle, F is the load on the spindle, E is the elastic modulus, I is the moment of inertia of the section, and d is the diameter of the spindle section.

[0046] From formulas (1) and (2), the spindle load F is:

[0047] (3)

[0048] Due to the mid-span bending moment of the traction machine main shaft:

[0049] (4)

[0050] Stress at the outermost edge of the cross section:

[0051] (5)

[0052] In formula (5), W is the torsional section modulus:

[0053] (6)

[0054] Substituting F back with δ, we obtain the stress-bending deformation relationship as follows:

[0055] (7)

[0056] (8)

[0057] In formulas (7) and (8), The stiffness-stress conversion factor is used, therefore the symmetrical cyclic stress amplitude is... Thus, the spindle fatigue life can be obtained from the Basquin equation:

[0058] (9)

[0059] In formula (9) denoted as the material fatigue strength coefficient, and b as the Basquin index.

[0060] As can be seen from formula (9), when the spindle undergoes elastic deformation, the fatigue life of the spindle is inversely proportional to the bending deformation by a power law, and the slenderness ratio L 2 The larger the / d value, the more sensitive the service life is to the amount of bending deformation.

[0061] When the traction machine spindle undergoes plastic deformation, low-cycle fatigue and stress relaxation / redistribution have already occurred, necessitating analysis using the strain-life method.

[0062] From the basic formula of Manson-Coffin:

[0063] (10)

[0064] In formula (10), For plastic strain amplitude (single-sided), is the fatigue ductility coefficient, and c is the fatigue ductility index.

[0065] permanent bending deformation δ of the spindle p The relationship with the maximum curvature in the beam is as follows:

[0066] (11)

[0067] The maximum plastic strain is:

[0068] (12)

[0069] For symmetrical cyclic loading, the plastic strain amplitude is:

[0070] (13)

[0071] Therefore, we get:

[0072] (14)

[0073] Therefore, the relationship between the plastic deformation of the traction machine spindle and its lifespan is as follows:

[0074] (15)

[0075] As can be seen from formula (15), plastic deformation is inversely proportional to fatigue life exponentially; under the Manson-Coffin framework, δp For every doubling of the original size, lifespan decreases to (1 / 2) of its original value. 1 / c The ratio is approximately 34% when c = -0.6.

[0076] In summary, when the traction machine spindle undergoes elastic or plastic deformation, the amount of spindle deformation is inversely proportional to the fatigue life of the spindle. Therefore, in this embodiment of the invention, the amount of elastic deformation of the spindle is measured when the high-parameter elevator is fully loaded to predict the fatigue life of the spindle in this state, and the amount of plastic deformation of the spindle is measured when the high-parameter elevator is unloaded to predict the fatigue life of the spindle in this state, thus realizing the prediction of the fatigue life of the spindle under different states. Therefore, in an exemplary embodiment, in S10, the digital model of the spindle is constructed by distinguishing whether the traction machine spindle is in the elastic deformation stage or the plastic deformation stage: for the elastic deformation stage, the fatigue life of the spindle is represented by formula (9); for the plastic deformation stage, the fatigue life of the spindle is represented by formula (15).

[0077] In some implementations, a digital model of the traction machine spindle is generated by acquiring its structural parameters (e.g., diameter, span, etc.) and operating parameters (e.g., load capacity, starting speed, running speed) and combining them with formulas (9) and (15). The digital model of the traction machine spindle is used to analyze the different bending deformations of the spindle under no-load and full-load conditions. The results are then compared with the bending deformations measured under the same actual working conditions to correct the constructed digital model and improve its analytical accuracy.

[0078] In some embodiments, at S20, a laser displacement sensor mounted on the upper end cover of the traction machine is used to obtain the actual bending deformation of the spindle under at least one load condition based on a three-point testing method. For example... Figure 4 As shown, the laser displacement sensor includes: laser displacement sensor A and laser displacement sensor B, respectively installed at the support points of the two bearings (i.e., the first bearing 403 and the second bearing 404) near the main shaft of the upper end cover 406 of the traction machine, and laser displacement sensor C installed at the center of the upper end cover of the traction machine. The three-point testing method includes: measuring the initial displacement value and subsequent displacement values ​​under the elevator's unloaded state during initial use using laser displacement sensors A, B, and C; and calculating the bending deformation of the main shaft based on the initial displacement value and the subsequent displacement values.

[0079] Since sensors A and B are installed on both sides of the main shaft near the two bearings, and because this location is close to the main shaft's fulcrum, the resulting main shaft deformation is relatively small. Therefore, points a and b on the main shaft measured by sensors A and B are taken as the zero point. Sensor C is installed at the center of the upper end cover. Because its measurement position is at the center between the two fulcrums of the traction machine's main shaft, the resulting main shaft deformation is the largest. Therefore, point c on the main shaft measured by sensor C is the point of maximum bending deformation. Specifically, the laser displacement sensor C measures the displacement of the traction sheave 402. Since the traction sheave 402 is mounted on the main shaft 401, the displacement at point c essentially reflects the displacement of the main shaft and can be considered as the displacement of the main shaft 401. The specific testing process is as follows:

[0080] First, when the traction machine spindle is used for the first time, under no-load conditions on a high-parameter elevator, the vertical displacement values ​​at points a, b, and c are measured, i.e., the initial displacement values ​​are a0, b0, and c0, respectively. In some implementations, during the detection, the spindle is rotated one revolution, causing the three sensors A, B, and C to measure the maximum vertical displacement values ​​corresponding to points a, b, and c. This is repeated three times, and the average of the maximum values ​​is taken, resulting in values ​​of a0, b0, and c0.

[0081] Secondly, during subsequent elevator use, the displacement values ​​of points a, b, and c under both no-load and full-load conditions of the main shaft are randomly detected. The displacement values ​​under no-load conditions are a1, b1, and c1, and the displacement values ​​under full-load conditions are a2, b2, and c2. In some implementations, each detection involves rotating the main shaft one revolution, allowing the three sensors A, B, and C to measure the maximum vertical displacement of points a, b, and c. This process is repeated three times, and the average of these maximum values ​​is taken as the final detection value.

[0082] Finally, the difference between the initial displacement value measured and the subsequent displacement values ​​is calculated, where the displacement differences under no-load conditions are as follows:

[0083] Δa1=a1-a0, Δb1=b1-b0, Δc1=c1-c0 (16)

[0084] The displacement differences under full load are as follows:

[0085] Δa2=a2-a0, Δb2=b2-b0, Δc2=c2-c0. (17)

[0086] The bending deformation at mid-span of the main shaft under no-load and full-load conditions was then calculated as follows:

[0087] When unloaded: δ1=Δb1-0.5╳(Δa1+Δc1) (18)

[0088] At full load: δ2=Δb2-0.5╳(Δa2+Δc2) (19)

[0089] Thus, the bending deformation of the spindle under no-load and full-load conditions can be obtained.

[0090] In this embodiment of the invention, on one hand, in S30, the spindle bending deformation obtained by the laser displacement sensor and the three-point test method is used to verify and correct the spindle digital model. On the other hand, in S70, when predicting the spindle fatigue life, the spindle bending deformation obtained in real time by the laser displacement sensor and the three-point test method is input into the fatigue life prediction model constructed based on the random forest regression algorithm (described later), and the prediction model outputs a predicted value for the remaining fatigue life of the traction machine spindle.

[0091] In some implementations, in S40, the step of using the modified spindle digital model to simulate and obtain fatigue life data of the traction machine spindle under different bending deformation amounts includes: the different bending deformation amounts are taken within a predetermined range with preset increments, and fatigue life data covering the elastic deformation and plastic deformation stages are obtained through the spindle digital model.

[0092] In some implementations, in S50, training the fatigue life prediction model based on the random forest regression algorithm using the fatigue life data obtained from simulation includes: analyzing the fatigue life of the traction machine main shaft under the elevator's no-load and full-load conditions using the fatigue life data obtained from simulation, and corresponding the analyzed fatigue life values ​​with each bending deformation to form a dataset; using the dataset, with bending deformation and load state as input features and fatigue life as output, to train the fatigue life prediction model.

[0093] In some implementations, the step of correlating the analyzed fatigue life values ​​with each bending deformation to form a dataset includes: classifying the fatigue life values ​​into initial, intermediate, and final stages, setting labels and corresponding bending deformation and load states to form a dataset.

[0094] In some implementations, the method further includes: dividing the dataset into a training set and a test set according to a predetermined ratio, and using stratified sampling to maintain data distribution consistency.

[0095] In some implementations, training the fatigue life prediction model using the dataset includes: extracting multiple training subsets with replacement from the training set using a bootstrap method, with each decision tree generated based on one training subset; when splitting at each node of a single decision tree, randomly selecting a predetermined number of candidate features from all input features, and selecting the optimal split from the predetermined number of candidate features to minimize the squared error; evaluating the fatigue life prediction model based on the data from the test set using the out-of-bag coefficient of determination and the root mean square error to determine the effectiveness of the prediction model. In some implementations, the predetermined number is the maximum number of features (max_feature).

[0096] In some implementations, the method further includes: performing hyperparameter optimization using a grid search method during training, wherein the optimized hyperparameters include at least: the number of decision trees (n_estimators), the maximum number of features (max_feature), the maximum depth of the tree (max_depth), the minimum number of samples required for node splitting (min_samples_split), and the minimum number of samples required for leaf nodes (min_samoles_leaf).

[0097] Specifically, such as Figure 2 As shown, the high-parameter elevator traction machine spindle fatigue life prediction method of the present invention includes: inputting the structural parameters and operating parameters of the traction machine spindle obtained through actual measurement, and generating a digital model of the traction machine spindle; analyzing the bending deformation of the traction machine spindle under no-load and full-load conditions using the digital model, and comparing the results with the bending deformation of the spindle measured under the same actual working conditions using a laser displacement sensor and a three-point test method to analyze the accuracy of the spindle digital model; when the analysis shows that the spindle digital model is inaccurate, adjusting the relevant parameters of the model to correct the spindle digital model; when the analysis shows that the spindle digital model is accurate, using the spindle digital model to perform elevator no-load and full-load operation. The system performs fatigue life analysis on the main shaft under different deformation amounts under different loads, including full load elevator operation, to obtain fatigue life data for the main shaft under different deformation amounts. This data is then used to generate training data labeled with the main shaft deformation amount. A random forest regression algorithm is used to construct a fatigue life prediction model based on the main shaft deformation amount, and the training data is used to train this model. Real-time deformation of the main shaft is acquired using a laser displacement sensor and a three-point testing method, and input into the trained fatigue life prediction model. This model enables rapid and accurate prediction of the fatigue life of the traction machine's main shaft, outputting the predicted value of the remaining fatigue life of the traction machine's main shaft.

[0098] In some implementations, training a traction machine spindle fatigue life prediction model based on a random forest regression analysis algorithm includes at least data preparation, generation of individual decision regression trees, ensemble prediction, and model evaluation.

[0099] (1) Preparation of training and testing data

[0100] Based on the analysis of the digital model of the traction machine spindle, fatigue life data of the spindle is obtained. This fatigue life data is then cleaned to obtain useful training and testing data. Each time training data is input, N data points are input, each containing an M-dimensional feature vector x and the corresponding spindle fatigue life target y. In data sampling, for the b-th tree (b∈(1,2,…B)), the Bootstrap method is applied to extract N subsets with replacement from the original data, obtaining a training subset D(b) for the b-th tree. Therefore, for a prediction model with B decision trees, the Bootstrap method is used to perform B samplings with replacement.

[0101] (20)

[0102] Where S b y is the sample index of the tree for lesson b, x is the feature vector of the traction machine spindle (including the spindle load value and the spindle bending deformation), y is the fatigue life of the traction machine spindle, and approximately 36.8% of the out-of-bag (OOB) samples do not participate in the training of this tree.

[0103] (2) Generation of a single decision regression tree

[0104] At the current node t, find the optimal partition. Minimize the squared error:

[0105] (twenty one)

[0106] Where R L R R The left and right sub-regions, The tree continues to split until the mean of the samples is met. The stopping conditions for node splitting are set as follows: tree depth ≥ max_depth (None (representing natural growth), 10-30); number of node samples ≤ min_samples_split (commonly 2–20); number of leaf node samples ≤ min_samples_leaf (commonly 1–8). Where max_depth is the maximum depth of the tree, min_samples_split is the minimum number of node samples required for splitting, and min_samples_leaf is the minimum number of leaf node samples required for splitting.

[0107] At each node, the best split is selected from only a random selection of max_feature candidate features to reduce inter-tree correlation. For example, in this implementation, the number of feature vectors is 2 (including the principal axis loading condition and principal axis deformation), so max_feature = 1 is set.

[0108] (3) Integrated prediction

[0109] A total of B trees are trained, where a larger B value indicates lower variance. The empirical value B ∈ [200, 1000]. For a new sample's feature vector x, the average value of the forest output from the B trees is:

[0110] (twenty two)

[0111] (4) Model Evaluation

[0112] Based on the training and prediction results, the prediction model is evaluated using the out-of-bag (OOB) determination coefficient R² and the root mean square error (RMSE).

[0113] (twenty three)

[0114] (twenty four)

[0115] In formulas (23) and (24), y i It is the true value of the i-th sample. It is the out-of-bag prediction value of the i-th sample. It is the average of all true values, and N is the total number of samples.

[0116] The following is combined Figure 3 Exemplary embodiments of the present invention will be described.

[0117] (I) Construction of digital models

[0118] During the initial operation phase of a high-parameter elevator, the structural and operational parameters of its traction machine spindle are acquired. Based on the acquired parameters, a digital model of the corresponding traction machine spindle is quickly generated.

[0119] (II) Detection of Deformation of Traction Machine Main Shaft Based on Laser Displacement Sensor

[0120] By installing laser displacement sensors on the end caps of the traction machine, during the initial operation phase of the high-parameter elevator, the test points of the traction machine main shaft under no-load and full-load conditions are detected and the initial displacement values ​​of each point are recorded. In the later operation of the elevator, the displacement values ​​of each point of the high-parameter elevator under different conditions are obtained in real time by laser displacement sensors. The bending deformation of the main shaft under the corresponding conditions is calculated by formulas (16)-(19).

[0121] (III) Data Acquisition

[0122] Based on the generated digital model of the traction machine spindle and formulas (9) and (15), the bending deformation of the traction machine spindle under no-load and full-load conditions is analyzed. For example, the spindle deformation is set in the range of 0.001-0.1mm, and 100 spindle deformation values ​​are set in increments of 0.001mm. The analysis and testing are carried out under no-load and full-load conditions of the traction machine spindle to obtain 200 spindle fatigue life data. At the same time, the fatigue life level of the traction machine spindle is divided into initial stage (use times ≥10). 6 times), mid-term (10 3 ≤Number of uses<10 6 (Number of uses <10) End of period (number of uses <10) 3 The spindle fatigue life values ​​obtained from the analysis are categorized into three stages (times), and labeled with corresponding spindle deformation and loading conditions to form a dataset. It should be understood that the specific values ​​listed herein are merely examples and do not constitute a limitation of the invention. Depending on the specific application, other values ​​may be used.

[0123] (iv) Construction of the prediction model

[0124] The obtained dataset is divided into a training set and a test set, with a 7:3 hierarchical split (140 records for the training set and 60 records for the test set). The random seed is fixed at 2024 to ensure repeatability. It should be understood that the specific values ​​listed herein are merely examples and do not constitute a limitation of the invention. Depending on the specific application, other values ​​may be used.

[0125] (v) Training of the prediction model

[0126] (1) Initially set the relevant parameters of the model. The number of trees in the forest, n_estimators (i.e. B value), is 500. The maximum number of features to be considered during each split is max_feature, which is 1. The maximum depth of each decision tree is max_depth (let the trees grow naturally first). The minimum number of samples required for splitting is min_samples_split, which is 5. The minimum number of samples required for splitting is min_samoles_leaf, which is 2.

[0127] (2) In the data replacement sampling, 500 samplings are performed based on B being 500 and formula (20), and it is ensured that approximately 36.8% of the out-of-bag (OOB) samples do not participate in the tree training.

[0128] (3) In the generation of a single decision regression tree, at each node, the best split is selected from only 1 (max_feature=1) randomly selected candidate features, and the best split is found according to formula (21). The splitting process continues until the tree depth, the number of node samples, and the number of leaf node samples meet the initial set values, minimizing the squared error.

[0129] (4) Training and out-of-bag validation: In prediction training, monitoring is performed using out-of-bag (OOB) samples, and OOB-R is performed using formula (23). 2 If the value is close to 1, then there is no need to add more trees.

[0130] (5) Hyperparameter mesh fine-tuning and cross-validation: The mesh range is set as follows: n_estimators∈{300,500, 800}, max_features∈{1, 2}, max_depth∈{None, 15, 25}, min_samples_split∈{2, 5, 10}. The negative root mean square error NRSME (NRSME=-RSME) of the prediction results is calculated by formula (24), and the optimal combination of n_estimators, max_feature, max_depth, min_samples_split, and min_samoles_leaf is finally obtained.

[0131] (6) Final model training and evaluation: Using the entire training set, retrain the model with the obtained optimal parameters to obtain the final fatigue life prediction model based on the random forest regression algorithm.

[0132] (vi) Evaluation of the predictive model test

[0133] During the model testing phase, the obtained prediction model is evaluated based on the relevant data of the test set and according to formulas (23) and (24) to determine the effectiveness of the constructed model.

[0134] (vii) Prediction of fatigue life of traction machine spindle

[0135] The displacement of each test point of the monitored traction machine spindle is obtained by laser displacement sensor. The bending deformation of the spindle under the actual state is calculated by formula (16)-(19). The obtained deformation and the load condition of the spindle are input into the fatigue life prediction model. The output of the prediction model is the prediction of the fatigue life of the high-parameter elevator traction machine spindle.

[0136] Based on the high-parameter elevator traction machine spindle fatigue life prediction method of the present invention, the present invention also provides a high-parameter elevator traction machine spindle fatigue life prediction system, which is used to implement the method described in any one of the embodiments of the present invention. In an exemplary embodiment, the system includes: a digital modeling module 501, used to construct a digital model of the main shaft based on the structural and operational parameters of the traction machine main shaft; a sensor detection module 502, including a laser displacement sensor installed on the upper end cover of the traction machine, used to detect the actual bending deformation of the main shaft under at least one load state; a digital model verification module 503, used to compare the actual bending deformation with the simulated bending deformation obtained by simulating the main shaft digital model to verify and correct the main shaft digital model; a data analysis module 504, used to use the corrected main shaft digital model to simulate and obtain fatigue life data of the traction machine main shaft under different bending deformations; a prediction model training module 505, used to train a fatigue life prediction model constructed based on a random forest regression algorithm using the fatigue life data obtained by simulation; and a real-time prediction module 506, used to input the actual bending deformation and current load state obtained in real time by the sensor detection module 502 into the trained fatigue life prediction model, and the fatigue life prediction model outputs a predicted value for the remaining fatigue life of the traction machine main shaft.

[0137] It should be understood that the steps, processes, and operations described in the above embodiments can be implemented by computer software programs. Therefore, the embodiments of the present invention also relate to a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the method or the processes and operations of the system described in any embodiment of the present disclosure. In some embodiments, program code for executing the computer program of the embodiments of the present invention can be written in any combination of one or more programming languages. For example, the computer program can be written using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on a personal computing device, partially on a personal computing device, partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the personal computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0138] It should also be understood that the computer program product may be stored on a computer-readable storage medium, such as hard disks, floppy disks, magnetic tapes, optical disks, solid-state drives, flash memory, etc.

[0139] Furthermore, it can be understood that the method of the present invention can be executed by a computer device. Therefore, embodiments of the present invention also relate to a computer device comprising: a memory storing a computer program thereon; and a processor that, when executing the computer program, implements the steps (processing, operations, etc.) of the method described in any embodiment of the present disclosure.

[0140] Figure 6 An exemplary structure of a computer device is shown. Figure 6 As shown, the computer device 100 may include at least a processor 101 and a memory 102, wherein the memory 102 may be or include a storage medium storing a computer program (or computer-readable instructions) thereon, and the processor 101 executes the computer program to perform some or all of the operations described in any embodiment of the present disclosure.

[0141] In some embodiments, the computer device 100 may further include a data storage device 103, a display 104, a speaker 105, and a communication module 106. The processor 101, memory 102, and data storage device 103 communicate with each other via a bus and interact with peripheral devices, such as the display 104 and communication module 106, via the bus and I / O module 107.

[0142] Specifically, data storage device 103 can store application programs and their various configuration files and data. Computer programs are stored on memory 102. Processor 101 executes the computer programs to perform various processes, operations, or steps based on the configuration center-based interface simulation method. Display 104 can be used to display various data, images, etc.

[0143] In optional embodiments, the communication module 106 may be omitted where communication with the other systems, devices, or equipment is not required. In some embodiments, the computer device 100 may further include a speaker 105 for voice output.

[0144] In some implementations, processor 101 may include any suitable semiconductor-based electronic processing unit, chip, microchip, or integrated circuit (IC). Memory 102 is a programmable memory, which may include any suitable electronic storage device configured to store instructions and be reprogrammable. For example, programmable memory may include an erasable programmable read-only memory (EPROM) device. Data storage device 103 may be a persistent storage device, which may include any suitable electronic storage device configured to retain stored information as power cycles. For example, data storage device 103 may include a hard disk drive, solid-state drive (SSD), flash memory drive, hybrid drive, etc., or any combination thereof.

[0145] Therefore, processor 101 can control the output of information on I / O module 107 by storing information in memory and / or executing programs / instructions stored in memory. For example, several aspects of the methods described herein can be executed by processor 101 according to programs / instructions stored in memory (e.g., memory 102 and / or data storage device 103).

[0146] Furthermore, the processor 101 can electronically communicate with the I / O module 107 and / or the communication module 106 to receive or send relevant instructions and information. The I / O module 107 may include any suitable mechanical or virtual user interface configured to enable a user to interact with the computer device 100 or to allow the user to perform one or more functions of the computer device 100 itself, such as a graphical user interface (GUI) on a screen or other display. In some instances, the user interface may include a voice interface capable of speech recognition, through which an operator can provide voice commands to the processor.

[0147] The communication module 106 may include any suitable devices and / or structures configured to facilitate information exchange between the computer device 100 and external electronic devices (e.g., sensor detection modules). The communication module 106 may include devices configured to send and / or receive wireless or wired information with other devices. For example, the communication module 106 may include an antenna, a transceiver, a connector for wired data reception and / or transmission, a data exchange device, etc., or any combination thereof. The communication module 106 may also include ancillary components, such as filtering circuitry, encryption / decryption circuitry, and / or integrated circuit (IC) chips (e.g., Bluetooth® chips) for processing signals. In some embodiments, the communication module 106 may include a WiFi device configured to connect to a local wireless network.

[0148] In some embodiments, computer device 100 may include smartphones, wearable computers, portable / mobile electronic devices, tablet computers, smartwatches, personal digital assistants (PDAs), personal computers (PCs), desktop computers, laptop computers, servers, and so on. Computer device 100 may include or have installed one or more application programs (APPs), one of which is configured to perform the dynamic data service system and methods described herein.

[0149] Although not shown, it should be understood that the computer device 100 also includes a power supply component, which may include any suitable device and / or structure configured to provide an electrical interface between the computer device 100 and a power source. The power source may include any suitable source of electrical energy, such as a battery, socket, capacitor, fuel cell, etc., or any combination thereof. Additionally or optionally, the power source may be included within the power supply component. For example, a battery or battery pack may be included within the computer device 100. In some embodiments, the battery may be rechargeable, for example, charged via a cable or an interface provided by the power supply component. In some embodiments, the power supply component may share features with or repeat features of the communication module 106. For example, a USB or micro-USB cable connector may be included in the computer device 100, so that power supply or data communication is performed by the same component.

[0150] Those skilled in the art should understand that the above disclosure is merely illustrative of embodiments of the present invention, and the scope of patent protection claimed in this application is not limited thereto. Various modifications, alterations, substitutions, and other changes can be made to the disclosed embodiments without departing from the spirit of this disclosure, and such changes are within the scope covered by the claims of this application.

Claims

1. A method for predicting the fatigue life of a high-parameter elevator traction machine spindle, characterized in that, include: Based on the structural and operational parameters of the traction machine spindle, a digital model of the spindle is constructed. Using a laser displacement sensor installed on the upper end cover of the traction machine, the actual bending deformation of the spindle under at least one load condition is obtained, and the actual bending deformation is compared with the simulated bending deformation obtained by the spindle digital model simulation in order to verify and correct the spindle digital model. Using the modified digital model of the main shaft, fatigue life data of the traction machine main shaft under different bending deformations were obtained through simulation. The fatigue life prediction model based on the random forest regression algorithm was trained using the fatigue life data obtained from the simulation. The laser displacement sensor is used to obtain the actual bending deformation and current load status of the traction machine spindle in real time. The actual bending deformation and current load status acquired in real time are input into the trained fatigue life prediction model, and the fatigue life prediction model outputs the predicted value of the remaining fatigue life of the traction machine spindle. Specifically, the digital model of the traction machine spindle is constructed by distinguishing whether the spindle is in the elastic deformation stage or the plastic deformation stage, wherein: For the elastic deformation stage, the fatigue life of the main shaft is represented by the Basquiat equation, as follows: N f Indicates the spindle fatigue life. denoted as the material fatigue strength coefficient, b as the Basquiat index, L as the main shaft span, d as the main shaft diameter, δ as the bending deformation of the main shaft, and E as the elastic modulus. During the plastic deformation stage, the fatigue life of the principal shaft is represented by the Manson-Coffey equation, as follows: N f Indicates the spindle fatigue life, δ p This indicates the amount of permanent bending deformation of the spindle. is the fatigue ductility coefficient, c is the fatigue ductility index, L represents the spindle span, and d represents the spindle diameter.

2. The method for predicting the fatigue life of a high-parameter elevator traction machine spindle according to claim 1, characterized in that, The laser displacement sensor includes: a first laser displacement sensor and a second laser displacement sensor respectively installed at the two bearing support points of the main shaft near the upper end cover of the traction machine, and a third laser displacement sensor installed at the center of the upper end cover of the traction machine. The actual bending deformation of the spindle is obtained using a three-point test method, including: The first laser displacement sensor, the second laser displacement sensor, and the third laser displacement sensor are used to measure the initial displacement value when the elevator is unloaded during its first use and the displacement value detected subsequently. The bending deformation of the spindle is calculated using the following formula: δ = Δb - 0.5 × (Δa + Δc) Where δ is the bending deformation of the principal axis, Δa represents the difference between the displacement value currently detected by the first laser displacement sensor and the corresponding initial displacement value, Δb represents the difference between the displacement value currently detected by the second laser displacement sensor and the corresponding initial displacement value, and Δc represents the difference between the displacement value currently detected by the third laser displacement sensor and the corresponding initial displacement value.

3. The method for predicting the fatigue life of a high-parameter elevator traction machine spindle according to claim 1, characterized in that, The step of using the modified digital model of the main shaft to simulate and obtain fatigue life data of the traction machine main shaft under different bending deformations includes: The different bending deformation amounts are taken within a predetermined range with preset increments, and fatigue life data covering the elastic deformation and plastic deformation stages are obtained through the main shaft digital model.

4. The method for predicting the fatigue life of a high-parameter elevator traction machine spindle according to claim 3, characterized in that, Training the fatigue life prediction model based on the random forest regression algorithm using the fatigue life data obtained from simulation includes: The fatigue life of the traction machine main shaft under elevator no-load and full-load conditions is analyzed using the fatigue life data obtained from the simulation. The fatigue life values ​​obtained from the analysis are correlated with each bending deformation to form a dataset. Using the dataset, with bending deformation and load state as input features and fatigue life as output, the fatigue life prediction model is trained.

5. The method for predicting the fatigue life of a high-parameter elevator traction machine spindle according to claim 4, characterized in that, The dataset, which is constructed by correlating the obtained fatigue life values ​​with various bending deformations, includes: The fatigue life values ​​are classified into initial, middle and final stages, and labeled with corresponding bending deformation and load states to form a dataset.

6. The method for predicting the fatigue life of a high-parameter elevator traction machine spindle according to claim 4, characterized in that, Also includes: The dataset is divided into training and testing sets according to a predetermined ratio, and stratified sampling is used to maintain data distribution consistency.

7. The method for predicting the fatigue life of a high-parameter elevator traction machine spindle according to claim 6, characterized in that, Training the fatigue life prediction model using the dataset includes: Multiple training subsets are extracted from the training set with replacement using a bootstrap method, and each decision tree is generated based on one training subset. When splitting at each node of a single decision tree, a predetermined number of candidate features are randomly selected from all input features, and the best split is selected from the predetermined number of candidate features to minimize the squared error. Based on the data from the test set, the fatigue life prediction model is evaluated using the out-of-bag determination coefficient and root mean square error to determine the effectiveness of the prediction model.

8. The method for predicting the fatigue life of a high-parameter elevator traction machine spindle according to claim 7, characterized in that, The method further includes: using a grid search method to optimize hyperparameters during training, and the optimized hyperparameters include at least: the number of decision trees, the maximum number of features, the maximum depth of the tree, the minimum number of samples required for node splitting, and the minimum number of samples required for leaf nodes.

9. A fatigue life prediction system for a high-parameter elevator traction machine spindle, used to implement the fatigue life prediction method for a high-parameter elevator traction machine spindle as described in any one of claims 1-8, characterized in that, include: The digital modeling module is used to construct a digital model of the main shaft based on the structural and operational parameters of the traction machine main shaft; The sensor detection module includes a laser displacement sensor installed on the upper end cover of the traction machine, used to detect the actual bending deformation of the main shaft under at least one load condition; The digital model verification module is used to compare the actual bending deformation with the simulated bending deformation obtained from the simulation of the spindle digital model, so as to verify and correct the spindle digital model. The data analysis module is used to simulate and obtain fatigue life data of the traction machine spindle under different bending deformations using the modified spindle digital model. The prediction model training module is used to train the fatigue life prediction model based on the random forest regression algorithm using the fatigue life data obtained from the simulation. The real-time prediction module is used to input the actual bending deformation and current load state obtained in real time by the sensor detection module into the trained fatigue life prediction model, and the fatigue life prediction model outputs the predicted value of the remaining fatigue life of the traction machine spindle.

10. A computer program product comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the high-parameter elevator traction machine spindle fatigue life prediction method according to any one of claims 1 to 8.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the high-parameter elevator traction machine spindle fatigue life prediction method according to any one of claims 1 to 8.

12. A computer device, characterized in that, include: A memory on which computer programs are stored; A processor, when executing the computer program, implements the steps of the high-parameter elevator traction machine spindle fatigue life prediction method according to any one of claims 1 to 8.

Images