A method and system for predicting the residual life of a shearer drum

Abstract

The application provides a coal cutter drum residual life prediction method and system, the method comprises the following steps: collecting drum multi-source operation data and physical design data; wherein the drum comprises pick; analyzing the characteristics of the multi-source operation data, and based on different coal and rock media and cutting parameters, a pick wear rate sensitivity model is established; based on a deep self-adaptive fuzzy clustering algorithm, the multi-source operation data is processed to construct a drum data-driven model; based on the physical design data, a drum mechanism model is established; a dynamic Bayesian network is used to establish a drum full life cycle digital twin; a drum fault monitoring model is obtained; based on the drum fault monitoring model, the drum fault data is analyzed, a drum life prediction curve is established, and the residual life of the drum is predicted. The technical scheme of the application provides a theoretical basis for realizing drum state recognition and full life cycle prediction.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of fault monitoring, and particularly relates to a coal mining machine drum residual life prediction method and system. BACKGROUND

[0002] Real-time performance is one of the key indicators of a fault monitoring system. However, there may be problems such as data processing delay and slow analysis speed in the prior art, which leads to insufficient timely fault warning and inability to effectively avoid production interruption caused by sudden failure. The prediction and early warning capability for potential faults also needs to be improved. The prior art may not accurately predict the future health status or potential failure risk of the drum, resulting in inaccurate maintenance plan. The fault monitoring system in the prior art has problems such as low system integration and poor compatibility between components. This may lead to problems such as decreased overall system performance and increased maintenance costs. As the most important cutting component of the fully mechanized coal mining face, the intelligent perception and intelligent decision of the drum failure are an important part of measuring the coal mining efficiency and evaluating the intelligent level. SUMMARY

[0003] In view of the deficiencies of the prior art, the application provides a coal mining machine drum residual life prediction method and system, which can monitor the running state of the drum in real time and improve the coal mining efficiency.

[0004] To achieve the above-mentioned purpose, the application provides the following solutions:

[0005] A coal mining machine drum residual life prediction method, comprising the following steps:

[0006] Collecting multi-source running data and physical design data of the drum; wherein the drum comprises a cutting pick;

[0007] Analyzing the characteristics of the multi-source running data, and establishing a cutting pick wear rate sensitivity model based on different coal and rock media and cutting parameters;

[0008] Processing the multi-source running data based on a deep adaptive fuzzy clustering algorithm, and constructing a drum data-driven model;

[0009] Establishing a drum mechanism model based on the physical design data;

[0010] Based on the matching relationship between the drum data-driven model and the drum mechanism model, a drum full life cycle digital twin is established by using a dynamic Bayesian network;

[0011] Integrating the cutting pick wear rate sensitivity model into the drum full life cycle digital twin to obtain a drum fault monitoring model;

[0012] Based on the drum fault monitoring model, analyzing drum failure data, establishing a drum life prediction curve, and predicting the residual life of the drum.

[0013] Preferably, the multi-source operation data includes cutting parameters, cutting vibration data, temperature data, current and voltage data, cutting force data, acoustic emission data, infrared thermal imaging data, and operation time and cycle data.

[0014] The cutting parameters include cutting speed, cutting depth, and pick angle.

[0015] The physical design data includes drum diameter and length data, spiral blade parameters, pick arrangement data, material parameters, transmission system design parameters, and bearing and sealing design parameters.

[0016] Preferably, the method for establishing the pick wear rate sensitivity model is as follows:

[0017] Based on different cutting parameters and coal rock media, corresponding cutting vibration data, acoustic emission data, and temperature data are obtained.

[0018] The cutting vibration data and the acoustic emission data corresponding to different cutting parameters and coal rock media are numerically simulated to obtain boundary conditions.

[0019] Based on the boundary conditions, drum vibration energy spectrum maps and acoustic emission energy spectrum maps are reconstructed.

[0020] Based on the drum vibration energy spectrum maps and the acoustic emission energy spectrum maps, frequency spectrum features are obtained; the frequency spectrum features include vibration data frequency spectrum features and acoustic emission data frequency spectrum features.

[0021] The temperature data corresponding to different cutting parameters and coal rock media are numerically simulated to obtain flash temperature distributions.

[0022] In combination with the infrared thermal imaging data and the flash temperature distributions, temperature and frequency distribution maps are obtained; the frequency is the relative frequency or probability of temperature falling within a certain specific interval.

[0023] Based on the temperature and frequency distribution maps, temperature features are obtained.

[0024] Based on the frequency spectrum features, the temperature features, and the pick wear rate, a pick wear rate sensitivity model is established.

[0025] Preferably, based on the frequency spectrum features, the temperature features, and the pick wear rate, a specific method for establishing the pick wear rate sensitivity model is as follows:

[0026] Based on the frequency spectrum features and the corresponding pick wear degree and the temperature features and the corresponding pick wear degree, a multi-source signal feature-pick wear singularity influence curve is obtained.

[0027] Obtain an influence curve of different cutting parameters and coal rock medium on the wear rate of the cutting pick based on the wear degree of the cutting pick corresponding to different cutting parameters and the coal rock medium;

[0028] Analyze a first correlation coefficient of the wear singularity influence curve and the wear rate influence curve by using a Spearman rank correlation coefficient;

[0029] Obtain a second correlation coefficient of the spectrum feature and the temperature feature and the wear rate influence curve based on the first correlation coefficient;

[0030] Assign an influence weight to the spectrum feature and the temperature feature based on the second correlation coefficient;

[0031] Establish a cutting pick wear rate sensitivity model based on the influence weight and the wear rate influence curve.

[0032] Preferably, the method for constructing the drum data-driven model is as follows:

[0033] Normalize the multi-source operation data, and reduce the dimension of the normalized multi-source operation data to obtain low-dimensional multi-source operation data;

[0034] Construct a sparse auto-encoder layer, and the sparse auto-encoder layer learns a low-dimensional representation of the low-dimensional multi-source operation data by minimizing a reconstruction error to obtain a trained sparse auto-encoder layer;

[0035] Stack a preset number of trained sparse auto-encoder layers to obtain a deep network;

[0036] Obtain low-dimensional features of the multi-source operation data based on the deep network;

[0037] Cluster the low-dimensional features based on a deep adaptive fuzzy clustering algorithm;

[0038] Obtain compressed multi-source operation data based on the low-dimensional feature clustering result;

[0039] Construct the drum data-driven model based on the compressed multi-source operation data.

[0040] The application also provides a coal cutter drum residual life prediction system for implementing the method, comprising:

[0041] A data acquisition module is configured to acquire multi-source operation data and physical design data of a drum, wherein the drum comprises a cutting pick;

[0042] A cutting pick wear rate sensitivity model construction module is configured to analyze the multi-source operation data features, and establish a cutting pick wear rate sensitivity model based on different coal rock media and cutting parameters;

[0043] A data-driven model construction module is configured to process the multi-source operation data based on a deep adaptive fuzzy clustering algorithm, and construct a drum data-driven model.

[0044] A mechanism model construction module is configured to establish a drum mechanism model based on the physical design data.

[0045] A digital twin module is configured to establish a drum full-life-cycle digital twin based on a matching relationship between the drum data-driven model and the drum mechanism model, and adopt a dynamic Bayesian network.

[0046] A fault monitoring and early warning module is configured to integrate the cutting tooth wear rate sensitivity model into the drum full-life-cycle digital twin, and obtain a drum fault monitoring model.

[0047] A life prediction module is configured to analyze drum fault data based on the drum fault monitoring model, establish a drum life prediction curve, and predict the remaining life of the drum.

[0048] Preferably, the method further comprises a life prediction module configured to analyze drum fault data based on the drum fault monitoring model, establish a drum life prediction curve, and predict the remaining life of the drum.

[0049] Compared with the prior art, the present application has the following beneficial effects: the present application integrates multi-models into a fault monitoring model by collecting and analyzing multi-source operation data of a drum of a coal mining machine, not only considers the wear rate of a cutting tooth, but also establishes a drum full-life-cycle digital twin, so that fault monitoring is more accurate and more efficient, and the service life of the drum is prolonged. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the present application, the following briefly introduces the drawings needed to be used in the embodiments, and obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0051] Figure 1 A flow chart of a drum remaining life prediction method of a coal mining machine according to an embodiment of the present application;

[0052] Figure 2 A schematic diagram of a drum fault monitoring model construction according to an embodiment of the present application. DETAILED DESCRIPTION

[0053] With reference to the accompanying drawings: the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0054] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] Embodiment one

[0056] As shown in Figure 1 , Figure 2 , a method for predicting the residual life of a shearer drum, comprising the following steps:

[0057] S1: collecting drum multi-source operation data and physical design data; wherein the drum includes pick; further embodiments are that the multi-source operation data includes cutting parameters, cutting vibration data, temperature data, current and voltage data, cutting force data, acoustic emission data, infrared thermal imaging data, and operation time and cycle data; in the process of cutting coal rock by the drum, the generation of acoustic emission signals is closely related to the breaking process of coal rock. Therefore, the data collected by the acoustic emission sensor can be used to analyze the interaction between the drum and coal rock, and the potential damage inside the drum.

[0058] Wherein the cutting parameters include cutting speed, cutting depth and pick angle;

[0059] The physical design data includes drum diameter and length data, spiral blade parameters, pick arrangement data, material parameters, transmission system design parameters, and bearing and sealing design parameters.

[0060] S2: analyze the characteristics of multi-source operation data, and establish a pick wear rate sensitivity model based on different coal rock media and cutting parameters; the pick wear rate sensitivity model is used to predict the influence of different influencing factors on the pick wear rate.

[0061] Further embodiments are that the method for establishing the pick wear rate sensitivity model is:

[0062] Based on different cutting parameters and coal rock media, the corresponding cutting vibration data, acoustic emission data and temperature data are obtained;

[0063] numerical simulation is performed on the cutting vibration data and acoustic emission data corresponding to different cutting parameters and coal rock media to obtain boundary conditions; in this embodiment, the boundary conditions include: force conditions of the drum and the cutting pick during the cutting process; breaking mode and stress distribution of the coal rock; propagation path and intensity distribution of the vibration and acoustic emission.

[0064] Based on the boundary conditions, the drum vibration energy spectrum graph and the acoustic emission energy spectrum graph are reconstructed;

[0065] Based on the drum vibration energy spectrum graph and the acoustic emission energy spectrum graph, the frequency spectrum features are obtained; wherein the frequency spectrum features include vibration data frequency spectrum features and acoustic emission data frequency spectrum features; the preprocessed vibration data is subjected to frequency spectrum analysis by using the fast Fourier transform (FFT) method to obtain the distribution of the vibration energy at different frequencies; the energy of each acoustic emission event is calculated, the acoustic emission waveform is subjected to energy measurement by using the integral method, and the energy of the acoustic emission event is distributed to different frequencies to obtain the acoustic emission energy spectrum graph.

[0066] Numerical simulation is performed on the temperature data corresponding to different cutting parameters and coal rock media to obtain flash temperature distribution;

[0067] The temperature and frequency distribution graph is obtained by combining the infrared thermal imaging data and the flash temperature distribution; wherein the frequency is the relative frequency or probability of the temperature falling in a certain specific interval;

[0068] Based on the temperature and frequency distribution graph, the temperature features are obtained;

[0069] Based on the frequency spectrum features (amplitude, energy distribution, frequency spectrum entropy), the temperature features (average temperature, temperature gradient, flash temperature distribution) and the pick wear rate, a pick wear rate sensitivity model is established.

[0070] Further embodiments are that, based on the frequency spectrum features, the temperature features and the pick wear rate, the specific method for establishing the pick wear rate sensitivity model is:

[0071] Based on the frequency spectrum features and the corresponding pick wear degree and the temperature features and the corresponding pick wear degree, the multi-source signal feature-pick wear singularity influence curve is obtained; in this embodiment, all feature data (frequency spectrum features, temperature features, wear degree) are aligned in time and subjected to standardization processing, so that features of different dimensions can be compared and analyzed on the same scale.

[0072] Based on the wear degree data at continuous time points, the pick wear rate is calculated;

[0073] By anomaly detection, the singularity point of sudden increase in the wear rate is identified;

[0074] Obtaining the contribution degree of the multi-source signal features (spectrum features and temperature features) to the wear singularity of the cutting tooth by analyzing the changes of the spectrum features and the temperature features near the singularity point;

[0075] Obtaining the wear singularity influence curve of the multi-source signal features on the cutting tooth based on the contribution degree.

[0076] Obtaining the wear rate influence curve of different cutting parameters and coal rock medium on the cutting tooth based on the wear degrees of the cutting tooth corresponding to different cutting parameters and coal rock medium;

[0077] Analyzing the first correlation coefficient of the wear singularity influence curve and the wear rate influence curve by using the Spearman rank correlation coefficient;

[0078] Obtaining the second correlation coefficient of the spectrum features and the temperature features and the wear rate influence curve based on the first correlation coefficient;

[0079] Assigning the influence weight to the spectrum features and the temperature features based on the second correlation coefficient;

[0080] Establishing a cutting tooth wear rate sensitivity model based on the influence weight and the wear rate influence curve.

[0081] S3: processing multi-source operation data based on a deep adaptive fuzzy clustering algorithm, and constructing a roller data-driven model;

[0082] S4: establishing a roller mechanism model based on physical design data;

[0083] Further implementation is that the method for constructing the roller data-driven model is:

[0084] Normalizing the multi-source operation data, and reducing the dimension of the normalized multi-source operation data to obtain low-dimensional multi-source operation data;

[0085] Constructing a sparse auto-encoder layer, the sparse auto-encoder layer learns the low-dimensional representation of the low-dimensional multi-source operation data by minimizing the reconstruction error, and obtains a trained sparse auto-encoder layer; during the training process, a sparsity constraint is added to encourage the average activation of the hidden units to approach a preset sparsity level.

[0086] Stacking a preset number of trained sparse auto-encoder layers to obtain a deep network;

[0087] Obtaining the low-dimensional features of the multi-source operation data based on the deep network;

[0088] Clustering the low-dimensional features based on the deep adaptive fuzzy clustering algorithm;

[0089] Obtaining compressed multi-source operation data based on the low-dimensional feature clustering result;

[0090] Based on the compressed multi-source operation data, a drum data-driven model is constructed.

[0091] In this embodiment, according to the extracted low-dimensional features, the fuzzy clustering centers are initialized by K-means++, for each data point, the fuzzy membership of each cluster is calculated according to the distance between the data point and the clustering center, the clustering weight is obtained based on the fuzzy membership, and the clustering weight is used to update the center of each cluster according to the fuzzy membership and the features of the data point. Repeat the steps of calculating the fuzzy membership and updating the clustering center until the change of the clustering center is less than the preset threshold or the preset iteration number is reached. Each data point is assigned to the corresponding cluster according to its maximum fuzzy membership, and a cluster label is assigned to it. For each cluster, only the data of its clustering center is retained as a representative, thereby greatly reducing the data storage amount. The clustering statistical information (such as mean, variance, median, etc.) is calculated and stored, instead of storing all data points. When new data arrives, first determine which cluster it belongs to, then only update the statistical information or clustering center of that cluster, instead of the entire data set. According to the clustering label and the stored clustering center or statistical information, an approximate representation of the original data is reconstructed.

[0092] S5: based on the matching relationship between the drum data-driven model and the drum mechanism model, a dynamic Bayesian network is used to establish a drum full life cycle digital twin;

[0093] In this embodiment, the method of constructing a dynamic Bayesian network includes: first, define DBN nodes, each node represents a state or variable in the life cycle of the drum, such as vibration intensity, temperature, wear degree, cutting efficiency, etc. The directed edge represents the causal relationship or conditional dependence relationship between the nodes. According to the mechanism model and operation data of the drum, the connection mode and direction between the nodes are determined. Based on the drum mechanism model, define the conditional probability table for each node to describe the probability distribution of the node under different parent node states. The life cycle of the drum is divided into multiple time slices, each time slice corresponds to a time layer of the DBN. The nodes between the time layers are connected through time transition probabilities to capture the dynamic changes of the drum state. Analyze the matching relationship between the data-driven model and the mechanism model to determine which data or features are consistent in both models and which need to be aligned through calibration. Use the actually obtained multi-source operation data to calibrate the parameters in the DBN to ensure that the model can accurately reflect the actual state and behavior of the drum, and complete the construction of the dynamic Bayesian network.

[0094] Finally, based on the constructed dynamic Bayesian network, the construction and update of the digital twin are carried out: according to the physical design parameters of the drum and the initial running state, the initial state of the DBN is set, and the real-time collected multi-source operation data is input into the DBN to drive the dynamic update of the digital twin during the operation of the drum.

[0095] S6: integrate the pick wear rate sensitivity model into the drum full life cycle digital twin to obtain a drum fault monitoring model;

[0096] S7: based on the drum fault monitoring model, analyze the drum fault data, establish a drum life prediction curve, and predict the remaining life of the drum.

[0097] Example two

[0098] The application also provides a coal cutter drum remaining life prediction system for realizing the method, comprising:

[0099] A data acquisition module is configured to acquire drum multi-source operation data and physical design data; wherein the drum comprises a pick.

[0100] A pick wear rate sensitivity model construction module is configured to analyze multi-source operation data characteristics, and establish a pick wear rate sensitivity model based on different coal and rock media and cutting parameters.

[0101] A data-driven model construction module is configured to process multi-source operation data based on a deep adaptive fuzzy clustering algorithm, and construct a drum data-driven model.

[0102] A mechanism model construction module is configured to establish a drum mechanism model based on physical design data.

[0103] A digital twin module is configured to establish a drum full life cycle digital twin based on the matching relationship between the drum data-driven model and the drum mechanism model, and adopt a dynamic Bayesian network.

[0104] A fault monitoring and early warning module is configured to integrate the pick wear rate sensitivity model into the drum full life cycle digital twin to obtain a drum fault monitoring model; and based on the drum fault monitoring model, complete fault monitoring and fault early warning of the drum.

[0105] A life prediction module is configured to analyze drum fault data based on the drum fault monitoring model, establish a drum life prediction curve, and predict the remaining life of the drum.

[0106] The above-described embodiments are only descriptions of the preferred modes of the application, and do not limit the scope of the application. Without departing from the design spirit of the application, various modifications and improvements to the technical solutions of the application made by those of ordinary skill in the art shall fall within the protection scope of the claims of the application.

Claims

1. A method for predicting the residual life of a shearer drum, characterized by, The method comprises the following steps: Collecting multi-source operation data and physical design data of the drum, wherein the drum comprises a cutting tooth; Analyzing characteristics of the multi-source operation data, and establishing a cutting tooth wear rate sensitivity model based on different coal and rock media and cutting parameters; Processing the multi-source operation data based on a deep adaptive fuzzy clustering algorithm, and constructing a drum data-driven model; Establishing a drum mechanism model based on the physical design data; Based on the matching relationship between the drum data-driven model and the drum mechanism model, a drum full life cycle digital twin is established by using a dynamic Bayesian network; Integrating the cutting tooth wear rate sensitivity model into the drum full life cycle digital twin to obtain a drum fault monitoring model; Based on the drum fault monitoring model, analyzing drum fault data, establishing a drum life prediction curve, and predicting the remaining life of the drum; The method for constructing a drum data-driven model is as follows: Normalizing the multi-source operation data, and reducing the dimensionality of the normalized multi-source operation data to obtain low-dimensional multi-source operation data; Constructing a sparse autoencoder layer, which learns a low-dimensional representation of the low-dimensional multi-source operation data by minimizing reconstruction error to obtain a trained sparse autoencoder layer; Stacking a preset number of trained sparse autoencoder layers to obtain a deep network; Based on the deep network, obtaining low-dimensional features of the multi-source operation data; Based on the deep adaptive fuzzy clustering algorithm, clustering the low-dimensional features; Based on the low-dimensional feature clustering result, obtaining compressed multi-source operation data; Based on the compressed multi-source operation data, constructing the drum data-driven model.

2. The method for predicting the remaining life of a shearer drum according to claim 1, wherein The multi-source operation data includes cutting parameters, cutting vibration data, temperature data, current and voltage data, cutting force data, acoustic emission data, infrared thermal imaging data, and operation time and cycle data; The cutting parameters include cutting speed, cutting depth, and cutting tooth angle; The physical design data includes drum diameter and length data, spiral blade parameters, cutting tooth arrangement data, material parameters, transmission system design parameters, and bearing and seal design parameters.

3. The coal cutter drum remaining life prediction method according to claim 2, characterized in that, The method for establishing a cutting tooth wear rate sensitivity model is as follows: Based on different cutting parameters and coal and rock media, corresponding cutting vibration data, acoustic emission data, and temperature data are obtained; Numerical simulation is performed on the cutting vibration data and acoustic emission data corresponding to different cutting parameters and coal and rock media to obtain boundary conditions; Based on the boundary conditions, drum vibration energy spectrum and acoustic emission energy spectrum are reconstructed; Based on the drum vibration energy spectrum and acoustic emission energy spectrum, frequency spectrum features are obtained; wherein the frequency spectrum features include vibration data frequency spectrum features and acoustic emission data frequency spectrum features; Numerical simulation is performed on the temperature data corresponding to different cutting parameters and coal and rock media to obtain flash temperature distribution; Combining the infrared thermal imaging data and the flash temperature distribution, a temperature and frequency distribution map is obtained; wherein the frequency is the relative frequency or probability of temperature falling within a certain specific interval. obtaining a temperature feature based on the temperature-frequency distribution map; establishing a pick wear rate sensitivity model based on the frequency spectrum feature, the temperature feature, and the pick wear rate.

4. The coal cutter drum remaining life prediction method according to claim 3, characterized in that, The specific method for establishing a pick wear rate sensitivity model based on the frequency spectrum feature, the temperature feature, and the pick wear rate is: obtaining a multi-source signal feature-pick wear singularity influence curve based on the frequency spectrum feature and the corresponding pick wear degree and the temperature feature and the corresponding pick wear degree; obtaining a different cutting parameter-coal rock medium pick wear rate influence curve based on different cutting parameters and the corresponding pick wear degree of the coal rock medium; analyzing a first correlation coefficient of the wear singularity influence curve and the wear rate influence curve by using a Spearman rank correlation coefficient; obtaining a second correlation coefficient of the frequency spectrum feature and the temperature feature and the wear rate influence curve based on the first correlation coefficient; assigning an influence weight to the frequency spectrum feature and the temperature feature based on the second correlation coefficient; establishing a pick wear rate sensitivity model based on the influence weight and the wear rate influence curve.

5. A system for predicting the residual lifetime of a shearer drum for implementing the method according to any one of claims 1 to 4, characterized in that comprise: a data acquisition module for acquiring drum multi-source operation data and physical design data; wherein the drum comprises picks; a pick wear rate sensitivity model construction module for analyzing the multi-source operation data features and establishing a pick wear rate sensitivity model based on different coal rock media and cutting parameters; a data-driven model construction module for processing the multi-source operation data based on a deep adaptive fuzzy clustering algorithm to construct a drum data-driven model; a mechanism model construction module for establishing a drum mechanism model based on the physical design data; a digital twin module for establishing a drum full-life-cycle digital twin by using a dynamic Bayesian network based on the matching relationship between the drum data-driven model and the drum mechanism model; a fault monitoring and early warning module for integrating the pick wear rate sensitivity model into the drum full-life-cycle digital twin to obtain a drum fault monitoring model; a life prediction module for analyzing drum fault data based on the drum fault monitoring model, establishing a drum life prediction curve, and predicting the remaining life of the drum.

Images