Method, device and equipment for evaluating health state of coal mine equipment

Abstract

The embodiment of the invention provides a method, device and equipment for evaluating the health state of coal mine equipment. The method comprises the following steps: acquiring multi-dimensional features of current coal mine equipment; fusing the multi-dimensional features to obtain a fused feature vector of the current coal mine equipment; inputting the fusion feature vector of the current coal mine equipment into a bidirectional long and short time memory network to obtain bidirectional time sequence features of the current coal mine equipment; obtaining structure enhancement features of the current coal mine equipment according to the bidirectional time sequence features of the current coal mine equipment and a graph neural network; and according to the bidirectional time sequence characteristics of the current coal mine equipment and the structure enhancement characteristics of the current coal mine equipment, evaluating the health state of the current coal mine equipment. In this way, comprehensive monitoring, accurate prediction and efficient maintenance of current coal mine equipment can be achieved through the multi-modal data collection and fusion technology.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of power equipment operation monitoring, and particularly to the field of coal mine equipment health state evaluation. BACKGROUND

[0002] Coal mine equipment plays a core role in mine production, and the stable operation of key equipment such as conveying, hoisting and ventilation directly affects production efficiency and safety. With the increase of coal mining depth and the complication of operating environment, the operating state of the equipment becomes more difficult to monitor, and the occurrence of faults has the characteristics of suddenness and diversity. In order to ensure the continuity and safety of coal mine production, real-time monitoring and scientific maintenance of the equipment state are particularly important.

[0003] At present, the maintenance of coal mine equipment mostly adopts the way of periodic maintenance or after-failure maintenance. Although periodic maintenance can reduce the risk of failure to a certain extent, it is easy to cause over-maintenance or insufficient maintenance due to the failure to combine the actual operating state of the equipment. After-failure maintenance may cause long-term shutdown of the equipment and serious economic losses due to the unpredictability of the failure.

[0004] Therefore, there is an urgent need for a method that can comprehensively evaluate the health state of coal mine equipment to improve the accuracy and timeliness of fault prediction, so as to effectively solve the maintenance problems in the operation of coal mine equipment. SUMMARY

[0005] The present disclosure provides a coal mine equipment health state evaluation method, device, equipment and storage medium.

[0006] According to a first aspect of the present disclosure, a coal mine equipment health state evaluation method is provided. The method comprises:

[0007] obtaining multi-dimensional features of a current coal mine equipment;

[0008] fusing the multi-dimensional features to obtain a fusion feature vector of the current coal mine equipment;

[0009] inputting the fusion feature vector of the current coal mine equipment into a bidirectional long short-term memory network to obtain bidirectional time sequence features of the current coal mine equipment;

[0010] obtaining structure-enhanced features of the current coal mine equipment according to the bidirectional time sequence features of the current coal mine equipment and a graph neural network;

[0011] evaluating the health state of the current coal mine equipment according to the bidirectional time sequence features of the current coal mine equipment and the structure-enhanced features of the current coal mine equipment.

[0012] In the aspect and any possible implementation manner as above, an implementation manner is further provided, and the inputting the fusion feature vector of the current coal mine equipment into the bidirectional long short-term memory network to obtain bidirectional time sequence features of the current coal mine equipment comprises:

[0013] The fusion feature vector of the current coal mine equipment is inputted into the bidirectional long short-term memory network to obtain a forward hidden state and a reverse hidden state of the current coal mine equipment.

[0014] The forward hidden state and the reverse hidden state of the current coal mine equipment are spliced to obtain the bidirectional time sequence features of the current coal mine equipment.

[0015] In the aspect and any possible implementation manner as above, an implementation manner is further provided, and the obtaining the structure enhanced features of the current coal mine equipment according to the bidirectional time sequence features of the current coal mine equipment and a graph neural network comprises:

[0016] An associated equipment related to the current coal mine equipment is determined, wherein the associated equipment is connected with the current coal mine equipment and / or has a logical control relationship.

[0017] A device topology graph is constructed according to the current coal mine equipment and the associated equipment.

[0018] The structure enhanced features of the current coal mine equipment are obtained according to the bidirectional time sequence features of the current coal mine equipment, the device topology graph and the graph neural network.

[0019] In the aspect and any possible implementation manner as above, an implementation manner is further provided, and the evaluating the health state of the current coal mine equipment according to the bidirectional time sequence features of the current coal mine equipment and the structure enhanced features of the current coal mine equipment comprises:

[0020] The bidirectional time sequence features of the current coal mine equipment and the structure enhanced features of the current coal mine equipment are spliced to obtain comprehensive features of the current coal mine equipment.

[0021] The remaining life of the current coal mine equipment is predicted according to the comprehensive features of the current coal mine equipment.

[0022] The health state of the current coal mine equipment is evaluated according to the remaining life of the current coal mine equipment.

[0023] In the aspect and any possible implementation manner as above, an implementation manner is further provided, and the method further comprises:

[0024] The prediction probability of the current coal mine equipment belonging to various fault types is determined according to the remaining life of the current coal mine equipment.

[0025] selecting, from the various fault types, a fault type with a maximum predicted probability as a target fault type of the current coal mine equipment.

[0026] According to the aspect and any possible implementation manner as described above, further provided is an implementation manner, wherein the evaluating the health status of the current coal mine equipment according to the remaining life of the current coal mine equipment comprises:

[0027] obtaining an equipment importance index of the current coal mine equipment;

[0028] obtaining an environmental risk factor;

[0029] calculating a health score of the current coal mine equipment according to the remaining life of the current coal mine equipment, the equipment importance index of the current coal mine equipment and the environmental risk factor;

[0030] evaluating the health status of the current coal mine equipment according to the health score of the current coal mine equipment.

[0031] According to the aspect and any possible implementation manner as described above, further provided is an implementation manner, wherein the method further comprises:

[0032] obtaining health statuses of each preset coal mine equipment;

[0033] comparing the health status of the current coal mine equipment with the health statuses of each preset coal mine equipment to determine a health priority ranking of each coal mine equipment in the current coal mine equipment and each preset coal mine equipment;

[0034] maintaining each coal mine equipment according to the health priority ranking of each coal mine equipment.

[0035] According to a second aspect of the present disclosure, a coal mine equipment health status evaluation device is provided. The device comprises:

[0036] a first obtaining module configured to obtain multi-dimensional features of a current coal mine equipment;

[0037] a fusion module configured to fuse the multi-dimensional features to obtain a fusion feature vector of the current coal mine equipment;

[0038] a second obtaining module configured to input the fusion feature vector of the current coal mine equipment into a bidirectional long short-term memory network to obtain bidirectional time sequence features of the current coal mine equipment;

[0039] a third obtaining module configured to obtain structure enhanced features of the current coal mine equipment according to the bidirectional time sequence features of the current coal mine equipment and a graph neural network;

[0040] An evaluation module is configured to evaluate the health status of the current coal mine equipment according to the bidirectional time sequence feature of the current coal mine equipment and the structure enhancement feature of the current coal mine equipment.

[0041] According to a third aspect of the present disclosure, an electronic device is provided. The electronic device comprises a memory and a processor, the memory having stored thereon a computer program, the processor implementing the method as described above when executing the program.

[0042] According to a fourth aspect of the present disclosure, a computer readable storage medium is provided, having stored thereon a computer program, the program being executed by a processor to implement the method according to the first aspect of the present disclosure.

[0043] In the present disclosure, after obtaining the multi-dimensional features of the current coal mine equipment, the multi-dimensional features can be fused to obtain a fusion feature vector of the current coal mine equipment, and then the fusion feature vector of the current coal mine equipment is input into a bidirectional long short-term memory network to obtain a bidirectional time sequence feature of the current coal mine equipment. According to the bidirectional time sequence feature of the current coal mine equipment and a graph neural network, a structure enhancement feature of the current coal mine equipment can be obtained. Furthermore, according to the bidirectional time sequence feature of the current coal mine equipment and the structure enhancement feature of the current coal mine equipment, the health status of the current coal mine equipment can be accurately evaluated. Thus, the multi-modal data acquisition and fusion technology can be used to realize comprehensive monitoring, accurate prediction and efficient maintenance of the current coal mine equipment.

[0044] It should be understood that the content described in the summary section is not intended to limit or define key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0045] The above and other features, advantages and aspects of embodiments of the present disclosure will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings in which:

[0046] Figure 1 A flow chart of a coal mine equipment health status evaluation method according to an embodiment of the present disclosure is shown;

[0047] Figure 2 A block diagram of a coal mine equipment health status evaluation device according to an embodiment of the present disclosure is shown;

[0048] Figure 3 A block diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0049] To make the purposes, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be described clearly and completely below with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are some but not all of the embodiments of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present disclosure.

[0050] In addition, the term "and / or" herein is only to describe the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can mean that A exists alone, A and B exist together, and B exists alone. In addition, the character " / " herein generally represents an "or" relationship between the front and rear associated objects.

[0051] Figure 1 A flowchart of a coal mine equipment health state evaluation method 100 according to an embodiment of the present disclosure is shown. The method 100 can include:

[0052] Step 110, acquiring multi-dimensional features of a current coal mine equipment;

[0053] The data source of the collected multi-dimensional features is multi-modal data, which can be: equipment operation data and environmental parameters. The operation data includes the vibration amplitude, temperature, current, and voltage fluctuation of the current coal mine equipment, and the environmental parameters include humidity and dust concentration.

[0054] At the same time, in order to further improve the comprehensiveness and accuracy of the data, the acoustic signal of the current coal mine equipment and the image information of the key components can also be collected for identifying potential abnormalities (such as cracks or wear).

[0055] Step 120, fusing the multi-dimensional features to obtain a fusion feature vector of the current coal mine equipment;

[0056] The collected multi-dimensional features are subjected to data cleaning, normalization, and feature extraction to form a standardized feature set, wherein,

[0057] Data cleaning eliminates noise and outliers in the collection process, and normalization processing adjusts various data to a unified range;

[0058] And by using the self-attention mechanism to weight and fuse the multi-modal multi-dimensional features, a fusion feature vector of the comprehensive operation state of the equipment can be generated.

[0059] For example: the operation state of equipment i in the last time window (such as the past 24 hours) is collected, including vibration, temperature, current, voltage, image, and acoustic signal, etc., to form its multi-modal fusion input sequence:

[0060] X i = [X i (t-T+1),...,X i (t)]

[0061] Parameter interpretation:

[0062] X i (t): represents the fusion feature vector of device i at time t, which is composed of data collected by multiple sensors, including vibration amplitude, environmental temperature and humidity, motor image, running sound signal, etc.

[0063] T: represents the length of the time window, which is used to provide a continuous time background;

[0064] X i : represents the time series operation data of device i, which is the basic feature sequence input to the time modeling module.

[0065] Step 130, input the fusion feature vector of the current coal mine equipment into the bidirectional long short-term memory network (i.e. Bi-LSTM), and obtain the bidirectional time sequence feature of the current coal mine equipment;

[0066] The bidirectional time sequence feature of the current coal mine equipment reflects not only the past operation trend of the current coal mine equipment, but also the influence of the future window on the current state of the current coal mine equipment.

[0067] Step 140, according to the bidirectional time sequence feature of the current coal mine equipment and the graph neural network (i.e. GNN, Graph Neural Network), obtain the structure enhanced feature of the current coal mine equipment;

[0068] The structure enhanced feature represents the structure feature after the operation feature of the current coal mine equipment and the influence of the neighbor device are fused, which is used to reflect the operation status and mutual dependence state of the current coal mine equipment in the whole system.

[0069] Step 150, according to the bidirectional time sequence feature of the current coal mine equipment and the structure enhanced feature of the current coal mine equipment, evaluate the health status of the current coal mine equipment.

[0070] After obtaining the multi-dimensional features of the current coal mine equipment, the multi-dimensional features can be fused to obtain a fusion feature vector of the current coal mine equipment, and then the fusion feature vector of the current coal mine equipment is input into a bidirectional long short-term memory network to obtain a bidirectional time sequence feature of the current coal mine equipment. According to the bidirectional time sequence feature of the current coal mine equipment and a graph neural network, a structure enhancement feature of the current coal mine equipment can be obtained. Then, according to the bidirectional time sequence feature of the current coal mine equipment and the structure enhancement feature of the current coal mine equipment, the health status of the current coal mine equipment can be accurately evaluated. Thus, the multi-modal data acquisition and fusion technology can realize comprehensive monitoring, accurate prediction and efficient maintenance of the current coal mine equipment.

[0071] In some embodiments, the inputting the fusion feature vector of the current coal mine equipment into the bidirectional long short-term memory network to obtain the bidirectional time sequence feature of the current coal mine equipment comprises:

[0072] inputting the fusion feature vector of the current coal mine equipment into the bidirectional long short-term memory network to obtain a forward hidden state and a reverse hidden state of the current coal mine equipment;

[0073] splicing the forward hidden state and the reverse hidden state of the current coal mine equipment to obtain the bidirectional time sequence feature of the current coal mine equipment.

[0074] By inputting the fusion feature vector of the current coal mine equipment into the bidirectional long short-term memory network, the forward hidden state and the reverse hidden state of the current coal mine equipment can be obtained. Then, the forward hidden state and the reverse hidden state of the current coal mine equipment are spliced to accurately obtain the bidirectional time sequence feature of the current coal mine equipment.

[0075] The steps of obtaining the bidirectional time sequence feature are as follows:

[0076] The fusion feature vector X i is processed by using the bidirectional long short-term memory network to extract the change trend of the equipment running state in the time dimension, and a hidden state sequence H i (t) is obtained.

[0077] H i (t) = LSTM(X i (t), H i (t-1))

[0078] Parameter explanation:

[0079] H i (t): represents the hidden state of the equipment i at time t, which is an internal feature output by the Bi-LSTM network according to the current input feature and the memory state at the previous moment, reflecting the historical legacy effect and dynamic trend of the equipment running state.

[0080] H i (t-1): represents the hidden state of the previous moment, which is the memory of the model for the past running state of the device;

[0081] The hidden state reflects the running trend and historical legacy effect of the device, although it is not directly used for predicting the output, but constitutes an important basis for the subsequent Bi-LSTM sequence output .

[0082] This mechanism can identify trends such as "vibration amplitude continues to rise" or "periodic current anomalies", and provide temporal context for subsequent prediction.

[0083] To further model the running trend in the entire time window, the forward and reverse LSTM structures are fused to obtain complete time series features:

[0084]

[0085] represents the bidirectional time series feature vector of the device i under the given time sequence, which reflects both the past running trend and the influence of the future window on the current state.

[0086] In fact, The bidirectional hidden state at the end of the time window can be composed, that is:

[0087]

[0088] Where:

[0089] the final hidden state of the forward LSTM (i.e., the forward hidden state);

[0090] the final hidden state of the reverse LSTM (i.e., the reverse hidden state);

[0091] represents the vector splicing operation.

[0092] In some embodiments, the structure-enhanced features of the current coal mine equipment are obtained according to the bidirectional time series features of the current coal mine equipment and the graph neural network, comprising:

[0093] determining an associated device related to the current coal mine equipment, wherein the associated device is connected with the current coal mine equipment and / or has a logical control relationship;

[0094] constructing a device topology graph according to the current coal mine equipment and the associated device;

[0095] A device topology graph can be constructed by considering the physical connection and control logic relationship between devices (such as a conveyor and a crusher, a motor and a drive controller). Specifically, the devices directly connected to the current coal mine device or the devices having a control relationship with the current coal mine device are associated devices. The current coal mine device and the associated devices are taken as a node, and the relationship between the current coal mine device and the associated devices is taken as an edge, so as to construct the device topology graph. In this way, the devices directly connected to the current coal mine device in the associated devices form the neighbor nodes of the current coal mine device, or all the associated devices are the neighbor nodes of the current coal mine device.

[0096] According to the bidirectional time sequence feature of the current coal mine device, the device topology graph and the graph neural network, a structure-enhanced feature of the current coal mine device is obtained.

[0097] The bidirectional time sequence feature of the current coal mine device and the device topology graph are input into the graph neural network, so as to obtain the structure-enhanced feature output by the graph neural network.

[0098] In some embodiments, the health status of the current coal mine device is evaluated according to the bidirectional time sequence feature of the current coal mine device and the structure-enhanced feature of the current coal mine device, including:

[0099] The bidirectional time sequence feature of the current coal mine device and the structure-enhanced feature of the current coal mine device are spliced to obtain a comprehensive feature of the current coal mine device.

[0100] The comprehensive feature integrates the running trend of the current coal mine device itself and the structural dependency information between the current coal mine device and other devices.

[0101] According to the comprehensive feature of the current coal mine device, the remaining life of the current coal mine device is predicted.

[0102] According to the remaining life of the current coal mine device, the health status of the current coal mine device is evaluated.

[0103] By automatically splicing the bidirectional time sequence feature of the current coal mine device and the structure-enhanced feature of the current coal mine device, the comprehensive feature of the current coal mine device can be obtained. Then, according to the comprehensive feature of the current coal mine device, the remaining life of the current coal mine device can be predicted by using a regression model. According to the remaining life of the current coal mine device, the health score of the current coal mine device can be calculated in time and accurately, and then the health status of the current coal mine device can be accurately evaluated.

[0104] In some embodiments, the method further includes:

[0105] According to the remaining life of the current coal mine device, the prediction probability of the current coal mine device belonging to various fault types is determined.

[0106] selecting a fault type with the maximum predicted probability from the various fault types as the target fault type of the current coal mine equipment.

[0107] According to the remaining life of the current coal mine equipment, the predicted probability of the current coal mine equipment belonging to various fault types can be determined, and then a fault type with the maximum predicted probability is selected from the various fault types as the target fault type of the current coal mine equipment, so that the fault type of the current coal mine equipment is accurately predicted by using the multi-dimensional features of the current coal mine equipment.

[0108] Fault type prediction (multi-classification output):

[0109] P i = softmax(W c ·Z i +b c ), T i = argmax j P ij

[0110] P i : represents a predicted probability vector of the equipment i belonging to each fault type;

[0111] P ij : represents the probability of the equipment i being determined as the jth fault type, and the value range is between [0, 1];

[0112] T i = argmax j P ij : represents selecting a category with the maximum predicted probability from all fault types as the target fault type output of the equipment i;

[0113] W c ,b c : are training parameters of the fault classifier, which are obtained by learning historical labeled fault samples;

[0114] Z i : is the integrated equipment feature (i.e., the integrated feature of the current coal mine equipment) fused, which is input into the classifier.

[0115] In some embodiments, the health state of the current coal mine equipment is evaluated according to the remaining life of the current coal mine equipment, including:

[0116] obtaining an equipment importance index of the current coal mine equipment;

[0117] obtaining an environmental risk factor;

[0118] According to the remaining life of the current coal mine equipment, the equipment importance index of the current coal mine equipment, and the environmental risk factor, a health score of the current coal mine equipment is calculated;

[0119] According to the health score of the current coal mine equipment, the health status of the current coal mine equipment is evaluated.

[0120] According to the remaining life of the current coal mine equipment, the equipment importance index of the current coal mine equipment, and the environmental risk factor, a health score of the current coal mine equipment is accurately calculated, and then according to the health score of the current coal mine equipment, the health status of the current coal mine equipment is accurately evaluated.

[0121] Based on the predicted remaining life RUL i , combined with the equipment importance index W i and the environmental risk factor E i , the health score of the equipment can be calculated:

[0122]

[0123] Parameter explanation:

[0124] S i : is the comprehensive health score of the equipment i, the higher the value, the better the state;

[0125] T max : the upper limit of the design life of the equipment;

[0126] W i : the weight score of the equipment i, such as core, maintenance cost, etc.;

[0127] W max : importance normalization benchmark;

[0128] E i : environmental risk factor, such as dust concentration, temperature and humidity, vibration and impact frequency;

[0129] α,β,γ: are score factor weights, used to balance the influence of each index on the result, which can be adjusted to match the actual scene strategy preference.

[0130] In some embodiments, the method further comprises:

[0131] Obtaining the health status of each preset coal mine equipment;

[0132] Comparing the health status of the current coal mine equipment with the health status of each preset coal mine equipment to determine the health priority ranking of each coal mine equipment in the current coal mine equipment and the each preset coal mine equipment;

[0133] The better the health state of the coal mine equipment, the higher the health priority ranking, specifically: the health score is used to measure the health state, the higher the health score, the higher the health priority ranking.

[0134] According to the health priority ranking of each coal mine equipment, the maintenance of each coal mine equipment is carried out.

[0135] By obtaining the health state of each preset coal mine equipment, the health state of the current coal mine equipment can be automatically compared with the health state of each preset coal mine equipment, and then the health priority ranking of each coal mine equipment in the current coal mine equipment and each preset coal mine equipment is determined, and then according to the health priority ranking of each coal mine equipment, the maintenance of each coal mine equipment can be carried out in turn, so that the maintenance sequence of the coal mine equipment is more accurate.

[0136] The coal mine equipment health state evaluation method of the present application will be further described below:

[0137] I. Input feature construction

[0138] Firstly, the running state of equipment i in the latest time window (such as the past 24 hours) is collected, including vibration, temperature, current, voltage, image and acoustic signal, etc., to form its multi-modal fusion input sequence:

[0139] X i = [X i (t-T+1),...,X i (t)]

[0140] Parameter explanation:

[0141] X i (t): represents the fusion feature vector of equipment i at time t, which is composed of data collected by multiple sensors, including vibration amplitude, environmental temperature and humidity, motor image, running sound signal, etc.

[0142] T: represents the length of the time window, which is used to provide a continuous time background;

[0143] X i : represents the time series running data of equipment i, which is the basic feature sequence input to the time modeling module.

[0144] II. Time series modeling (Bi-LSTM)

[0145] X i is processed by using a bidirectional long short-term memory network to extract the change trend of the equipment running state in the time dimension, and a hidden state sequence H i (t) is obtained:

[0146] H i (t) = LSTM(Xi (t),H i (t-1))

[0147] Parameter interpretation:

[0148] H i (t): represents the hidden state of device i at time t, which is the internal feature output by the Bi-LSTM network according to the current input feature and the memory state at the previous moment, reflecting the historical legacy effect and dynamic trend of the device running state;

[0149] H i (t-1): represents the hidden state at the previous moment, which is the memory of the model for the past running state of the device; this hidden state reflects the running trend and historical legacy effect of the device, although it is not directly used for prediction output, but constitutes an important basis for the subsequent Bi-LSTM sequence output .

[0150] This mechanism can identify trends such as "vibration amplitude continuously rising" or "periodic current anomalies", providing temporal context for subsequent prediction.

[0151] To further model the running trend within the entire time window, the forward and reverse LSTM structures are fused to obtain complete time series features:

[0152]

[0153] represents the bidirectional time sequence feature vector of device i under the given time sequence, reflecting both past running trends and future window effects on the current state.

[0154] In fact, can be composed by splicing the bidirectional hidden state at the end of the time window, that is

[0155]

[0156] Where:

[0157] the final hidden state of the forward LSTM;

[0158] the final hidden state of the reverse LSTM;

[0159] represents the vector splicing operation.

[0160] This result is used as the input for graph neural network modeling, that is:

[0161]

[0162] III. Structure correlation modeling (graph neural network, GNN)

[0163] In the present application, the physical connection and control logic relationship between devices (such as conveyors and crushers, motors and drive controllers) are considered, and a device topology graph is constructed, and the running state of the neighbor nodes is modeled based on graph neural network. The initial input is the feature of the time modeling output

[0164]

[0165] Parameter explanation:

[0166] represents the state feature of device i at the l-th layer of GNN, which integrates the running state of neighbor devices;

[0167] represents the state feature of device j at the l-1-th layer of graph neural network, which is initially from its time series feature is assigned and updated layer by layer in the graph convolution process, reflecting the information interaction state between it and its neighbor devices;

[0168] In the graph neural network modeling, device i represents the target device that needs to be predicted and scored, and device j ∈ N(i) represents the neighbor device connected to it, whose running state affects the structure feature construction of device i through feature propagation mechanism.

[0169] N(i): represents the neighbor device set directly connected to device i (such as an upstream and downstream device of a belt conveyor);

[0170] d i , d j : respectively the connection degree of device i and neighbor device j, i.e. the number of neighbors;

[0171] W (l) : is the trainable weight matrix of the l-th layer of graph convolution;

[0172] σ(·): is the activation function (such as ReLU), which is used to introduce non-linear enhancement modeling capability.

[0173] After L layers of propagation, the structure enhanced feature is obtained:

[0174]

[0175] represents the structure feature that integrates the running feature of device i and the influence of neighbor devices, which is used to reflect the running status and mutual dependence of the device in the whole system.

[0176] IV. Feature fusion and prediction modeling

[0177] Time features Structural features Fusion, integrated representation of device i:

[0178]

[0179] Parameter interpretation:

[0180] Z i : represents the fused feature vector, integrating both the device's own operational trends and the structural dependency information with other devices;

[0181] represents vector concatenation operation or weighted fusion operation.

[0182] Then use this fused feature for two prediction tasks:

[0183] 1. Remaining useful life prediction (regression output):

[0184] RUL i = W r · Z i + b r

[0185] RUL i : Remaining Useful Life prediction result of device i, unit can be hours or cycles;

[0186] W r , b r : weight parameters and bias terms in the regression model, can be obtained by training historical life data.

[0187] 2. Fault type prediction (multi-classification output):

[0188] P i = softmax(W c · Z i + b c ), T i = argmax j P ij

[0189] P i : represents the prediction probability vector of device i belonging to each fault type;

[0190] P ij : represents the probability of device i being judged as the jth fault, taking values in the range [0, 1];

[0191] T i = argmax j P ij: represents selecting the class with the maximum prediction probability from all fault types as the fault type output of device i;

[0192] W c ,b c : is a training parameter of the fault classifier, which is learned by historical labeled fault samples;

[0193] Z i : is the integrated feature of the device, which is input into the classifier.

[0194] Five, health score calculation and sorting

[0195] Based on the predicted remaining useful life RUL i , combined with the importance of the device W i and the environmental risk factor E i , the health score of the device is calculated:

[0196]

[0197] Parameter explanation:

[0198] S i : is the comprehensive health score of device i, and the higher the value, the better the state;

[0199] T max : the upper limit of the design life of the device;

[0200] W i : the weight score of device i, such as core nature, maintenance cost, etc.;

[0201] W max : importance normalization benchmark;

[0202] E i : environmental risk factor, such as dust concentration, temperature and humidity, vibration and impact frequency;

[0203] Alpha, beta, gamma: are the weight factors of the score, used to balance the influence of each index on the result, which can be adjusted to match the actual scene strategy preference.

[0204] The following is a supplementary explanation:

[0205] In the present application, the health score mechanism can be used for the whole device, or can be refined to the key components inside the device, depending on the granularity requirement of predictive maintenance in the actual application scene.

[0206] In the core method, the health score of device i is calculated by the following formula:

[0207]

[0208] The above score reflects the weighted results of the running state, remaining life and risk influence of the entire device, and is suitable for occasions where the structure is relatively simple or maintenance decisions are made for the entire machine.

[0209] For complex devices, such as a belt conveyor, which includes multiple key sub-components (such as bearings, motors, belts, etc.), the scoring method of the present application can be extended to the component level according to the same principle:

[0210]

[0211] where S i,k is the score of the kth component of device i; the meanings of the parameters correspond to those of the entire machine score but act on the component dimension. The score result can be used for priority ranking of the key components within the device, assisting in implementing more refined maintenance scheduling strategies.

[0212] Application Example: Intelligent Predictive Maintenance of Belt Conveyor

[0213] In order to better illustrate the practical application of the present application, this example takes the belt conveyor, a key device in coal mines, as an example to describe how to use the intelligent predictive maintenance method and device to achieve precise maintenance.

[0214] The belt conveyor in a coal mine is the core equipment for conveying coal, and it is prone to bearing failure, belt tearing and drive motor abnormalities due to long-term operation in a high-humidity and high-dust environment. If these problems are not addressed in a timely manner, it may lead to a coal mine shutdown or safety accidents, so intelligent predictive maintenance of the belt conveyor is particularly important.

[0215] Implementation Steps

[0216] 1. Data Collection and Processing

[0217] Vibration sensors, temperature sensors and acoustic sensors are installed on the drive motor, bearings and tensioning devices of the belt conveyor, and industrial cameras are installed at key positions of the conveyor belt to collect real-time equipment operation data and environmental parameters. The data includes the vibration amplitude and temperature rise of the motor, bearing noise signals, belt surface images, as well as humidity and dust concentration in the environment. The collected data is cleaned, normalized and feature fused by the data processing module to form a comprehensive feature vector, which fully reflects the running state of the conveyor.

[0218] 2. Fault Prediction

[0219] Through the deep learning model of the fault prediction module, combined with historical data and real-time collected data, the bearing failure, motor abnormality and belt tearing risk of the belt conveyor are predicted.

[0220] For example, the remaining useful life (RUL) of the bearing is predicted through time series features of vibration and temperature data.

[0221] The source of abnormal noise of the motor is identified through spectral analysis of acoustic data.

[0222] The image data collected by the camera is combined with image recognition algorithms to detect tears or wear on the surface of the belt.

[0223] 3. Health score calculation

[0224] According to the prediction results, the comprehensive health score of the key components of the belt conveyor is calculated.

[0225] Assuming the remaining useful life of the bearing is RUL bearing = 500, the importance weight of the motor is W motor = 0.8, the environmental risk factor is E env = 0.3, and the health score formula is:

[0226]

[0227] Similarly, the health scores of the bearings, motors, and belts in the belt conveyor can be calculated in combination with specific parameters, thereby prioritizing each component.

[0228] It should be noted that for the foregoing method embodiments, in order to simply describe, they are all expressed as a series of action combinations, but those skilled in the art should know that the present disclosure is not limited by the order of the described actions, because according to the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present disclosure.

[0229] The above is an introduction to the method embodiments, and the following device embodiments will further illustrate the scheme described in the present disclosure.

[0230] Figure 2 A block diagram of a coal mine equipment health state evaluation device 200 according to an embodiment of the present disclosure is shown. As Figure 2 shown, the device 200 includes:

[0231] A first acquisition module 210 is configured to acquire multi-dimensional features of a current coal mine equipment.

[0232] A fusion module 220 is configured to fuse the multi-dimensional features to obtain a fusion feature vector of the current coal mine equipment.

[0233] The second obtaining module 230 is configured to input the fusion feature vector of the current coal mine equipment into a bidirectional long short-term memory network to obtain bidirectional time sequence features of the current coal mine equipment.

[0234] The third obtaining module 240 is configured to obtain structure enhanced features of the current coal mine equipment according to the bidirectional time sequence features of the current coal mine equipment and a graph neural network.

[0235] The evaluation module 250 is configured to evaluate the health status of the current coal mine equipment according to the bidirectional time sequence features of the current coal mine equipment and the structure enhanced features of the current coal mine equipment.

[0236] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described modules can refer to the corresponding process in the foregoing method embodiments, which will not be described herein.

[0237] According to embodiments of the present disclosure, the present disclosure further provides an electronic device and a non-transitory computer-readable storage medium having computer instructions stored therein.

[0238] Figure 3 A schematic block diagram of an electronic device 800 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices, and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present disclosure described and / or claimed in this document.

[0239] The device 800 includes a computing unit 801 that can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. Various programs and data required for the operation of the device 800 can also be stored in the RAM 803. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other through a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0240] A plurality of components in the device 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a magnetic disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0241] The computing unit 801 can be various general and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 801 performs various methods and processes described above, such as the method 100. For example, in some embodiments, the method 100 can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 808. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the method 100 described above can be performed. Alternatively, in other embodiments, the computing unit 801 can be configured to perform the method 100 by any other appropriate means, such as by means of firmware.

[0242] Various implementations of the systems and techniques described above herein can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip system (SOC), a programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0243] Program code for carrying out methods of the present disclosure can be written in any combination of one or more programming languages. The program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, produces the functions / operations specified in the flowcharts and / or block diagrams. The program code can be executed entirely on a machine, partially on a machine, partially on a machine as a standalone software package, or entirely on a remote machine or server.

[0244] In the context of the present disclosure, a machine-readable medium can be a tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium will include one or more lines of electrical connections, portable computer disks, hard disk drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0245] To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0246] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0247] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0248] It should be understood that the steps shown in the various forms above can be reordered, added to, or removed. For example, the steps described in the present disclosure can be performed in parallel, in series, or in a different order, as long as the desired results of the present disclosure are achieved, and the present disclosure is not limited in this regard.

[0249] The specific embodiments described above do not constitute an exhaustive list of all possible embodiments as modifications can be made by one skilled in the art, especially in light of the above teachings. Any and every modification, equivalent substitution, and improvement not described above that falls within the spirit and scope of the disclosure should be included.

Claims

1. A method for evaluating the health status of coal mine equipment, characterized in that: include: Obtain multi-dimensional characteristics of current coal mining equipment; Fusing the multi-dimensional features to obtain a fused feature vector of the current coal mining equipment; Inputting the fusion feature vector of the current coal mining equipment into a bidirectional long short-term memory network to obtain a bidirectional time series feature of the current coal mining equipment; Obtaining structural enhancement features of the current coal mining equipment based on the bidirectional time series features of the current coal mining equipment and a graph neural network; The health status of the current coal mining equipment is evaluated according to the bidirectional time sequence characteristics of the current coal mining equipment and the structural enhancement characteristics of the current coal mining equipment.

2. The method according to claim 1, wherein Inputting the fused feature vector of the current coal mining equipment into a bidirectional long short-term memory network to obtain the bidirectional time series features of the current coal mining equipment includes: Inputting the fused feature vector of the current coal mining equipment into the bidirectional long short-term memory network to obtain the forward hidden state and the reverse hidden state of the current coal mining equipment; The forward hidden state and the reverse hidden state of the current coal mining equipment are spliced ​​to obtain a bidirectional time series feature of the current coal mining equipment.

3. The method according to claim 1, wherein The step of obtaining the structural enhancement features of the current coal mining equipment based on the bidirectional time series features of the current coal mining equipment and the graph neural network includes: Determining associated equipment related to the current coal mining equipment, wherein the associated equipment is connected to and / or has a logical control relationship with the current coal mining equipment; Constructing a device topology map based on the current coal mine equipment and the associated equipment; According to the bidirectional timing characteristics of the current coal mine equipment, the equipment topology diagram and the graph neural network, the structural enhancement characteristics of the current coal mine equipment are obtained.

4. The method according to claim 1, wherein The evaluating the health status of the current coal mining equipment according to the bidirectional time sequence characteristics of the current coal mining equipment and the structural enhancement characteristics of the current coal mining equipment includes: splicing the bidirectional time sequence characteristics of the current coal mining equipment and the structural enhancement characteristics of the current coal mining equipment to obtain a comprehensive characteristic of the current coal mining equipment; predicting the remaining life of the current coal mining equipment based on the comprehensive characteristics of the current coal mining equipment; The health status of the current coal mining equipment is evaluated according to the remaining life of the current coal mining equipment.

5. The method according to claim 4, wherein The method further comprises: determining, based on the remaining life of the current coal mining equipment, a predicted probability that the current coal mining equipment belongs to various fault types; A fault type with the greatest predicted probability is selected from the various fault types as the target fault type of the current coal mine equipment.

6. The method according to claim 4, wherein The evaluating the health status of the current coal mining equipment according to the remaining life of the current coal mining equipment includes: Obtaining an equipment importance index of the current coal mine equipment; Obtain environmental risk factors; Calculating a health score of the current coal mine equipment according to the remaining life of the current coal mine equipment, the equipment importance index of the current coal mine equipment, and the environmental risk factor; The health status of the current coal mine equipment is evaluated according to the health score of the current coal mine equipment.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: Get the health status of each preset coal mine equipment; comparing the health status of the current coal mining equipment with the health status of each of the preset coal mining equipment, and determining a health priority ranking between the current coal mining equipment and each of the preset coal mining equipment; The coal mine equipment is repaired according to the health priority ranking of the coal mine equipment.

8. A coal mine equipment health status assessment device, characterized in that: include: The first acquisition module is used to obtain the multi-dimensional characteristics of the current coal mine equipment; A fusion module, configured to fuse the multi-dimensional features to obtain a fusion feature vector of the current coal mining equipment; A second acquisition module is configured to input the fused feature vector of the current coal mining equipment into a bidirectional long short-term memory network to obtain a bidirectional time series feature of the current coal mining equipment; a third acquisition module, configured to obtain a structural enhancement feature of the current coal mining equipment based on the bidirectional time series feature of the current coal mining equipment and a graph neural network; An evaluation module is configured to evaluate the health status of the current coal mining equipment according to the bidirectional time sequence characteristics of the current coal mining equipment and the structural enhancement characteristics of the current coal mining equipment.

9. An electronic device, characterized in that: include: memory and processor, The memory stores a computer program, and when the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that When the instructions in the storage medium are executed by a processor corresponding to the electronic device, the electronic device is enabled to implement the coal mine equipment health status assessment method as described in any one of claims 1 to 7.

Images