A method, device and related components for intelligent assessment of health status of coal mining machine

Through the combination of K-mean algorithm and bidirectional long and short-term memory network model, dynamic clustering and classification are performed, combined with fuzzy comprehensive evaluation, the problem of multi-time change information in the assessment of coal mining machine health status is solved, and the evaluation is improved Accuracy and efficiency.

CN116226751BActive Publication Date: 2024-12-03SHANXI KEDA AUTOMATION CONTROL
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202310087191.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-01-31
Publication Date
2024-12-03
Estimated Expiration
2043-01-31

AI Technical Summary

Technical Problem

The existing coal mining machine health status evaluation methods lack consideration for the changes in the equipment's state during evaluation, resulting in low reliability of the evaluation results.

Method used

Dynamic clustering analysis is used for dynamic clustering analysis to remove redundant data, and the monitoring data is classified in combination with the pre-trained two-way long and short-term memory network model, and the health score of the coal miner is calculated through the fuzzy comprehensive evaluation rules, fully considering the time period information.

Benefits of technology

It improves the accuracy and efficiency of the health score of coal mining machines, reduces the interference of similar monitoring data on the evaluation results, and enhances the reliability of the evaluation results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116226751B_ABST
    Figure CN116226751B_ABST
Patent Text Reader

Abstract

The present invention discloses an intelligent evaluation method, device and related components for the health status of a coal mining machine. The method includes obtaining monitoring data of each target position of the coal mining machine in multiple adjacent time periods; using the K-means algorithm to perform dynamic cluster analysis on all the corresponding monitoring data in each of the time periods, and removing similar monitoring data based on the clustering results obtained by the dynamic clustering analysis; using a pre-trained bidirectional long short-term memory network model to classify the monitoring data after removing similar monitoring data, and output the classification results; based on the preset coal mining machine health status evaluation rules, all the classification results are used for fuzzy comprehensive evaluation to calculate the coal mining machine health score of the current time period. The method is based on the monitoring data of the continuous time period of the coal mining machine, and through bidirectional long short-term network model classification and fuzzy comprehensive evaluation, the rationality and accuracy of the health score of the coal mining machine in the current time period are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of equipment maintenance, and in particular to a method and device for intelligently evaluating the health status of a coal mining machine and related components. Background Art

[0002] With the development of technologies such as big data and artificial intelligence, the coal industry is gradually entering the era of intelligence. In order to improve the efficiency and intelligence level of coal development, coal mine intelligent equipment, as the core of coal mine intelligence, must have good adaptability and reliability. The health status evaluation, prediction and maintenance of comprehensive mining equipment are the guarantee for the development of coal mine intelligence.

[0003] In actual application scenarios, coal mining machines integrate mechanical, electrical and hydraulic components, and are responsible for both coal cutting and coal loading. They operate for a long time in working faces with complex geological conditions and harsh environments. Accurate assessment of their health status can provide a basis for the maintenance of the coal mining machines and thus ensure the normal operation of the equipment.

[0004] At present, the existing coal mining machine health status assessment method only focuses on the health status of the coal mining machine in the current period, that is, the health status of the coal mining machine is determined based on the health status monitoring data of the current period, which will result in low reliability of the evaluation results. Summary of the invention

[0005] The purpose of the present invention is to provide a method, device and related components for intelligently evaluating the health status of a coal mining machine, aiming to solve the problem of low accuracy in existing health status evaluation of coal mining machines.

[0006] In order to solve the above technical problems, the object of the present invention is to achieve the following technical solutions: to provide a method for intelligently evaluating the health status of a coal mining machine, which comprises:

[0007] Acquire monitoring data of each target position of the coal mining machine in multiple adjacent time periods;

[0008] Performing dynamic cluster analysis on all the monitoring data corresponding to each of the time periods using a K-means algorithm, and removing similar monitoring data based on clustering results obtained by the dynamic cluster analysis;

[0009] Using a pre-trained bidirectional long short-term memory network model to classify the monitoring data after removing similar monitoring data, and output the classification result;

[0010] Based on the preset coal mining machine health status evaluation rules, all the classification results are used to perform fuzzy comprehensive evaluation to calculate the coal mining machine health score for the current time period.

[0011] Through the above technical scheme, firstly, by collecting monitoring data of multiple continuous time periods of the coal mining machine and removing redundant data in all monitoring data, the interference of similar monitoring data on the health status assessment results is reduced and the efficiency of outputting the health score results is improved. Then, the monitoring data of each target position in all time periods is classified through a bidirectional long short-term memory network model to obtain the probability of the health status classification of each monitoring data. Finally, a fuzzy comprehensive evaluation is performed using the probability of the health status classification to calculate the health score of the coal mining machine. Since this application obtains the health score of the coal mining machine in the current time period through monitoring data of multiple continuous time periods, it fully takes into account the characteristic that the health status change process of the coal mining machine is continuous in time, and faces the influence of time information on the health score, thereby ensuring the accuracy of the health score of the coal mining machine.

[0012] Furthermore, the K-means algorithm is used to perform dynamic cluster analysis on all the monitoring data corresponding to each of the time periods, including:

[0013] storing all the monitoring data in each of the time periods in a data set in the form of a state matrix, wherein the monitoring data of different target locations are located in different rows of the state matrix;

[0014] The state matrix is ​​normalized as follows:

[0015]

[0016] Among them, s i represents the normalized monitoring data of the ith row, x i Represents the monitoring data of the i-th row;

[0017] Based on the normalized state matrix, n initial cluster centers are set, and each row of monitoring data s of the state matrix is ​​used as a sample;

[0018] The initial cluster center and sample s are iteratively calculated. If the following conditions are met: ||sz j (m)||<||sz i (m)||, (i, j = 1, 2, ..., n and i ≠ j), then s∈f j (m), where z i (m) represents the i-th cluster center at the m-th iteration; f j (m) represents the clustering domain of the jth cluster center at the mth iteration;

[0019] Calculate the new cluster center according to the following formula:

[0020]

[0021] Among them, Ni is the i-th clustering domain f i The number of samples in (m);

[0022] In satisfying z i (m+1)=z i After the condition (m), all samples are clustered into n categories.

[0023] Through the above technical solution, all monitoring data are clustered.

[0024] Furthermore, the clustering results obtained based on the dynamic clustering analysis remove similar monitoring data, including:

[0025] The distance d(x, c) from each sample to the center of its cluster is calculated as follows:

[0026] d(x,c)=(xc)[(xc)] T ;

[0027] Among them, the centroid c is the mean of each sample in the clustering domain;

[0028] For each of the target cluster centers, all corresponding samples are traversed, the sample corresponding to the smallest d(x, c) is retained, and the original monitoring data corresponding to the sample is used as the target monitoring data, and the original monitoring data corresponding to the remaining samples are deleted.

[0029] Through the above technical solution, similar monitoring data are removed to reduce the interference of this part of similar monitoring data on the health score, and at the same time, the calculation efficiency of the health score is improved.

[0030] Furthermore, the monitoring data after similar monitoring data are removed is classified using the pre-trained bidirectional long short-term memory network model, and the classification result is output, including:

[0031] The state matrix after removing similar monitoring data is input into the sequence input layer of the pre-trained bidirectional long short-term network model, and passes through the first BiLSTM layer, the first dropout layer, the second BiLSTM layer, the second dropout layer, the fully connected layer, the softmax function activation layer and the classification layer in sequence to output the classification result vector.

[0032] Through the above technical solution, the bidirectional long short-term memory network model is used to classify the remaining monitoring data to obtain the corresponding health status classification probability.

[0033] Furthermore, the monitoring data after removing similar monitoring data using the pre-trained bidirectional long short-term memory network model is classified, and before the classification result is output, the method includes:

[0034] The particle swarm algorithm is used to optimize the target hyperparameters in the bidirectional long short-term network model, wherein the target hyperparameters include the number of hidden units in the first BiLSTM layer, the dropout rate of the first dropout layer, the number of hidden units in the second BiLSTM layer, the dropout rate of the second dropout layer, the initial learning rate, the maximum number of iterations, and the mini-batch size.

[0035] Through the above technical solution, the particle swarm algorithm is used to optimize the hyperparameters of the bidirectional long-short-term network model to improve the classification accuracy of the bidirectional long-short-term network model.

[0036] Furthermore, the use of a particle swarm algorithm to optimize the target hyperparameters in the bidirectional long-term and short-term network model includes:

[0037] Based on the value range of the target hyperparameter [h min ,h max ], initialize the speed and position of each particle in the population, where the initial value is calculated as follows:

[0038] h=rand*(h max -h min )+h min ;

[0039] The optimal position of the particle at the initial time p best As the fitness of the first particle in the particle swarm, the fitness of the particle is calculated as follows:

[0040]

[0041] Among them, Y pred is the label classified by the BiLSTM network model, Y test is the label of the test set;

[0042] Set the particle's global optimal position g best is min(p best );

[0043] Based on the preset update rules, the position and velocity of the particle are updated, and the fitness of the updated particle is compared with the historical individual optimal position p of the particle. best The position corresponding to the smaller fitness is selected as the particle's p best ; Compare the fitness of the updated particle with the global optimal position g of the historical particle best The position corresponding to the smaller fitness is selected as the particle's g best ;

[0044] Continue to iterate until the maximum number of iterations of the particle swarm is reached. max, and the optimization result is used as the corresponding value of the target hyperparameter.

[0045] Furthermore, the preset coal mining machine health status evaluation rules include:

[0046] The monitoring data collected by the coal mining machine in each time period is determined as the evaluation factor, and based on all the evaluation factors, the time period domain U is obtained:

[0047]

[0048] All the time periods are divided into q subsets in chronological order, as follows:

[0049]

[0050] The domain of comment levels for each level of coal mining machine health status is set as:

[0051] X = {state 1, state 2, state 3, state 4, state 5}

[0052] The fuzzy comprehensive evaluation is performed using all the classification results to calculate the health score of the coal mining machine in the current time period, including:

[0053] The classification result vectors corresponding to different time periods are combined in chronological order to establish a fuzzy relationship matrix;

[0054] Based on the time period domain and comment level domain, the evaluation matrix R is obtained from the fuzzy relationship matrix as follows:

[0055]

[0056] Set the first-level evaluation weight distribution vector A as follows: median (q):

[0057]

[0058] Set the first-level evaluation weight distribution vector A as follows: res :

[0059]

[0060] Calculate each element b in the first-level evaluation vector B as follows:

[0061]

[0062] Among them, a is the element in the first-level evaluation weight allocation vector, and r is the element in the evaluation matrix R;

[0063] The first-level evaluation vector B is calculated as follows:

[0064]

[0065] in, Representing fuzzy operators A median (q) only appears in B median In the calculation of (q), the other first-level evaluation vectors are all A res calculate;

[0066] The single factor judgment matrix T of the time period domain U is calculated as follows:

[0067]

[0068] Set the secondary evaluation weight distribution vector as follows:

[0069]

[0070] The secondary evaluation vector is calculated as follows:

[0071]

[0072] The score vector S of each health state of the coal mining machine is calculated as follows:

[0073] S=(0,25,50,75,100);

[0074] Calculate the health score of the coal mining machine as follows:

[0075]

[0076] In addition, the technical problem to be solved by the present invention is to provide an intelligent evaluation device for the health status of a coal mining machine, which comprises:

[0077] An acquisition unit, used for acquiring monitoring data of each target position of the coal mining machine in a plurality of adjacent time periods;

[0078] A removal unit, configured to perform a dynamic cluster analysis on all the monitoring data corresponding to each of the time periods using a K-means algorithm, and remove similar monitoring data based on a clustering result obtained by the dynamic cluster analysis;

[0079] A classification unit, used to classify the monitoring data after similar monitoring data are removed using a pre-trained bidirectional long short-term memory network model to obtain a classification result;

[0080] The scoring unit is used to perform fuzzy comprehensive evaluation based on the preset coal mining machine health status evaluation rules using all the classification results to calculate the coal mining machine health score in the current time period.

[0081] Through the above technical scheme, firstly, by collecting monitoring data of multiple continuous time periods of the coal mining machine and removing redundant data in all monitoring data, the interference of similar monitoring data on the health status assessment results is reduced and the efficiency of outputting the health score results is improved. Then, the monitoring data of each target position in all time periods is classified through a bidirectional long short-term memory network model to obtain the probability of the health status classification of each monitoring data. Finally, a fuzzy comprehensive evaluation is performed using the probability of the health status classification to calculate the health score of the coal mining machine. Since this application obtains the health score of the coal mining machine in the current time period through monitoring data of multiple continuous time periods, it fully takes into account the characteristic that the health status change process of the coal mining machine is continuous in time, and faces the influence of time information on the health score, thereby ensuring the accuracy of the health score of the coal mining machine.

[0082] In addition, an embodiment of the present invention provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the intelligent assessment method for the health status of the coal mining machine described in the first aspect above is implemented.

[0083] In addition, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the intelligent assessment method for the health status of a coal mining machine described in the first aspect above.

[0084] The embodiment of the present invention discloses a method, device and related components for intelligently evaluating the health status of a coal mining machine, wherein the method includes: obtaining monitoring data of each target position of the coal mining machine in multiple adjacent time periods; using the K-means algorithm to perform dynamic cluster analysis on all the corresponding monitoring data in each of the time periods, and removing similar monitoring data based on the clustering results obtained by the dynamic clustering analysis; using a pre-trained bidirectional long short-term memory network model to classify the monitoring data after removing similar monitoring data, and output the classification results; based on the preset coal mining machine health status evaluation rules, using all the classification results to perform fuzzy comprehensive evaluation, and calculating the coal mining machine health score of the current time period. The method is based on the monitoring data of the continuous time period of the coal mining machine, and through bidirectional long short-term network model classification and fuzzy comprehensive evaluation, it improves the rationality and accuracy of the health score of the coal mining machine in the current time period. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0086] Figure 1 A schematic diagram of a process flow of a method for intelligently evaluating the health status of a coal mining machine provided by an embodiment of the present invention;

[0087] Figure 2 A schematic diagram of the structure of a bidirectional long short-term memory network model in the intelligent evaluation method for the health status of a coal mining machine provided in an embodiment of the present invention;

[0088] Figure 3 A schematic block diagram of a device for intelligently evaluating the health status of a coal mining machine provided by an embodiment of the present invention;

[0089] Figure 4 A schematic block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0090] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0091] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0092] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0093] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0094] See also Figure 1 , Figure 1 A flow chart of a method for intelligently evaluating the health status of a coal mining machine provided by an embodiment of the present invention; Figure 1 As shown, the method includes steps S101 to S104.

[0095] S101, obtaining monitoring data of each target position of the coal mining machine in multiple adjacent time periods;

[0096] S102, performing dynamic cluster analysis on all the monitoring data corresponding to each of the time periods using a K-means algorithm, and removing similar monitoring data based on clustering results obtained by the dynamic cluster analysis;

[0097] S103, using a pre-trained bidirectional long short-term memory network model to classify the monitoring data after similar monitoring data are removed, and outputting a classification result;

[0098] S104. Based on the preset coal mining machine health status evaluation rules, all the classification results are used to perform fuzzy comprehensive evaluation to calculate the coal mining machine health score for the current time period.

[0099] In this embodiment, the existing health status assessment method of the coal mining machine only evaluates the health status of the coal mining machine in the current time period by collecting monitoring information of the coal mining machine's operation in the current time period. This method does not fully take into account the characteristic that the health status change process of the coal mining machine is continuous in multiple time periods, that is, it ignores the impact of multi-time period information on the health score. Therefore, this application collects monitoring data of the coal mining machine at each target position in multiple adjacent time periods, and uses the K-means algorithm to remove similar monitoring data from all the collected monitoring data, thereby reducing the interference of a large amount of similar monitoring data on the health score, and to a certain extent improves the scoring efficiency and reduces the calculation time; uses a pre-trained bidirectional long short-term memory network model to classify the monitoring data after removing similar monitoring data to obtain a classification result, and finally, based on the preset coal mining machine health status evaluation rules, uses the classification result to perform fuzzy comprehensive evaluation to obtain the health score of the coal mining machine in the current time period.

[0100] The evaluation method of the present application takes into account the impact of the coal mining machine's operating status at different time periods on the coal mining machine's health status, and quantifies the final score of the evaluation, making the evaluation result, i.e., the final health score, more reasonable.

[0101] It should be understood that the coal mining machine can obtain the above monitoring data through sensors or other monitoring equipment, so this application will not elaborate on it. Specifically, the monitoring data of this application include but are not limited to: traction motor temperature data, traction motor current data, traction motor vibration data, cutting motor temperature data, cutting motor current data, cutting motor vibration data, heightening pump motor temperature data, heightening pump motor current data, heightening pump working pressure data, crushing motor current data, crushing motor temperature data. In other words, the target positions of the coal mining machine include but are not limited to: traction motor, cutting motor, heightening pump motor, crushing motor.

[0102] In step S101, monitoring data of various parts of the coal mining machine are collected within one minute at a sampling frequency of 100 Hz. In other words, each time period is 1 minute, and 6000 sampling point data (monitoring data) generated by various parts of the coal mining machine within one minute can be collected. However, in the actual evaluation process, the duration and sampling frequency of each time period can be other values, which are not limited in this application.

[0103] In a specific embodiment, the step S102 of “using the K-means algorithm to perform dynamic cluster analysis on all the monitoring data corresponding to each of the time periods” includes the following steps:

[0104] S10, storing all the monitoring data corresponding to each of the time periods in a data set in the form of a state matrix, wherein the monitoring data of different target locations are located in different rows of the state matrix;

[0105] In this embodiment, for example, all the traction motor temperature data (6000) collected in the current time period (for example, within one minute) are arranged in the first row of the state matrix in the order of collection time, and then all the traction motor current data collected in the current time period are arranged in the second row of the state matrix in the order of installation and collection time, and so on, to obtain the state matrix of the current time period, wherein each time period corresponds to a state matrix.

[0106] S11, normalize the state matrix according to the following formula:

[0107]

[0108] Among them, s i represents the normalized monitoring data of the ith row, x i Represents the monitoring data of the i-th row;

[0109] In this embodiment, the influence of dimensions between monitoring data can be eliminated by normalizing the state matrix.

[0110] S12, based on the normalized state matrix, setting n initial cluster centers, and taking each row of monitoring data s of the state matrix as a sample;

[0111] It should be noted that "n" in this step can also represent the number of samples remaining after removing similar monitoring samples. In a specific embodiment, the number of normalized samples s is set to k, and the initial clustering centers are set to n. Cluster analysis is performed by traversing n=(1, 2, ..., k-1). After each cluster analysis, the silhouette coefficient of the current clustering result is calculated to determine the value of n.

[0112] More specifically, the sample s under the current cluster is calculated as follows: i The average distance d to other samples in the same cluster domain a (i), d a (i) represents sample s i Dissimilarity within the clustering domain:

[0113]

[0114] Among them, N a For i The number of samples in the same cluster domain.

[0115] Calculate the sample s as follows i The average distance d to all samples in other cluster domains m bm (i)

[0116]

[0117] Among them, N bm is the number of samples in other clustering domains m; s mj are samples in other clustering domains m.

[0118] If we exclude sample s i If there are m clustering domains outside the clustering domain where it is located, then sample s can be defined i The dissimilarity between clustering domains is d b (i)

[0119] d b (i) = min{d b1 (i), d b2 (i), ..., d bm (i)} to calculate the sample s i The silhouette coefficient c i

[0120]

[0121] Among them, c i The larger the value, the greater the sample size. i The more reasonable the clustering is.

[0122] When the current initial cluster center is set to n, the silhouette coefficient C of the clustering result is calculated as follows: n :

[0123]

[0124] Among them, C n The larger it is, the more reasonable the clustering result is.

[0125] Set the initial number of cluster centers n to: clustering result silhouette coefficient C n The maximum number of initial cluster centers set, that is, max{C n} corresponding to n.

[0126] For ease of understanding, this application sets the number of normalized samples s to 10, and the initial cluster centers should be 1, 2, ..., 9 respectively for cluster analysis, that is, clustering once when n=2, clustering once when n=3, clustering once when n=4, ..., clustering once when n=9.

[0127] Assuming that n=4, clustering generates 4 cluster domains, each of which has 5 samples, as shown in Table 1 below:

[0128] Table 1

[0129]

[0130] Calculate the sample s1(s a1 ) to other samples s in the same cluster domain a2 ,s a3 ,s a4 ,s a5 The average distance d a (1) is:

[0131]

[0132] Calculate the average distance d from sample s1 to all samples in other clustering domains B1, B2, and B3 b (1) d b (2) d b (3):

[0133]

[0134]

[0135] The dissimilarity d between clustering domains of sample s1 b (1) is:

[0136] d b (1) = min{d b1 (1), d b2 (1), d b3 (1)}The silhouette coefficient c1 of s1 is:

[0137]

[0138] Similarly, we can calculate each sample s i The silhouette coefficient c i , that is, c1, c2, ..., c 10.

[0139] When n=4, the silhouette coefficient C4 of the clustering result is:

[0140]

[0141] Similarly, the silhouette coefficient C of the clustering result can be calculated when n is other values n , i.e. C2, C3, ..., C9. If max{C1, C2, ..., C9} = C4, the number of initial cluster centers is set to 4.

[0142] S13, iteratively calculate the initial cluster center and sample s, if the following conditions are met: ||sz j (m)||<||sz i (m)||, (i, j = 1, 2, ..., n and i ≠ j), then s∈f j (m), where z i (m) represents the i-th cluster center at the m-th iteration; f j (m) represents the clustering domain of the jth cluster center at the mth iteration;

[0143] It should be noted that the "s" in the formula of this step represents the monitoring data s of each row after the state matrix is ​​normalized.

[0144] S14. Calculate the new cluster center according to the following formula:

[0145]

[0146] Among them, N i is the i-th clustering domain f i The number of samples in (m);

[0147] S15, in the satisfaction of z i (m+1)=z i After the condition (m), all samples are clustered into n categories.

[0148] In this embodiment, when z is satisfied i (m+1)=z i After the condition (m) is met, the algorithm converges and eventually n new cluster centers can be obtained.

[0149] In a specific embodiment, the step S102 of “removing similar monitoring data based on clustering results obtained by dynamic clustering analysis” includes the following steps:

[0150] S20, calculate the distance d(x, c) from each sample to the cluster center according to the following formula:

[0151] d(x,c)=(xc)[(xc)]T ;

[0152] Among them, the centroid c is the mean of each sample in the clustering domain; “[(xc)] T ” represents the matrix transpose of (xc).

[0153] S21. For each of the target cluster centers, traverse all corresponding samples, retain the sample corresponding to the smallest d(x, c), use the original monitoring data corresponding to the sample as the target monitoring data, and delete the original monitoring data corresponding to the remaining samples.

[0154] By removing similar monitoring data as redundant data through the embodiment of the present application, the accuracy of the final score of the health status of the coal mining machine can be effectively improved, and at the same time, the computing efficiency of subsequent links can be improved.

[0155] Combination Figure 2 In a specific embodiment, the step S103 includes the following steps:

[0156] S30. Input the state matrix after removing similar monitoring data into the sequence input layer of the pre-trained bidirectional long short-term network model, and pass through the first BiLSTM layer, the first discard layer, the second BiLSTM layer, the second discard layer, the fully connected layer, the softmax function activation layer and the classification layer in sequence, and output the classification result vector.

[0157] It should be noted that the Long Short-Term Memory (LSTM) network is a special type of Recurrent Neural Network (RNN), which makes up for the shortcoming of RNN in learning long-term information dependencies and is often used to process time series data. BiLSTM is composed of forward LSTM and backward LSTM, which overcomes the shortcoming of LSTM that it cannot encode information from back to front, and can better discover the dependencies of bidirectional long-term information.

[0158] The bidirectional long short-term memory network model of the present application is constructed by a sequence input layer (sequenceInputLayer), a first BiLSTM layer, a first dropout layer (dropoutLayer), a second BiLSTM layer, a second dropout layer (dropoutLayer), a fully connected layer (fullyConnectedLayer), a softmax function activation layer and a classification layer (classificationLayer). Specifically, the n-row state matrix after removing similar monitoring data is input into the pre-trained BiLSTM network model, wherein the input size of the input layer of the sequence input layer is set to n, and the first BiLSTM layer and the second BiLSTM layer are set to n. The number of hidden units in the iLSTM layer, the dropout rates of the first dropout layer and the second dropout layer are all calculated by the particle swarm algorithm when the network model is trained. The first BiLSTM layer outputs the complete sequence, the first dropout layer sets the input elements to 0 according to the first dropout rate, the second BiLSTM layer outputs the last time step of the sequence, the second dropout layer sets the input elements to 0 according to the second dropout rate, the output size of the fully connected layer is set to 5 (this application sets 5 health levels: severely ill, ill, general, sub-healthy, healthy), and the abstract feature vector Z=(z1, z2, z3, z4, z5) with a dimension of 5 is output, and then the softmax function (normalized exponential function) is applied to obtain the probability of multi-classification results:

[0159]

[0160] Where K is the number of health status levels of the coal mining machine; z i It indicates the model's prediction value for the sample belonging to level i;

[0161] Specifically, the classification result vector R = (r1, r2, r3, r4, r5) output by the softmax function activation layer, where r n (n=1, 2, 3, 4, 5) represents the probability that the state matrix Q belongs to class n.

[0162] It should be noted that the training data used in the pre-training process of the bidirectional long short-term memory network of the present application should be: the state matrix after removing similar monitoring data formed by the monitoring data of each target position when the coal mining machine is operating at different health status levels (i.e. the above-mentioned 5 health levels), and the current health status level number is used as the label of its corresponding state matrix Q.

[0163] In a specific embodiment, before step S103, the following steps are included:

[0164] S40. Optimize the target hyperparameters in the bidirectional long-short term network model using a particle swarm algorithm, wherein the target hyperparameters include the number of hidden units in the first BiLSTM layer, the dropout rate of the first dropout layer, the number of hidden units in the second BiLSTM layer, the dropout rate of the second dropout layer, the initial learning rate, the maximum number of iterations, and the mini-batch size (MiniBatchSize).

[0165] In this embodiment, the hyperparameters in the BiLSTM network model are optimized by the particle swarm algorithm (PSO algorithm) to improve the classification accuracy of the model. It should be noted that the hyperparameters directly affect the actual performance of the BiLSTM network model. Therefore, based on the good optimization ability of the PSO algorithm itself, for complex deep learning network models, it can quickly find the appropriate hyperparameter values ​​through iterative optimization calculation.

[0166] In a specific embodiment, the step S40 includes the following steps:

[0167] S50, based on the value range of the target hyperparameter [h min ,h max ], initialize the speed and position of each particle in the population, where the initial value is calculated as follows:

[0168] h=rand*(h max -h min )+h min ;

[0169] In this embodiment, rand represents a random value in a uniform distribution in the interval (0,1);

[0170] S51, the initial optimal position of the particle individual p best As the fitness of the first particle in the particle swarm, the fitness of the particle is calculated as follows:

[0171]

[0172] Among them, Y pred is the label of the bidirectional long-term and short-term network model classification, Y test is the label of the test set; “Y pred = = Y test " indicates whether the classification label and the test set label are equal; numel(Y test ) indicates that the number of test sets is calculated using the numel function, where the classification labels are the five preset health levels: 1 (severely ill), 2 (ill-healthy), 3 (normal), 4 (sub-healthy), and 5 (healthy).

[0173] S52, set the particle global optimal position g bestis min(p best );

[0174] S53, based on the preset update rules, update the position and speed of the particle, and compare the fitness of the updated particle with the historical individual optimal position p of the particle best The position corresponding to the smaller fitness is selected as the particle's p best ; Compare the fitness of the updated particle with the global optimal position g of the historical particle best The position corresponding to the smaller fitness is selected as the particle's g best ;

[0175] Through the comparison and replacement operation in step S53, the particle position that meets both the global optimum and the local optimum is found.

[0176] In this embodiment, the specific steps of "updating the positions and velocities of all particles" in step S53 are as follows:

[0177] S531. Update the particle position as follows:

[0178] X i,j (t+1)=X i,j (t)+V i,j (t+1);

[0179] Among them, the population X composed of n particles is (X1, X2, ..., X n ); the position of the i-th particle represents a D-dimensional vector X i =(X i1 , X2, ..., X iD ) T ; The velocity of the i-th particle represents a D-dimensional vector V i =(V i1 , V2, ..., V iD ) T ;

[0180] S91. Update the particle velocity as follows:

[0181]

[0182] Among them, c1 and c2 (both not less than 0) are acceleration constants; r1 and r2 are uniformly distributed random numbers and are both greater than 0 and less than 1; ω is the inertia weight coefficient; the position of the i-th particle represents a D-dimensional vector X i =(X i1 , X2, ..., X iD ) T ; The velocity of the i-th particle represents a D-dimensional vector V i=(V i1 , V2, ..., V iD ) T ; The individual extreme value is P i =(P i1 , P2, ..., P iD ) T ; The global extreme value is P g =(P g1 , P g2 , ..., P gD ) T ;

[0183] S532, the inertia weight coefficient ω is calculated according to the following formula:

[0184]

[0185] Among them, t represents the current iteration number, t max Indicates the maximum number of iterations of the particle swarm.

[0186] In this embodiment, the maximum inertia weight ω max =1.2; minimum inertia weight ω min =0.8.

[0187] S54, when the number of iterations reaches the preset maximum number of iterations of the particle swarm t max Then, the optimization result is used as the value of the corresponding target hyperparameter.

[0188] It should be noted that the fuzzy comprehensive evaluation method is a method of comprehensively evaluating things from multiple factors (indicators) by applying the principle of fuzzy relationship synthesis. In a specific embodiment, the coal mining machine health status evaluation rule in step S104 includes the following steps:

[0189] S60, determining the monitoring data collected by the coal mining machine in each time period as the evaluation factor, and obtaining the time period domain U based on all the evaluation factors:

[0190]

[0191] In this embodiment, the monitoring data collected in the current time period of the coal mining machine, for example, one minute, is recorded as "time i", and data of j minutes before and after "time i" are collected respectively. The number of all time periods k = 1 + 2j (k is an integer multiple of q), and then all the collected monitoring data are used to determine the evaluation factors.

[0192] S61, divide all the time period domains into q subsets in chronological order, as follows:

[0193]

[0194] It should be noted that q is an odd number not less than 3.

[0195] S62. Set the comment level domains for each level of the coal mining machine health status as follows:

[0196] X = {state 1, state 2, state 3, state 4, state 5}

[0197] Specifically, this application divides the health status of the coal mining machine into state 1, state 2, state 3, state 4 and state 5 according to the mapping relationship between the health status level and health status of the coal mining machine shown in Table 2, and uses the state sequence number as the label of the corresponding state matrix Q.

[0198] Table 2

[0199]

[0200] Among them, the newly commissioned coal mining machines are generally in a healthy state, i.e., state 5 in Table 2.

[0201] In a specific embodiment, the step S104 includes the following steps:

[0202] S70, performing fuzzy comprehensive evaluation using all the classification results to calculate the health score of the coal mining machine in the current time period, including:

[0203] S71. Combine the classification result vectors corresponding to different time periods in chronological order to establish a fuzzy relationship matrix. Specifically, the fuzzy relationship matrix is ​​shown in Table 3 below, where (0≤r≤1)

[0204] Table 3

[0205]

[0206]

[0207] S72, based on the time period domain and the comment level domain, obtain the evaluation matrix R from the fuzzy relationship matrix according to the following formula:

[0208]

[0209] It should be added that “time i” is the current time period, which best reflects the health status of the coal mining machine in the current time period.

[0210] S73, set the first-level evaluation weight allocation vector A according to the following formula median (q):

[0211]

[0212] Among them, Amedian(q) The element at the median(k / q)th position in The elements in the remaining positions are median means calculating the median.

[0213] S74, set the first-level evaluation weight allocation vector A according to the following formula res :

[0214]

[0215] Among them, A res The length of is k / q, and each element in the vector is

[0216] S75. Calculate each element b in the first-level evaluation vector B as follows:

[0217]

[0218] Among them, a is the element in the first-level evaluation weight allocation vector, and r is the element in the evaluation matrix R;

[0219] S76. Calculate the first-level evaluation vector B according to the following formula:

[0220]

[0221] Among them, the symbol Representing fuzzy operators A median(q) Only appears in B median(q) In the calculation of , the other first-level evaluation vectors are all A res calculate;

[0222] It should be noted that the weighted average fuzzy operator It is more reasonable to consider all factors according to their weights, so this operator is used to calculate each element b in the first-level evaluation vector B:

[0223] S77. Calculate the single factor judgment matrix T of the time period domain U as follows:

[0224]

[0225] S78. Set the secondary evaluation weight allocation vector according to the following formula

[0226]

[0227] in, The element at the median(q)th position in The elements in the remaining positions are median means calculating the median.

[0228] In this embodiment, since the time period of “time i” is: It can best reflect the state of the coal mining machine in the current period, so the secondary evaluation weight distribution vector is set according to the above formula

[0229] S79, calculate the secondary evaluation according to the following formula

[0230]

[0231] Among them, the symbol Denotes the fuzzy operator M(,·⊕).

[0232] S80. Calculate the fractional vector S of each state of the coal mining machine according to the following formula:

[0233] S=(0,25,50,75,100);

[0234] Among them, the elements at each position of the score vector S represent the score values ​​of state 1 to state 5 respectively.

[0235] S81. Calculate the health score of the coal mining machine according to the following formula:

[0236]

[0237] For ease of understanding, this application records the current time period "time 5" of the coal mining machine, and collects data for 4 minutes before and after "time 5" (j=4), so these 9 (k=9) consecutive time periods are used to illustrate the calculation of the health score of the coal mining machine:

[0238] The time period domain U is:

[0239] {time 1, time 2, time 3, time 4, time 5, time 6, time 7, time 8, time 9}

[0240] Among them, time 5 is the current time period.

[0241] Divide U into 3 (q=3) subsets in chronological order:

[0242]

[0243] The corresponding fuzzy relationship matrix is ​​shown in Table 4:

[0244] Table 4

[0245] State 1 State 2 State 3 State 4 State 5 Time 1 <![CDATA[r 11 ]]> <![CDATA[r 12 ]]> <![CDATA[r 13 ]]> <![CDATA[r 14 ]]> <![CDATA[r 15 ]]> Time 2 <![CDATA[r 21 ]]> <![CDATA[r 22 ]]> <![CDATA[r 23 ]]> <![CDATA[r 24 ]]> <![CDATA[r 25 ]]> Time 3 <![CDATA[r 31 ]]> <![CDATA[r 32 ]]> <![CDATA[r 33 ]]> <![CDATA[r 34 ]]> <![CDATA[r 35 ]]> Time 4 <![CDATA[r 41 ]]> <![CDATA[r 42 ]]> <![CDATA[r 43 ]]> <![CDATA[r 44 ]]> <![CDATA[r 45 ]]> Time 5 <![CDATA[r 51 ]]> <![CDATA[r 52 ]]> <![CDATA[r 53 ]]> <![CDATA[r 54 ]]> <![CDATA[r 55 ]]> Time 6 <![CDATA[r 61 ]]> <![CDATA[r 62 ]]> <![CDATA[r 63 ]]> <![CDATA[r 64 ]]> <![CDATA[r 65 ]]> Time 7 <![CDATA[r 71 ]]> <![CDATA[r 72 ]]> <![CDATA[r 73 ]]> <![CDATA[r 74 ]]> <![CDATA[r 75 ]]> Time 8 <![CDATA[r 81 ]]> <![CDATA[r 82 ]]> <![CDATA[r 83 ]]> <![CDATA[r 84 ]]> <![CDATA[r 85 ]]> Time 9 <![CDATA[r 91 ]]> <![CDATA[r 92 ]]> <![CDATA[r 93 ]]> <![CDATA[r 94 ]]> <![CDATA[r 95 ]]>

[0246] According to U and X, the evaluation matrix R can be obtained respectively i(i=1, 2, 3) as follows:

[0247]

[0248] Set the first-level evaluation weight distribution vector A median(q) (A2) is:

[0249]

[0250] The first-level evaluation weight distribution vector A except A2 res The length of each vector is 3, and each element in the vector is

[0251] Calculate each element b in the first-level evaluation vector B as follows:

[0252] Calculate the first-level evaluation vector B i (i=1, 2, 3):

[0253]

[0254] The single factor judgment matrix T of U is:

[0255]

[0256] Set the secondary evaluation weight distribution vector for:

[0257]

[0258] Calculate the secondary evaluation as follows

[0259]

[0260] Finally, the shearer health score N is:

[0261]

[0262] The embodiment of the present invention also provides a coal mining machine health status intelligent assessment device, which is used to execute any embodiment of the above-mentioned coal mining machine health status intelligent assessment method. Figure 3 , Figure 3 It is a schematic block diagram of a coal mining machine health status intelligent assessment device provided by an embodiment of the present invention.

[0263] like Figure 3 As shown, the coal mining machine health status intelligent assessment device 200 includes:

[0264] An acquisition unit 201 is used to acquire monitoring data of each target position of the coal mining machine in a plurality of adjacent time periods;

[0265] A removal unit 202, configured to perform a dynamic cluster analysis on all the monitoring data corresponding to each of the time periods using a K-means algorithm, and remove similar monitoring data based on a clustering result obtained by the dynamic cluster analysis;

[0266] A classification unit 203 is used to classify all the remaining monitoring data using a pre-trained bidirectional long short-term memory network model to obtain a classification result;

[0267] The scoring unit 204 is used to perform fuzzy comprehensive evaluation based on the preset coal mining machine health status evaluation rules using all the classification results to calculate the coal mining machine health score in the current time period.

[0268] The device first collects monitoring data of the coal mining machine for multiple continuous time periods and removes redundant data in all monitoring data, thereby reducing the interference of similar monitoring data on the health status assessment results and improving the efficiency of outputting health score results. The device then classifies the monitoring data of each target position in all time periods through a bidirectional long short-term memory network model to obtain the probability of the health status classification of each monitoring data. Finally, the probability of the health status classification is used to perform fuzzy comprehensive evaluation to calculate the health score of the coal mining machine. Since the present application obtains the health score of the coal mining machine in the current time period through the monitoring data of multiple continuous time periods, it fully takes into account the characteristic that the health status change process of the coal mining machine is continuous in time, faces the influence of time information on the health score, and ensures the accuracy of the health score of the coal mining machine.

[0269] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0270] The above-mentioned coal mining machine health status intelligent assessment device can be implemented in the form of a computer program, which can be used in Figure 4 Runs on the computer device shown.

[0271] See also Figure 4 , Figure 4 11 is a schematic block diagram of a computer device provided by an embodiment of the present invention. The computer device 1100 is a server, which may be an independent server or a server cluster composed of multiple servers.

[0272] See also Figure 4The computer device 1100 includes a processor 1102 , a memory, and a network interface 1105 connected via a system bus 1101 , wherein the memory may include a non-volatile storage medium 1103 and an internal memory 1104 .

[0273] The non-volatile storage medium 1103 can store an operating system 11031 and a computer program 11032. When the computer program 11032 is executed, the processor 1102 can execute the intelligent health status assessment method of the coal mining machine.

[0274] The processor 1102 is used to provide computing and control capabilities to support the operation of the entire computer device 1100 .

[0275] The internal memory 1104 provides an environment for the operation of the computer program 11032 in the non-volatile storage medium 1103. When the computer program 11032 is executed by the processor 1102, the processor 1102 can execute the intelligent assessment method for the health status of the coal mining machine.

[0276] The network interface 1105 is used for network communication, such as providing data information transmission, etc. Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device 1100 to which the solution of the present invention is applied. The specific computer device 1100 may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0277] Those skilled in the art will understand that Figure 4 The embodiments of the computer device shown in the figure do not constitute a limitation on the specific composition of the computer device. In other embodiments, the computer device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. For example, in some embodiments, the computer device may only include a memory and a processor. In such embodiments, the structure and function of the memory and the processor are the same as those of the embodiment of the present invention. Figure 4 The embodiments shown are consistent and will not be described again here.

[0278] It should be understood that in the embodiment of the present invention, the processor 1102 may be a central processing unit (CPU), and the processor 1102 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0279] In another embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium may be a non-volatile computer-readable storage medium. The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the method for intelligently assessing the health status of a coal mining machine according to an embodiment of the present invention is implemented.

[0280] The storage medium is a physical, non-transient storage medium, for example, it can be a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk, etc., which can store program codes.

[0281] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0282] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. An intelligent assessment method for the health status of coal mining machines, It is characterized in that include: Acquire monitoring data of each target position of the coal mining machine in multiple adjacent time periods; Performing dynamic cluster analysis on all the monitoring data corresponding to each of the time periods using a K-means algorithm, and removing similar monitoring data based on clustering results obtained by the dynamic cluster analysis; Using a pre-trained bidirectional long short-term memory network model to classify the monitoring data after removing similar monitoring data, and output the classification result; Based on the preset coal mining machine health status evaluation rules, all the classification results are used to perform fuzzy comprehensive evaluation to calculate the coal mining machine health score for the current time period; The method of using the K-means algorithm to perform dynamic cluster analysis on all the monitoring data corresponding to each of the time periods includes: storing all the monitoring data in each of the time periods in a data set in the form of a state matrix, wherein the monitoring data of different target locations are located in different rows of the state matrix; The state matrix is ​​normalized as follows: Among them, s i represents the normalized monitoring data of the ith row, x i Represents the monitoring data of the i-th row; Based on the normalized state matrix, n initial cluster centers are set, and each row of monitoring data s of the state matrix is ​​used as a sample; The initial cluster center and sample s are iteratively calculated, if the following conditions are met: ‖sz j (m)‖<‖sz i (m)‖, (i, j = 1, 2, ..., n and i ≠ j), then s∈f j (m), where z i (m) represents the i-th cluster center at the m-th iteration; f j (m) represents the clustering domain of the jth cluster center at the mth iteration; Calculate the new cluster center according to the following formula: Among them, N i is the i-th clustering domain f i The number of samples in (m); In satisfying z i (m+1)=z i After the condition (m), all samples are clustered into n categories; The clustering results obtained based on the dynamic clustering analysis remove similar monitoring data, including: The distance d(x, c) from each sample to the center of its cluster is calculated as follows: d(x,c)=(x-c)[(x-c)] T ; Among them, the centroid c is the mean of each sample in the clustering domain; For each target cluster center, traverse all corresponding samples, retain the sample corresponding to the smallest d(x, c), and use the original monitoring data corresponding to the sample as the target monitoring data, and delete the original monitoring data corresponding to the remaining samples.

2. According to claim 1, the intelligent evaluation method for the health status of a coal mining machine, It is characterized in that The pre-trained bidirectional long short-term memory network model is used to classify the monitoring data after removing similar monitoring data, and the classification result is output, including: The state matrix after removing similar monitoring data is input into the sequence input layer of the pre-trained bidirectional long short-term network model, and passes through the first BiLSTM layer, the first dropout layer, the second BiLSTM layer, the second dropout layer, the fully connected layer, the softmax function activation layer and the classification layer in sequence to output the classification result vector.

3. The method for intelligently evaluating the health status of a coal mining machine according to claim 2, It is characterized in that The monitoring data after similar monitoring data are removed is classified by using the pre-trained bidirectional long short-term memory network model, and before the classification result is output, the method includes: The particle swarm algorithm is used to optimize the target hyperparameters in the bidirectional long short-term network model, wherein the target hyperparameters include the number of hidden units in the first BiLSTM layer, the dropout rate of the first dropout layer, the number of hidden units in the second BiLSTM layer, the dropout rate of the second dropout layer, the initial learning rate, the maximum number of iterations, and the mini-batch size.

4. The method for intelligently evaluating the health status of a coal mining machine according to claim 3, It is characterized in that The method of optimizing the target hyperparameters in the bidirectional long-short term network model by using a particle swarm algorithm includes: Based on the value range of the target hyperparameter [h min ,h max ], initialize the speed and position of each particle in the population, where the initial value is calculated as follows: h=rand*(h max -h min )+h min ; The optimal position of the particle at the initial time p best As the fitness of the first particle in the particle swarm, the fitness of the particle is calculated as follows: Among them, Y pred is the label classified by the BiLSTM network model, Y test is the label of the test set; set the global optimal position of the particle g best is min(p best ); Based on the preset update rules, the position and velocity of the particle are updated, and the fitness of the updated particle is compared with the historical individual optimal position p of the particle. best The position corresponding to the smaller fitness is selected as the particle's p best ; Compare the fitness of the updated particle with the global optimal position g of the historical particle best The position corresponding to the smaller fitness is selected as the particle's g best ; Continue to iterate until the maximum number of iterations of the particle swarm is reached. max , and the optimization result is used as the corresponding value of the target hyperparameter.

5. The method for intelligently evaluating the health status of a coal mining machine according to claim 1, It is characterized in that The preset coal mining machine health status evaluation rules include: The monitoring data collected by the coal mining machine in each time period is determined as the evaluation factor, and based on all the evaluation factors, the time period domain U is obtained: All the time periods are divided into q subsets in chronological order, as follows: The domain of comment levels for each level of coal mining machine health status is set as: X = {state 1, state 2, state 3, state 4, state 5} The fuzzy comprehensive evaluation is performed using all the classification results to calculate the health score of the coal mining machine in the current time period, including: The classification result vectors corresponding to different time periods are combined in chronological order to establish a fuzzy relationship matrix; Based on the time period domain and comment level domain, the evaluation matrix R is obtained from the fuzzy relationship matrix as follows: Set the first-level evaluation weight distribution vector A as follows: median(q) : Set the first-level evaluation weight distribution vector A as follows: res : Calculate each element b in the first-level evaluation vector B as follows: Among them, a is the element in the first-level evaluation weight allocation vector, and r is the element in the evaluation matrix R; The first-level evaluation vector B is calculated as follows: Among them, "°" represents the fuzzy operator A median(q) Only appears in B median(q) In the calculation of , the other first-level evaluation vectors are all A res calculate; The single factor judgment matrix T of the time period domain U is calculated as follows: Set the secondary evaluation weight distribution vector as follows: The secondary evaluation vector is calculated as follows: The score vector S of each health state of the coal mining machine is calculated as follows: S = (0, 25, 50, 75, 100); Calculate the health score of the coal mining machine as follows:

6. An intelligent assessment device for the health status of coal mining machines, It is characterized in that include: An acquisition unit, used to acquire monitoring data of each target position of the coal mining machine in a plurality of adjacent time periods; A removal unit, configured to perform a dynamic cluster analysis on all the monitoring data corresponding to each of the time periods using a K-means algorithm, and remove similar monitoring data based on a clustering result obtained by the dynamic cluster analysis; A classification unit, used to classify the monitoring data after similar monitoring data are removed using a pre-trained bidirectional long short-term memory network model to obtain a classification result; A scoring unit, used to perform fuzzy comprehensive evaluation using all the classification results based on a preset coal mining machine health status evaluation rule, and calculate the coal mining machine health score for the current time period; When the removal unit uses the K-means algorithm to perform dynamic cluster analysis on all the monitoring data corresponding to each of the time periods, it is specifically used to: storing all the monitoring data in each of the time periods in a data set in the form of a state matrix, wherein the monitoring data of different target locations are located in different rows of the state matrix; The state matrix is ​​normalized as follows: Among them, s i represents the normalized monitoring data of the ith row, x i Represents the monitoring data of the i-th row; Based on the normalized state matrix, n initial cluster centers are set, and each row of monitoring data s of the state matrix is ​​used as a sample; The initial cluster center and sample s are iteratively calculated. If the following conditions are met: ||sz j (m)||<||sz i (m)||, (i,j=1,2,...,n and i≠j), then s∈f j (m), where z i (m) represents the i-th cluster center at the m-th iteration; f j (m) represents the clustering domain of the jth cluster center at the mth iteration; Calculate the new cluster center according to the following formula: Among them, N i is the i-th clustering domain f i The number of samples in (m); In satisfying z i (m+1)=z i After the condition (m), all samples are clustered into n categories; When removing similar monitoring data based on the clustering results obtained by dynamic clustering analysis, the removal unit is specifically used to: The distance d(x, c) from each sample to the center of its cluster is calculated as follows: d(x,c)=(x-c)[(x-c)] T ; Among them, the centroid c is the mean of each sample in the clustering domain; For each target cluster center, traverse all corresponding samples, retain the sample corresponding to the smallest d(x, c), and use the original monitoring data corresponding to the sample as the target monitoring data, and delete the original monitoring data corresponding to the remaining samples.

7. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the computer program, the method for intelligently assessing the health status of a coal mining machine according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the intelligent assessment method for the health status of a coal mining machine as described in any one of claims 1 to 5.

Citation Information

Patent Citations

  • Improved optimization classification method for transformer state evaluation

    CN112580198A

  • Method and system for predicting residual service life of multiple rotary assemblies and storage medium

    CN114707431A