A method and device for determining the fault type of a cut tobacco drying machine in a cigarette factory

Abstract

The application provides a cigarette factory cut tobacco drying machine equipment fault type determination method and device, the method comprises the following steps: obtaining a first data set to be determined for fault type; obtaining a second data set from all historical sensor data features affecting the operation of the cut tobacco drying machine; using a fuzzy C-means clustering-based method, clustering the second data set L times to obtain L clustering results; using a hash table to construct an offline fault index library according to the L clustering results; and determining the fault type of the first data set according to the offline fault index library. The method provided by the application can quickly and efficiently determine the fault type of the first data set.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and apparatus for determining the fault type of a cigarette drying machine. Background Technology

[0002] In the tobacco production process, the stable operation of the tobacco drying machine plays a crucial role in the quality of the final tobacco product. In traditional tobacco factory equipment maintenance, due to the complexity of the cascaded system, the massive amount of data from various sensors, and the correlation between fault sources, accurately and quickly locating the fault source, determining the fault type, and performing appropriate maintenance when a production equipment malfunctions, given the vast amount of sensor data, heavily relies on the maintenance personnel's experience.

[0003] Currently, existing methods for determining whether tobacco factory equipment is malfunctioning by detecting sensor data typically employ wavelet analysis, SOD isolated point detection, and fault tree methods to perform fault detection and type determination on sensor data. However, when processing such high-dimensional data, there are drawbacks such as excessive time consumption and the inability to promptly provide fault determination results to maintenance personnel. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a method and device for determining the fault type of cigarette factory drying machine equipment that can promptly feed back fault results to maintenance personnel.

[0005] A method for determining the fault type of a cigarette factory drying machine includes:

[0006] Step 1: Obtain the first dataset for determining the type of fault; the first dataset consists of various sensor data detected by sensors during the operation of the wire drying machine.

[0007] Step 2: Obtain the second dataset [x1, x2, x3, x4, ..., x] from all historical sensor data features affecting the operation of the wire drying machine. i .,x N The second dataset is the fault dataset corresponding to when the wire drying machine malfunctions, x i This is the i-th fault sample point;

[0008] Step 3: Using the fuzzy C-means clustering method, perform L clustering operations on the second dataset to obtain L clustering results; each clustering result includes K clusters, and each cluster corresponds to a cluster center;

[0009] Step 4: Based on the L clustering results, construct an offline fault index library using hash tables; the offline fault index library includes: L hash mapping tables, each hash mapping table corresponding to one of the clustering results;

[0010] Step 5: Determine the fault type of the first dataset based on the offline fault index library.

[0011] Furthermore, in the method for determining the fault type of the cigarette factory drying machine as described above, step three includes:

[0012] Step 31: Randomly initialize the membership matrix u based on the number of clusters K and the number of data points N in the second dataset. ij This yields the initialized membership matrix; where i represents the i-th row of the membership matrix and j represents the j-th column of the membership matrix.

[0013] Step 32: Based on the membership factor m and the initialized membership matrix, perform preliminary clustering on the second dataset to obtain the initial cluster centers cj;

[0014] Step 33: Based on the initial cluster centers c j Update the membership matrix;

[0015] Step 34: Repeat step 33 until the absolute difference between each element in each row of the current membership matrix and the corresponding element in the corresponding row of the previous iteration membership matrix is ​​less than the preset membership threshold ε, thus obtaining the final membership matrix.

[0016] Step 35: Use the final membership matrix as the first clustering result of the second dataset. The first clustering result includes K clusters, and each cluster includes several fault sample points.

[0017] Step 36: Fix the number of cluster categories and the membership threshold. Use δ as the perturbation amplitude factor to randomly perturb the membership factor L-1 times. Repeat steps 31-35 with the membership factor after each perturbation to obtain the corresponding L-1 clustering results.

[0018] Furthermore, in the method for determining the fault type of the cigarette factory drying machine as described above, step four includes:

[0019] Step 41: Select one clustering result from the L clustering results;

[0020] Step 42: Place the K clusters corresponding to the selected clustering results into K hash buckets respectively;

[0021] Step 43: Use a hash table to store the cluster center set C1 = {c1, c2, ..., c3} corresponding to the K clusters. k The cluster centers are mapped to the K hash buckets to construct a cluster center-hash bucket mapping hash table;

[0022] Step 44: Repeat steps 41-43 to map the remaining clustering results one by one using the hash table to obtain L-1 cluster center-hash bucket mapping hash tables;

[0023] Step 45: Use the L cluster center-hash bucket mapping hash tables obtained from Step 43 and Step 44 as the offline fault index library.

[0024] Furthermore, in the method for determining the fault type of the cigarette factory drying machine as described above, step five includes:

[0025] Step 51: Select one of the L cluster center-hash bucket mapping hash tables;

[0026] Step 52: Construct a query vector based on the first dataset, and map the query vector to the cluster center set C1 = {c1, c2, ..., c3} in the selected cluster center-hash bucket mapping hash table. k Perform Euclidean distance calculations on each cluster, select the cluster center with the closest Euclidean distance as the target cluster center, and use the clusters in the hash buckets mapped by the target cluster center as candidate clusters for this query processing.

[0027] Step 53: Repeat step 52, and sequentially traverse the remaining hash tables in the L cluster center-hash bucket mapping hash tables to obtain L-1 candidate clusters;

[0028] Step 54: Perform a union of the L candidate clusters obtained in Steps 52 and 53 to obtain the target candidate set;

[0029] Step 54: Calculate the Euclidean distance between the query vector constructed from the first dataset and each fault sample point in the target candidate set, and select the data sample with the closest Euclidean distance as the target data sample.

[0030] Step 55: Determine the fault type of the first dataset based on the fault type corresponding to the target data sample.

[0031] Furthermore, in the method for determining the fault type of the cigarette factory drying machine as described above, the sensor data includes: motor temperature, motor speed, pipeline humidity, pipeline temperature, valve opening degree, and pipeline pressure.

[0032] A fault type determination device for a cigarette factory drying machine includes:

[0033] The acquisition unit is used to acquire a first dataset of fault types to be determined; the first dataset consists of various sensor data detected by sensors during the operation of the wire drying machine.

[0034] The acquisition unit further acquires a second dataset [x1, x2, x3, x4, ..., x] from all historical sensor data features affecting the operation of the wire drying machine. i .,x N The second dataset is the fault dataset corresponding to when the wire drying machine malfunctions, x i This is the i-th fault sample point;

[0035] The clustering unit is used to perform L clustering operations on the second dataset using a fuzzy C-means clustering method, resulting in L clustering results; each clustering result includes K clusters, and each cluster corresponds to a cluster center;

[0036] A construction unit is used to construct an offline fault index library using hash tables based on the L clustering results; the offline fault index library includes: L hash mapping tables, each hash mapping table corresponding to one of the clustering results;

[0037] The determining unit is used to determine the fault type of the first dataset based on the offline fault index library.

[0038] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the fault type determination method for cigarette factory drying machine as described above.

[0039] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the fault type determination method for cigarette factory drying machine equipment as described above.

[0040] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the fault type determination method for cigarette factory drying machine equipment as described above.

[0041] The method and apparatus for determining the fault type of a cigarette factory drying machine provided by this invention have the following beneficial effects:

[0042] 1. The second dataset is clustered using fuzzy C-means clustering to obtain clustering results. A hash table is then used to map the clustering results to construct an offline fault index library. The query vector is then searched in the offline fault index library using a high-dimensional nearest neighbor method to obtain the most similar data. The fault type of the first dataset is then determined based on the most similar data, which improves the efficiency of fault type determination.

[0043] 2. By using fuzzy C-means clustering to cluster the second dataset, the inherent structure of the second dataset is analyzed through clustering to obtain clusters. This allows the obtained clusters to be efficiently placed into the corresponding hash buckets, thereby improving the efficiency of hash bucket construction. Compared with the traditional locality-sensitive hashing algorithm based on random projection, hash bucket construction is more efficient. Moreover, improving the efficiency of hash bucket construction further improves the efficiency of building the offline fault index library, thus ultimately improving the efficiency of fault type determination.

[0044] 3. By comparing the constructed query vector with the cluster center set in the cluster center-hash bucket mapping hash table using Euclidean distance, target candidate clusters are obtained. Then, by calculating the Euclidean distance of each fault sample point in the union of the query vector and the target candidate clusters, target data samples are obtained. Finally, the fault type of the first dataset is determined by the fault type corresponding to the target data sample. Compared with traditional fault detection methods based on wavelet analysis, SOD outlier detection, and fault tree, this method is less time-consuming and can more quickly and timely determine the fault type of the equipment, thereby improving the stability of system operation.

[0045] 4. By constructing a query vector from the first dataset and matching the query vector in the fault index database to determine the fault type of the first dataset, multi-parameter joint determination is achieved, which improves the accuracy of fault type determination in complex system fault type determination.

[0046] 5. By efficiently and accurately determining the fault type of the first dataset, the efficiency of equipment maintenance is improved to a certain extent, making equipment maintenance more intelligent and timely, and reducing equipment maintenance costs and operational risks. Attached Figure Description

[0047] Figure 1 This is one of the flowcharts for the method of determining the fault type of a cigarette factory drying machine provided by the present invention;

[0048] Figure 2 The second flowchart of the method for determining the fault type of cigarette drying machine provided by the present invention;

[0049] Figure 3 A schematic diagram of the fault type determination device for cigarette drying machine equipment provided by the present invention;

[0050] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention are described clearly and completely below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0052] This invention employs a fuzzy clustering-based locally sensitive hashing algorithm. When a device malfunctions, it quickly retrieves high-dimensional fault data to promptly determine the type of fault. Simultaneously, this invention automatically learns the inherent structural distribution characteristics of the dataset using a fuzzy C-means algorithm, enabling more efficient and accurate hash binning and more precise retrieval results.

[0053] Figure 1 This is one of the flowcharts for the method of determining the fault type of a cigarette factory drying machine provided by the present invention, such as... Figure 1 Therefore, the method includes the following steps:

[0054] Step 1: Obtain the first dataset for determining the type of fault; the first dataset consists of various sensor data detected by the sensors during the operation of the wire drying machine.

[0055] Specifically, the first dataset consists of data features collected by sensors during the operation of the wire drying machine. These data features include, but are not limited to, motor temperature, motor speed, pipe humidity, pipe temperature, valve opening, and pipe pressure. Therefore, the first dataset can be represented as a set of equipment operation data values: [motor temperature, motor speed, valve opening, valve pressure, pipe outlet temperature, ..., moisture content at the wire drying outlet].

[0056] Step 2: Obtain the second dataset [x1, x2, x3, x4, ..., x] from all historical sensor data features affecting the operation of the wire drying machine. i .,x N The second dataset is the fault dataset corresponding to when the wire drying machine malfunctions, x i Let N be the i-th fault sample point, and N be the total number of fault sample points.

[0057] Specifically, all historical sensor data features can be obtained from the underlying CNC management platform. For example, the dataset constructed from these historical sensor data features can be represented as [45.1, 1286, 0.7, 3.42, 35, ..., 0.78], which contains a total of 35 sensor data points. Furthermore, the second dataset has the same data features as the first dataset, and the historical sensor data features also have the same data features as the first dataset. Historical operational fault data is selected from the historical sensor data features, and the selected historical operational fault data is labeled with corresponding fault type tags, thus forming the second dataset. The fault type tags refer to various types of faults that may occur during the production operation of the tobacco drying machine, including but not limited to excessively high motor temperature, excessive valve pressure, excessively low tobacco outlet humidity, and excessively low motor speed.

[0058] Step 3: Using the fuzzy C-means clustering method, perform L clustering on the second dataset to obtain L clustering results; each clustering result includes K clusters, and each cluster corresponds to a cluster center.

[0059] Specifically, this invention uses the fuzzy C-means clustering algorithm to cluster the second dataset in order to analyze its inherent structural distribution characteristics. Due to the perturbation of the membership factor, different clustering results can be generated.

[0060] The following is a detailed explanation of how to use fuzzy C-means clustering to perform L-fold clustering on the second dataset. This process includes the following steps:

[0061] Step 31: Randomly initialize the membership matrix u based on the number of clusters K and the number of data points N in the second dataset. ij This yields the initialized membership matrix; where i represents the i-th row of the membership matrix and j represents the j-th column of the membership matrix.

[0062] Specifically, the number of clusters K needs to be manually specified, usually based on the number of different types of malfunctions in the wire drying machine during production. K is generally chosen to be 20, and the membership threshold ε is set to 0.0000001. The membership matrix u is then randomly initialized. ij Membership matrices can be generated using random functions, for example, the following:

[0063]

[0064] Membership matrix u ijLet N be an N*K matrix, where N is the number of faulty sample points and K is the number of clusters. The membership matrix represents the degree to which each sample point belongs to each cluster. For the dataset [x1, x2, x3, x4, ..., x...] i .,x N For example, its i-th fault sample point x i The sum of the membership degrees of each cluster is 1, that is, the sum of the values ​​in the corresponding row of the membership degree matrix U is 1.

[0065] Step 32: Based on the membership factor m and the initialized membership matrix, perform preliminary clustering on the second dataset to obtain the initial cluster centers c. j .

[0066] Specifically, the membership factor m is generally in the range [1,3], and here m is taken as 2. The initial cluster centers c are obtained by calculating the membership matrix of the second dataset based on the initialized membership matrix. j Cluster center c j The following formula is used to calculate:

[0067]

[0068] Where c j Let the cluster center set be C = {c1, c2, ..., c...} k The j-th cluster center element in}.

[0069] Step 33: Based on the initial cluster centers c j Update the membership matrix;

[0070] Specifically, using the initial cluster centers c calculated in the previous iteration... j Recalculate the membership matrix u ij And update the membership matrix. Recalculate u. ij Use the following formula:

[0071]

[0072] Step 34: Repeat step 33 until the absolute difference between each element in each row of the current membership matrix and the corresponding element in the corresponding row of the previous membership matrix is ​​less than the preset membership threshold ε, thus obtaining the final membership matrix.

[0073] Specifically, calculate the membership matrix u for this operation. ij (t+1) And the membership matrix u from the previous iteration ij (t) Find the maximum change in the corresponding membership degree and iterate through it.

[0074]

[0075] Where t is the iteration number, and the membership matrix u is calculated in this iteration. ij (t+1) Each element in each row has a membership matrix u from the previous iteration ij (t) If the absolute difference between corresponding elements in the corresponding row is less than the preset membership threshold ε, the algorithm stops; otherwise, step 33 is repeated iteratively until the difference is less than the preset membership threshold ε. The final iterative membership matrix... It can be represented as:

[0076]

[0077] Step 35: Use the final membership matrix as the first clustering result of the second dataset. The first clustering result includes K clusters, and each cluster includes several fault sample points.

[0078] Specifically, each row of the membership matrix represents the degree to which the sample point belongs to each cluster. By comparing the membership values ​​in each row, the sample point is assigned to the cluster corresponding to the highest membership value, thus completing the clustering of the sample point. The final iterative membership matrix... To perform fuzzification, the maximum value in each row of the matrix is ​​set to 1, and the rest are set to 0. This results in the fuzzified membership matrix. like:

[0079] In this matrix, the i-th row represents the fault sample point x. i It belongs to the cluster with a value of 1. The membership matrix of the final iteration... This is the clustering result of the second dataset.

[0080] Step 36: Fix the number of cluster categories and the membership threshold. Use δ as the perturbation amplitude factor to randomly perturb the membership factor L-1 times. Repeat steps 31-35 with the membership factor after each perturbation to obtain the corresponding L-1 clustering results.

[0081] Specifically, through steps 35 and 36, a total of L clustering was performed, resulting in L clustering results. Each clustering result includes K clusters, and each cluster includes several fault sample points.

[0082] Step 4: Based on the L clustering results, construct an offline fault index library using hash tables; the offline fault index library includes: L hash mapping tables, each hash mapping table corresponding to one of the clustering results.

[0083] Specifically, the K clusters corresponding to a clustering result are placed into the corresponding K hash buckets, and a hash table is used to store the cluster center of each cluster and map it to the corresponding hash bucket, thus constructing a cluster center-hash bucket mapping hash table. By storing and mapping L clustering results separately using hash tables, L cluster center-hash bucket mapping hash tables can be obtained. These L cluster center-hash bucket mapping hash tables constitute the offline fault index library.

[0084] The following explains how to construct an offline fault index library using a hash table based on the L clustering results. This process includes the following steps:

[0085] Step 41: Select one clustering result from the L clustering results;

[0086] Step 42: Place the K clusters corresponding to the selected clustering results into K hash buckets respectively;

[0087] Step 43: Use a hash table to store the cluster center set C1 = {c1, c2, ..., c3} corresponding to the K clusters. k The cluster centers are mapped to the K hash buckets to construct a cluster center-hash bucket mapping hash table;

[0088] Step 44: Repeat steps 41-43 to map the remaining clustering results one by one using hash tables to obtain L-1 cluster center-hash bucket mapping hash tables;

[0089] Step 45: Use the L cluster center-hash bucket mapping hash tables obtained from Step 43 and Step 44 as the offline fault index library.

[0090] Specifically, the K clusters after the first clustering are placed into K hash buckets, and a hash table t1 is used to store the cluster center set C1 = {c1, c2, ..., c...}. k The cluster centers are mapped to K hash buckets, i.e., a hash table t1 is constructed to map the cluster centers to the hash buckets. The K clusters after the second clustering are placed into the K hash buckets, and a hash table t2 is used to store the cluster center set C2 = {c1, c2, ..., c...}. k The cluster centers are mapped to K hash buckets, thus constructing a cluster center-hash bucket mapping hash table t2... By continuously repeating this mapping, L cluster center-hash bucket mapping hash tables can be obtained, that is, the cluster center-hash bucket mapping hash table set T = {t1, t2, ..., t...}. LThe second dataset is clustered offline using a fuzzy C-means clustering algorithm, and its corresponding K×L hash buckets are stored to construct an offline fault index library. This offline fault index library is used to store the results of offline clustering of the second dataset based on the fuzzy C-means clustering algorithm, including hash mapping tables and hash bucket data, providing a data source for fault type determination.

[0091] Step 5: Determine the fault type of the first dataset based on the offline fault index library.

[0092] Specifically, the data most similar to the first dataset is retrieved from the offline fault index library, and the fault type label of that data is used as the query result output.

[0093] The following is a detailed explanation of how to determine the fault type of the first dataset based on the offline fault index library. This process includes the following steps:

[0094] Step 51: Select one of the L cluster center-hash bucket mapping hash tables;

[0095] Step 52: Construct a query vector based on the first dataset, and map the query vector to the cluster center set C1 = {c1, c2, ..., c3} in the selected cluster center-hash bucket mapping hash table. k Perform Euclidean distance calculations on each cluster, select the cluster center with the closest Euclidean distance as the target cluster center, and use the clusters in the hash buckets mapped by the target cluster center as candidate clusters for this query processing.

[0096] Specifically, real-time sensor data from the wire drying machine is collected, and a query vector is constructed. This involves collecting data from various sensors of the wire drying machine at regular time intervals during its operation, and then building a query vector based on a pre-designed data structure. This query vector is then constructed using the collected real-time operating data of the wire drying machine. q The query vector needs to be compared with the historical operating data sample x of the wire drying machine in the fault index database. i The attributes and dimensions remain consistent.

[0097] Construct a query vector such as: x q = [45.6, 1215, 0.7, 3.45, 35, ..., 0.78], the vector contains a total of 35 sensor data values. The meaning of each value in the vector corresponds to the historical sensor data feature list [motor temperature, motor speed, valve opening, valve air pressure, pipe outlet temperature, ..., drying wire outlet moisture] collected by the device sensors in step two. If there is a missing value in the sensor, -1 is used to replace it.

[0098] query vector x qThe hash table set T = {t1, t2, ..., t3} is mapped to the cluster center-hash bucket in the fault index. L The first hash table element t1 in the} stores the cluster center set C1 = {c1, c2, ..., c}. k Perform Euclidean distance calculation, traverse all elements in the cluster center set, and select the data in the hash bucket mapped by the cluster center with the closest Euclidean distance as the candidate cluster r1 for this query processing.

[0099] Step 53: Repeat step 52, and sequentially traverse the remaining hash tables in the L cluster center-hash bucket mapping hash tables to obtain L-1 candidate clusters;

[0100] Step 54: Perform a union of the L candidate clusters obtained in Steps 52 and 53 to obtain the target candidate set.

[0101] Specifically, the cluster center-hash bucket mapping hash table set T = {t1, t2, ..., t} is traversed sequentially. L All hash table elements in} are processed, and all candidate clusters are combined to form a candidate set R, where R = {r1∪r2∪...r}. L}

[0102] Step 54: Calculate the Euclidean distance between the query vector constructed from the first dataset and each fault sample point in the target candidate set, and select the data sample with the closest Euclidean distance as the target data sample.

[0103] Step 55: Determine the fault type of the first dataset based on the fault type corresponding to the target data sample.

[0104] Specifically, the query vector x q Perform Euclidean distance calculations on each data in the candidate set R obtained in step 53, retrieve the closest data, and output the retrieval result based on the fault type of that data.

[0105] Figure 2 This is the second flowchart of the method for determining the fault type of a cigarette factory drying machine provided by the present invention, as follows: Figure 2 As shown, the method includes the following steps:

[0106] Step S1: Acquire and analyze the sensor data characteristics affecting the operation of the wire drying machine, construct a historical operation characteristic sample of the wire drying machine, and obtain a sample set [x1,x2,x3,x4,...,x] p ], where x1,x2,x3,x4,...,x pThe sensor data features are the historical operation data of the wire drying machine; the sensor data features include, but are not limited to, motor temperature, motor speed, pipe humidity, pipe temperature, valve opening, pipe pressure, etc.

[0107] Step S2: For the sample set [x1,x2,x3,x4,...,x...] p Preprocessing was performed to filter out the fault sample dataset [x1, x2, x3, x4, ..., x] when the wire drying machine malfunctioned. i .,x N ], and label the sample data points with the corresponding fault type; where P≥N.

[0108] The fault type label refers to various types of faults that may occur in the tobacco drying machine during production and operation, including but not limited to motor overheating, valve pressure overpressure, tobacco outlet humidity under low humidity, and motor speed under low speed.

[0109] Step S3: Select the fault sample set [x1, x2, x3, x4, ..., x] selected in step S2. i .,x N As the training set, clustering is performed using the fuzzy C-means algorithm;

[0110] Step S4: Place each cluster processed in step S3 into the corresponding hash bucket, and use a hash table to store the cluster center of each cluster and map it to the corresponding hash bucket, thus constructing a cluster center-hash bucket mapping hash table.

[0111] Step S5: By perturbing the initial membership factor and repeating step S3 multiple times, generate multiple different fuzzy C-means clustering results, and perform step S4 operation on the clustering results generated in step S3 each time to construct a cluster center-hash bucket mapping hash table set and corresponding hash buckets, and establish an offline fault index library.

[0112] Step S6: Collect real-time operating data of the wire drying machine and construct a query vector based on the real-time operating data;

[0113] The query vector is constructed by collecting real-time operating data of the wire drying machine. This query vector must be consistent with the attributes and dimensions of the historical operating feature vector of the wire drying machine in the fault index database.

[0114] Step S7: Calculate the Euclidean distance between the query vector and the cluster centers stored in the cluster center-hash bucket mapping hash table in the fault index library, and select the hash bucket mapped to the nearest cluster center in the hash table as the candidate cluster;

[0115] Step S8: Repeat step S7 until all hash tables in the cluster center-hash bucket mapping hash table set have been traversed, and all candidate clusters are combined to form a candidate set.

[0116] Step S9: Calculate the similarity between the query vector and the data in the candidate set in step S8 one by one, retrieve the most similar data, and output the retrieval result based on the fault type of the data.

[0117] The following describes the fault type determination device for cigarette drying machine equipment provided by the present invention. The fault type determination device for cigarette drying machine equipment described below can be referred to in correspondence with the fault type determination method for cigarette drying machine equipment described above.

[0118] Figure 3 This is a schematic diagram of the fault type determination device for cigarette drying machines provided by the present invention, as shown in the figure. Figure 3 As shown, the device includes:

[0119] The acquisition unit 301 is used to acquire a first dataset of fault types to be determined; the first dataset consists of various sensor data detected by sensors during the operation of the wire drying machine.

[0120] The acquisition unit 301 is further configured to acquire a second dataset [x1, x2, x3, x4, ..., x] from all historical sensor data features affecting the operation of the wire drying machine. i .,x N The second dataset is the fault dataset corresponding to when the wire drying machine malfunctions;

[0121] Clustering unit 302 is used to perform L clustering operations on the second dataset using a fuzzy C-means clustering method, resulting in L clustering results; each clustering result includes K clusters, and each cluster corresponds to a cluster center;

[0122] Construction unit 303 is used to construct an offline fault index library using hash tables based on the L clustering results; the offline fault index library includes: L hash mapping tables, each hash mapping table corresponding to one of the clustering results;

[0123] The determining unit 304 is used to determine the fault type of the first dataset based on the offline fault index library.

[0124] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a fault type determination method for a cigarette factory drying machine, the method including:

[0125] Step 1: Obtain the first dataset for determining the type of fault; the first dataset consists of various sensor data detected by sensors during the operation of the wire drying machine.

[0126] Step 2: Obtain the second dataset [x1, x2, x3, x4, ..., x] from all historical sensor data features affecting the operation of the wire drying machine. i .,x N The second dataset is the fault dataset corresponding to when the wire drying machine malfunctions, x i This is the i-th fault sample point;

[0127] Step 3: Using the fuzzy C-means clustering method, perform L clustering operations on the second dataset to obtain L clustering results; each clustering result includes K clusters, and each cluster corresponds to a cluster center;

[0128] Step 4: Based on the L clustering results, construct an offline fault index library using hash tables; the offline fault index library includes: L hash mapping tables, each hash mapping table corresponding to one of the clustering results;

[0129] Step 5: Determine the fault type of the first dataset based on the offline fault index library.

[0130] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0131] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the fault type determination method for cigarette drying machine equipment provided by the above methods.

[0132] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the fault type determination method for cigarette factory drying machine provided by the above methods.

[0133] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0134] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0135] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining the fault type of a cigarette factory drying machine, characterized in that, include: Step 1: Obtain the first dataset for determining the fault type; The first dataset consists of various sensor data detected by sensors during the operation of the wire drying machine; the sensor data includes: motor temperature, motor speed, pipe humidity, pipe temperature, valve opening, and pipe pressure; Step 2: Obtain the second dataset [x1, x2, x3, x4, ..., x] from all historical sensor data features affecting the operation of the wire drying machine. i .,x N The second dataset is the fault dataset corresponding to when the wire drying machine malfunctions, x i This is the i-th fault sample point; Step 3: Using the fuzzy C-means clustering method, perform L clustering operations on the second dataset to obtain L clustering results; each clustering result includes K clusters, and each cluster corresponds to a cluster center; Step 4: Based on the L clustering results, construct an offline fault index library using hash tables; the offline fault index library includes: L hash mapping tables, each hash mapping table corresponding to one of the clustering results; Step 5: Determine the fault type of the first dataset based on the offline fault index library.

2. The method for determining the fault type of a cigarette factory drying machine according to claim 1, characterized in that, Step three includes: Step 31: Randomly initialize the membership matrix u based on the number of clusters K and the number of data points N in the second dataset. ij This yields the initialized membership matrix; where i represents the i-th row of the membership matrix and j represents the j-th column of the membership matrix. Step 32: Based on the membership factor m and the initialized membership matrix, perform preliminary clustering on the second dataset to obtain the initial cluster centers c. j ; Step 33: Based on the initial cluster centers c j Update the membership matrix; Step 34: Repeat step 33 until the absolute difference between each element in each row of the current membership matrix and the corresponding element in the corresponding row of the previous membership matrix is ​​less than the preset membership threshold ε, thus obtaining the final membership matrix. Step 35: Use the final membership matrix as the first clustering result of the second dataset. The first clustering result includes K clusters, and each cluster includes several fault sample points. Step 36: Fix the number of cluster categories and the membership threshold. Use δ as the perturbation amplitude factor to randomly perturb the membership factor L-1 times. Repeat steps 31-35 with the membership factor after each perturbation to obtain the corresponding L-1 clustering results.

3. The method for determining the fault type of a cigarette factory drying machine according to claim 2, characterized in that, Step four includes: Step 41: Select one clustering result from the L clustering results; Step 42: Place the K clusters corresponding to the selected clustering results into K hash buckets respectively; Step 43: Use a hash table to store the cluster center set C1 = {c1, c2, ..., c3} corresponding to the K clusters. k The cluster centers are mapped to the K hash buckets to construct a cluster center-hash bucket mapping hash table; Step 44: Repeat steps 41-43 to map the remaining clustering results one by one using the hash table to obtain L-1 cluster center-hash bucket mapping hash tables; Step 45: Use the L cluster center-hash bucket mapping hash tables obtained from Step 43 and Step 44 as the offline fault index library.

4. The method for determining the fault type of a cigarette factory drying machine according to claim 3, characterized in that, Step five includes: Step 51: Select one of the L cluster center-hash bucket mapping hash tables; Step 52: Construct a query vector based on the first dataset, and map the query vector to the cluster center set C1 = {c1, c2, ..., c3} in the selected cluster center-hash bucket mapping hash table. k Perform Euclidean distance calculations on each cluster, select the cluster center with the closest Euclidean distance as the target cluster center, and use the clusters in the hash buckets mapped by the target cluster center as candidate clusters for this query processing. Step 53: Repeat step 52, and sequentially traverse the remaining hash tables in the L cluster center-hash bucket mapping hash tables to obtain L-1 candidate clusters; Step 54: Perform a union of the L candidate clusters obtained in Steps 52 and 53 to obtain the target candidate set; Step 54: Calculate the Euclidean distance between the query vector constructed from the first dataset and each fault sample point in the target candidate set, and select the data sample with the closest Euclidean distance as the target data sample. Step 55: Determine the fault type of the first dataset based on the fault type corresponding to the target data sample.

5. An apparatus for determining the fault type of a cigarette factory drying machine as described in claim 1, characterized in that, include: The acquisition unit is used to acquire the first dataset of the fault types to be determined. The first dataset consists of various sensor data detected by sensors during the operation of the wire drying machine. The acquisition unit is further configured to acquire a second dataset [x1,x2,x3,x4,...,x] from all historical sensor data features affecting the operation of the wire drying machine. N The second dataset is the fault dataset corresponding to when the wire drying machine malfunctions; The clustering unit is used to perform L clustering operations on the second dataset using a fuzzy C-means clustering method, resulting in L clustering results; each clustering result includes K clusters, and each cluster corresponds to a cluster center; A construction unit is used to construct an offline fault index library using a hash table based on the L clustering results; The offline fault index library includes: L hash mapping tables, each hash mapping table corresponding to one of the clustering results; The determining unit is used to determine the fault type of the first dataset based on the offline fault index library.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for determining the fault type of the cigarette factory drying machine as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for determining the fault type of the cigarette factory drying machine as described in any one of claims 1 to 4.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for determining the fault type of the cigarette factory drying machine as described in any one of claims 1 to 4.

Images