Fault detection model evaluation index recommendation method and device based on clustering algorithm

Abstract

The embodiment of the invention provides a fault detection model evaluation index recommendation method and device based on a clustering algorithm. The fault detection model evaluation index recommendation method based on the clustering algorithm comprises the steps that equipment operation information corresponding to each piece of historical equipment in a plurality of pieces of historical equipment and a plurality of historical evaluation indexes are acquired; encoding the equipment operation information and the plurality of historical evaluation indexes to generate an encoding vector corresponding to each historical equipment; clustering the plurality of historical devices through a target clustering algorithm and based on the coding vector, and generating a corresponding clustering result; and determining a to-be-recommended evaluation index corresponding to each historical device according to the clustering result.

Description

Technical Field

[0001] The embodiments of the present specification relate to the field of computer technology, and in particular to a method for recommending evaluation indicators of a fault detection model based on a clustering algorithm. Background Art

[0002] Fault detection refers to the process in which the system can determine, locate and analyze the location of abnormal conditions under a certain working environment, find out the specific reasons that cause the system to be abnormal, determine the type and level of the fault, and even perform automatic repairs or give corresponding repair suggestions. Bearings, gears, etc., as key components for transmitting force and torque in mechanical equipment, are prone to local failures. Any minor failure may bring huge risks to the reliability of the entire mechanical system. Therefore, fault detection is crucial for the safe operation of mechanical equipment.

[0003] At present, the method of fault detection for equipment is usually to first determine some detection indicators related to the equipment, and then determine whether the equipment has a fault by determining the indicator value of the equipment under each detection indicator. Therefore, how to ensure the accuracy of the selected detection indicators has become an urgent problem to be solved. Summary of the invention

[0004] In view of this, the embodiments of this specification provide a method for recommending evaluation indicators of a fault detection model based on a clustering algorithm. One or more embodiments of this specification also involve a device for recommending evaluation indicators of a fault detection model based on a clustering algorithm, a computing device, a computer-readable storage medium, and a computer program to solve the technical defects existing in the prior art.

[0005] According to a first aspect of an embodiment of this specification, a method for recommending evaluation indicators of a fault detection model based on a clustering algorithm is provided, comprising:

[0006] Obtain device operation information and multiple historical evaluation indicators corresponding to each historical device among multiple historical devices;

[0007] Encoding the device operation information and the multiple historical evaluation indicators to generate a coding vector corresponding to each historical device;

[0008] Clustering the plurality of historical devices based on the encoding vectors by using a target clustering algorithm to generate corresponding clustering results;

[0009] Determine the evaluation index to be recommended corresponding to each historical device according to the clustering result.

[0010] Optionally, the target clustering algorithm includes a DBSCAN clustering algorithm;

[0011] Accordingly, clustering the plurality of historical devices by using a target clustering algorithm and based on the encoding vectors includes:

[0012] Determine the cluster radius corresponding to each device category and the number of historical devices contained in each device category;

[0013] The plurality of historical devices are clustered by using the DBSCAN clustering algorithm and based on the clustering radius, the number and the encoding vector.

[0014] Optionally, determining the evaluation index to be recommended corresponding to each historical device according to the clustering result includes:

[0015] Determine a first device category to which the first historical device belongs according to the clustering result, and determine second historical devices included in the first device category, wherein the first historical device is each of the plurality of historical devices, and the second historical device is each other historical device included in the first device category except the first historical device;

[0016] According to a plurality of historical evaluation indicators respectively corresponding to the first historical device and the second historical device, a to-be-recommended evaluation indicator corresponding to the first historical device is determined.

[0017] Optionally, determining the evaluation indicator to be recommended corresponding to the first historical device according to the plurality of historical evaluation indicators respectively corresponding to the first historical device and the second historical device includes:

[0018] Determine the overlapping historical evaluation indicators among the plurality of historical evaluation indicators respectively corresponding to the first historical device and the second historical device;

[0019] According to the multiple historical evaluation indicators corresponding to the first historical device and the overlapping historical evaluation indicators, the evaluation indicator to be recommended corresponding to the first historical device is determined.

[0020] Optionally, the clustering algorithm-based fault detection model evaluation index recommendation method further includes:

[0021] Acquire device operation information corresponding to the target device, and encode the device operation information corresponding to the target device to generate a target encoding vector corresponding to the target device;

[0022] Determining a second device category to which the target device belongs according to the target coding vector and the clustering result;

[0023] Determining the similarity between the target device and each historical device in the second device category according to the target coding vector and the coding vector corresponding to each historical device in the second device category;

[0024] Determine the evaluation index to be recommended corresponding to the target device according to the similarity.

[0025] Optionally, determining the evaluation index to be recommended corresponding to the target device according to the similarity includes:

[0026] determining a maximum value among similarities between the target device and each historical device included in the second device category;

[0027] The target historical device corresponding to the maximum value is determined, and the evaluation index to be recommended corresponding to the target historical device is determined as the evaluation index to be recommended corresponding to the target device.

[0028] Optionally, the equipment operation information includes usage condition information, equipment failure information, data processing information, etc.

[0029] According to a second aspect of an embodiment of this specification, a fault detection model evaluation index recommendation device based on a clustering algorithm is provided, comprising:

[0030] An acquisition module is configured to acquire device operation information and a plurality of historical evaluation indicators corresponding to each of a plurality of historical devices;

[0031] An encoding module, configured to encode the device operation information and the multiple historical evaluation indicators to generate an encoding vector corresponding to each historical device;

[0032] A clustering module, configured to cluster the plurality of historical devices based on the encoding vectors by using a target clustering algorithm, and generate corresponding clustering results;

[0033] The determination module is configured to determine the evaluation index to be recommended corresponding to each historical device according to the clustering result.

[0034] According to a third aspect of an embodiment of this specification, a computing device is provided, including:

[0035] Memory and processor;

[0036] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement any step of the method for recommending evaluation indicators of a fault detection model based on a clustering algorithm.

[0037] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of any one of the methods for recommending evaluation indicators of a fault detection model based on a clustering algorithm are implemented.

[0038] According to a fifth aspect of the embodiments of this specification, a computer program is provided, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the above-mentioned method for recommending evaluation indicators of fault detection models based on clustering algorithms.

[0039] The embodiments of this specification obtain the device operation information and multiple historical evaluation indicators corresponding to each historical device in multiple historical devices, encode the device operation information and multiple historical evaluation indicators, and generate a coding vector corresponding to each historical device; cluster the multiple historical devices based on the coding vector through the target clustering algorithm to generate corresponding clustering results; determine the evaluation indicators to be recommended corresponding to each historical device according to the clustering results. By analyzing the specific characteristics and working environment of the equipment, the matching of the recommendation results with the actual situation of the equipment is achieved, and corresponding evaluation indicators are given to different types of equipment under different working conditions, environments, etc., which is conducive to improving the accuracy of the recommendation results and verification results. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a flowchart of a method for recommending evaluation indicators of a fault detection model based on a clustering algorithm provided by an embodiment of this specification;

[0041] Figure 2 It is a process flow chart of a fault detection model evaluation index recommendation method based on a clustering algorithm provided by an embodiment of this specification;

[0042] Figure 3 It is a structural diagram of a fault detection model evaluation index recommendation device based on a clustering algorithm provided by an embodiment of this specification;

[0043] Figure 4 It is a structural block diagram of a computing device provided by an embodiment of this specification. DETAILED DESCRIPTION

[0044] Many specific details are described in the following description to facilitate a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of this specification, so this specification is not limited to the specific implementation disclosed below.

[0045] The terms used in one or more embodiments of this specification are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of this specification. The singular forms of "a", "said" and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0046] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0047] In this specification, a method for recommending evaluation indicators of a fault detection model based on a clustering algorithm is provided. This specification also involves a device for recommending evaluation indicators of a fault detection model based on a clustering algorithm, a computing device, a computer-readable storage medium, and a computer program, which are described in detail one by one in the following embodiments.

[0048] Figure 1 A flowchart of a method for recommending evaluation indicators of a fault detection model based on a clustering algorithm according to an embodiment of the present specification is shown, which specifically includes the following steps.

[0049] Step 102: Obtain device operation information and multiple historical evaluation indicators corresponding to each historical device in multiple historical devices.

[0050] Specifically, the historical equipment may be a mechanical equipment.

[0051] The equipment operation information of historical equipment may include usage condition information, equipment failure information, and data processing information.

[0052] The usage condition information may include engineering fields, equipment objects, usage conditions, and other usage background information.

[0053] Among them, the attribute parameters contained in the engineering field are usually the field type, and the field type refers to the industry field type to which the verification object belongs during the process of determining the test verification sample of the current verification system, such as aircraft, hydraulic machinery, wind turbines, ships, etc.

[0054] The attribute parameters contained in the equipment object are usually based on the object type, and the object type refers to the common fault components in the system object, such as the gears and bearings in the gearbox.

[0055] Operating conditions may include common environmental stresses and working stresses, etc. Environmental stresses may include temperature stress, humidity stress, vibration stress (maximum vibration, weighted vibration, minimum vibration, cruising continuous vibration, etc.), and working stresses may include electrical stress, speed, flow, pressure, load and other information.

[0056] Other usage background information mainly includes some background information not involved in the usage environment and operating conditions, such as constant speed conditions (single operating condition) / variable speed (multiple operating conditions) conditions of rotating machinery, marked sample conditions, single fault type / multiple fault types and other qualitative descriptions.

[0057] The equipment fault information of historical equipment may include fault type information and fault severity information. The attribute parameters included in the fault type information are diagnosable fault types, and the diagnosable fault types cover some common fault types of various historical equipment. Taking the gearbox as an example, they include gear tooth breakage, wear, crack faults, inner ring, outer ring, and rolling element faults of bearings, etc. The attribute parameters included in the fault severity information are diagnosable fault severity, which is used to describe the diagnostic capability of the diagnostic scheme. It can diagnose whether the fault is an initial fault or a severe fault, so as to determine whether it meets the actual needs of the project.

[0058] The data processing information of historical devices may include data transmission delay information (bandwidth, delay, transmission rate), data transmission packet loss information (packet loss rate, packet loss time, number of retransmissions), data transmission sampling rate fluctuation information (actual transmission rate, fluctuation and its impact on application performance) as well as CPU utilization, memory usage, disk I / O rate, network bandwidth, etc.

[0059] Historical evaluation indicators may include fault detection rate, missed detection rate, false alarm rate, F1 weighted value, and mean square error.

[0060] Step 104: Encode the device operation information and the multiple historical evaluation indicators to generate a coding vector corresponding to each historical device.

[0061] Specifically, after obtaining the device operation information and multiple historical evaluation indicators corresponding to each historical device, the device operation information and multiple historical evaluation indicators corresponding to each historical device can be encoded to cluster the historical devices based on the generated encoding vector.

[0062] In practical applications, the device operation information and multiple historical evaluation indicators corresponding to each historical device can also be input as sample data into the evaluation indicator recommendation model to be trained, so as to train the evaluation indicator recommendation model to be trained with this part of sample data to obtain the trained evaluation indicator recommendation model. Therefore, the device operation information and multiple historical evaluation indicators are encoded, specifically, the device operation information and multiple historical evaluation indicators can be input as input sets into the evaluation indicator recommendation model to be trained, and the evaluation indicator recommendation model to be trained encodes the device operation information and multiple historical evaluation indicators to generate a coding vector corresponding to each historical device.

[0063] Step 106: Clustering the multiple historical devices based on the encoding vectors using a target clustering algorithm to generate corresponding clustering results.

[0064] In an optional implementation, the target clustering algorithm includes a DBSCAN clustering algorithm;

[0065] Accordingly, clustering the plurality of historical devices by using a target clustering algorithm and based on the encoding vectors includes:

[0066] Determine the cluster radius corresponding to each device category and the number of historical devices contained in each device category;

[0067] The plurality of historical devices are clustered by using the DBSCAN clustering algorithm and based on the clustering radius, the number and the encoding vector.

[0068] Specifically, as mentioned above, after obtaining the equipment operation information and multiple historical evaluation indicators corresponding to each historical equipment, the equipment operation information and multiple historical evaluation indicators corresponding to each historical equipment can be input as sample data into the evaluation indicator recommendation model to be trained to obtain the evaluation indicator recommendation model.

[0069] Since the database will store the evaluation data of various types of equipment as well as information such as the equipment usage environment and usage conditions, different equipment or different usage conditions will have different evaluation focuses. Therefore, the embodiments of this specification capture the equipment operation information of each historical device and the features in the historical evaluation indicators, and convert them into coding vectors through the evaluation indicator recommendation model to be trained. These coding vectors can effectively represent each historical device, that is, they can approximately uniquely identify the feature set of each historical device. The coding vectors are then used to train the evaluation indicator recommendation model to be trained, so that the evaluation indicator recommendation model can extract key information from the equipment data of the historical equipment and accurately obtain the similarity between each device.

[0070] In addition, when training the model, a clustering algorithm can be used to classify devices with high similarity into the same category to save subsequent computing costs. A clustering algorithm is an unsupervised learning method that can classify objects in a data set into several categories based on similarities and differences, so that objects in the same category have high similarity, while objects in different categories have low similarity. The embodiments of this specification can use a clustering algorithm to classify multiple historical devices into several categories, and the historical devices in each category have high similarity in evaluation indicators.

[0071] In the embodiment of this specification, the target clustering algorithm includes the DBSCAN clustering algorithm.

[0072] DBSCAN is an algorithm that clusters data points in space by defining density. The neighborhood proposed in the DBSCAN algorithm is used to reflect the density of the data point set in space. The DBSCAN algorithm defines Eps and Minpts, and (Eps, Minpts) is used to describe the distribution of neighborhood data points. Among them, Eps describes the threshold of the neighborhood distance of a certain data point, that is, the neighborhood radius (clustering radius) introduced by the definition of density, and Minpts describes the threshold of the number of points contained in the neighborhood of a certain data point with a radius of Eps.

[0073] In the DBSCAN algorithm, the relationship between points is defined as: direct density reachability, density reachability, and density connectivity. Direct density reachability is defined as: if point p is in the Eps neighborhood of core point q, then point p is directly density reachable from point q. The opposite is not necessarily true unless p is also a core point. Density reachability is defined as: if there exists a sample point sequence p 1 、p 2 ,……,p n , p i+1 It is from p i Based on Eps and Minpts, it is directly density-reachable. From the characteristics of direct density reachability, it can be seen that the density reachability relationship also does not satisfy the symmetry. Density connectivity can be defined as: if there is a core point O such that p i and p j can be reached by O density, then p i and p j They are density connected. The density connection relationship satisfies symmetry.

[0074] According to the density definition of DBSCAN, the largest set that satisfies the density connection relationship derived from the density reachable relationship is the class after classification. The core idea of ​​the DBSCAN clustering algorithm can be described as follows: starting from a specified core point, slowly diffuse to the density reachable range, and finally obtain a maximum area containing core points and boundary points, and any two points in the area are density connected.

[0075] The specific execution steps are as follows:

[0076] (1) Calculate the ε neighborhood of all points: For each point P in the data set, calculate the number of neighbors in its ε neighborhood. The threshold of the number of neighbors is usually defined by a parameter MinPts.

[0077] (2) Marking core points: If the number of points in the ε neighborhood of a point is greater than or equal to MinPts, then the point is marked as a core point.

[0078] (3) Find density-connected points: For each core point, find all points that are density-connected to it. If point P is in the ε neighborhood of point O, and O is a core point, then P is a point that is density-connected to O.

[0079] (4) Marking noise points and boundary points: Points that are not marked as core points are marked as noise points. Points that are densely connected to a core point but are not core points are marked as boundary points.

[0080] Each core point or point densely connected to it is assigned a separate cluster label. If a point is densely connected to multiple core points, it will be assigned the cluster label of the first core point found.

[0081] (5) Noise points form independent clusters: All noise points form an independent cluster.

[0082] The DBSCAN algorithm clusters data points in space based on density, so as to cluster data points with higher density in the data set and exclude data points with lower density. The algorithm can classify data sets of any shape, and does not require pre-setting of division conditions before classification, such as setting the number of classifications, and can handle noise points well.

[0083] Step 108: Determine the evaluation index to be recommended corresponding to each historical device according to the clustering result.

[0084] Specifically, after the historical devices are clustered, the evaluation index to be recommended corresponding to each historical device can be determined according to the clustering result.

[0085] As mentioned above, the historical evaluation index corresponding to each historical device may be a fault detection rate, a missed detection rate, a false alarm rate, an F1 weighted value, a mean square error, etc. Then the evaluation index to be recommended may be one or at least two of the historical evaluation indexes, and the historical evaluation index of each historical device may be the same as or different from the evaluation index to be recommended.

[0086] In an optional implementation, determining the evaluation index to be recommended corresponding to each historical device according to the clustering result includes:

[0087] Determine a first device category to which the first historical device belongs according to the clustering result, and determine second historical devices included in the first device category, wherein the first historical device is each of the plurality of historical devices, and the second historical device is each other historical device included in the first device category except the first historical device;

[0088] According to a plurality of historical evaluation indicators respectively corresponding to the first historical device and the second historical device, a to-be-recommended evaluation indicator corresponding to the first historical device is determined.

[0089] Furthermore, the determining, based on the plurality of historical evaluation indicators respectively corresponding to the first historical device and the second historical device, the evaluation indicator to be recommended corresponding to the first historical device includes:

[0090] Determine the overlapping historical evaluation indicators among the plurality of historical evaluation indicators respectively corresponding to the first historical device and the second historical device;

[0091] According to the multiple historical evaluation indicators corresponding to the first historical device and the overlapping historical evaluation indicators, the evaluation indicator to be recommended corresponding to the first historical device is determined.

[0092] Specifically, for each historical device (first historical device) among multiple historical devices, when determining its corresponding evaluation indicator to be recommended, the multiple historical evaluation indicators corresponding to the first historical device can be combined with the historical evaluation indicators corresponding to other historical devices (second historical devices) included in the first device category to which it belongs, so as to jointly determine them, so as to improve the accuracy and practicality of the determined evaluation indicators to be recommended.

[0093] The embodiment of the present specification determines the evaluation index to be recommended corresponding to the first historical device by determining the overlapping historical evaluation indexes among the multiple historical evaluation indexes respectively corresponding to the first historical device and the second historical device.

[0094] For example, the device category to which the first historical device belongs contains three historical devices, and the three historical devices are device S1, device S2, and device S3, and the historical evaluation indicators corresponding to device S1 are indicator 1, indicator 2, and indicator 3, the historical evaluation indicators corresponding to device S2 are indicator 1, indicator 2, and indicator 4, and the historical evaluation indicators corresponding to device S3 are indicator 2, indicator 3, and indicator 5. Then, the historical evaluation indicators corresponding to the three historical devices overlap, namely indicator 2 and indicator 3. Therefore, indicator 1, indicator 2, and indicator 3 can be determined as the evaluation indicators to be recommended for device S1; indicator 1, indicator 2, and indicator 3 can be determined as the evaluation indicators to be recommended for device S2, or indicator 1, indicator 2, and indicator 4 can be determined as the evaluation indicators to be recommended for device S2, and indicator 2, indicator 3, and indicator 4 can also be determined as the evaluation indicators to be recommended for device S2; similarly, indicator 2, indicator 3, and indicator 5 can be determined as the evaluation indicators to be recommended for device S3.

[0095] In an optional implementation, the clustering algorithm-based fault detection model evaluation index recommendation method further includes:

[0096] Acquire device operation information corresponding to the target device, and encode the device operation information corresponding to the target device to generate a target encoding vector corresponding to the target device;

[0097] Determining a second device category to which the target device belongs according to the target coding vector and the clustering result;

[0098] Determining the similarity between the target device and each historical device in the second device category according to the target coding vector and the coding vector corresponding to each historical device in the second device category;

[0099] Determine the evaluation index to be recommended corresponding to the target device according to the similarity.

[0100] Furthermore, determining the evaluation index to be recommended corresponding to the target device according to the similarity includes:

[0101] determining a maximum value among similarities between the target device and each historical device included in the second device category;

[0102] The target historical device corresponding to the maximum value is determined, and the evaluation index to be recommended corresponding to the target historical device is determined as the evaluation index to be recommended corresponding to the target device.

[0103] Specifically, the device operation information corresponding to the target device is of the same information type as the device operation information corresponding to each of the aforementioned historical devices, that is, the device operation information corresponding to the target device may also include engineering fields, device objects, usage conditions, and other usage background information.

[0104] After obtaining the device operation information of the target device, the device operation information of the target device can be encoded to generate a corresponding target encoding vector. Then, the target device can be classified into a device category based on the target encoding vector to classify the target device into a target device category (second device category) among the multiple device categories generated by clustering the historical devices.

[0105] In practical applications, based on the clustering of multiple historical devices by the DBSCAN clustering algorithm in the aforementioned embodiment, it is assumed that n device categories are generated after clustering, namely, device category 1, device category 2, ..., device category n, then there is a central point device (core point) in each device category. When it is necessary to classify the target device, the central point device of each device category can be determined first, and the encoding vector of the central point device can be determined, and then the distance between the encoding vector of the target device and the encoding vector of each central point device can be calculated, and then according to the distance calculation result, the target device can be classified into the device category where the central point device closest to it is located.

[0106] Based on this, after determining the device category to which the target device belongs, the similarity between the target device and each historical device in the device category can be determined based on the target coding vector and the coding vector corresponding to each historical device in the device category, and the recommended evaluation indicator corresponding to the target device can be determined based on the similarity.

[0107] In practical applications, the cosine similarity algorithm can be used to calculate the similarity between the encoding vector of the target device and the encoding vector corresponding to each historical device in the device category.

[0108] In addition, after determining the similarity between the target device and each historical device in the device category, the maximum similarity value can be selected from multiple similarities, and the target historical device corresponding to the maximum similarity value can be determined, and then the evaluation indicator to be recommended corresponding to the target historical device can be determined as the evaluation indicator to be recommended corresponding to the target device.

[0109] In practical applications, the recommended evaluation indicators can be used to verify whether the target device has a fault.

[0110] In addition, based on the training of the evaluation index recommendation model, since the evaluation index recommendation model can extract key information from the device data of historical devices and accurately obtain the similarity between each device, when there is a new target device, the evaluation index recommendation model can determine the similarity between the target device and each historical device based on the device operation information of the target device, and determine the evaluation index to be recommended for the target device based on the similarity.

[0111] Based on this, when a new target device needs to be introduced, the device operation information of the target device is first encoded using the trained evaluation index recommendation model, and the target device is quickly classified according to the encoding vector to determine the device category to which it belongs. This process helps to narrow the scope of subsequent similarity calculations, and only the similarity calculation needs to be performed within the device category to which the target device belongs. Within the device category, the cosine similarity calculation method is used between the encoding vectors of the target device and the historical device to evaluate the similarity between the target device and the historical device. This method not only reduces the amount of calculation, but also helps to improve the pertinence and accuracy of the similarity evaluation.

[0112] By performing information mining in the above manner, the output information that can be obtained may include a recommended evaluation indicator set suitable for selection, evaluation indicator results, reference ranges, and devices similar to the target device and verification cases, etc.

[0113] In practice, the data corresponding to different devices are different in terms of engineering fields, object types, operating conditions, and usage environments. The evaluation index recommendation model usually presents different processing capabilities for different devices during use. Moreover, even for the same device, the processing capabilities of the model show great differences under different conditions such as different operating conditions and failure modes. This difference is not only reflected in the accuracy of the model's recommendation results for the evaluation indicators, but also affects the adaptability and stability of the verification evaluation.

[0114] Moreover, in actual application scenarios, there are a few devices with a small number of devices and a small number of fault samples. If only the data of these small devices are used for model training, the trained model will be "overfitted" and unable to judge the actual situation of the device from an objective perspective. Therefore, it is particularly important to make full use of historical data and existing evaluation and verification information (to be recommended evaluation indicators). Historical data contains rich and effective information that can be mined. This information not only records the performance of past algorithms, but also contains common characteristics under different equipment, working conditions, and environmental conditions.

[0115] The embodiments of this specification use a similarity algorithm to calculate the similarity between the target device and the historical device, and select the more relevant evaluation indicators to be recommended. The evaluation indicator recommendation model is based on the effective information of the historical device and combines the relevant data of the target device for comprehensive evaluation, so as to recommend evaluation indicators for the target device through more comprehensive and objective data. When the evaluation indicator recommendation for the target device is completed, the target device's recommended evaluation indicator information and device operation information can be used as historical verification information to further update and improve the historical verification information, and serve as a reference for verification of new devices introduced in the future.

[0116] The embodiments of this specification obtain the device operation information and multiple historical evaluation indicators corresponding to each historical device in multiple historical devices, encode the device operation information and multiple historical evaluation indicators, and generate a coding vector corresponding to each historical device; cluster the multiple historical devices based on the coding vector through the target clustering algorithm to generate corresponding clustering results; determine the evaluation indicators to be recommended corresponding to each historical device according to the clustering results. By analyzing the specific characteristics and working environment of the equipment, the matching of the recommendation results with the actual situation of the equipment is achieved, and corresponding evaluation indicators are given to different types of equipment under different working conditions, environments, etc., which is conducive to improving the accuracy of the recommendation results and verification results.

[0117] The following combination Figure 2 , the fault detection model evaluation index recommendation method based on clustering algorithm is further described. Figure 2 A processing flow chart of a fault detection model evaluation index recommendation method based on a clustering algorithm provided in an embodiment of the present specification is shown, which specifically includes the following steps.

[0118] Step 202: Obtain device operation information corresponding to the target device.

[0119] Specifically, equipment operation information includes operating condition information, equipment failure information, and data processing information.

[0120] Step 204: Encode the device operation information corresponding to the target device to generate a target coding vector corresponding to the target device.

[0121] Step 206: Determine clustering results corresponding to multiple historical devices, and determine a device code vector corresponding to the center point device in each clustering result.

[0122] Step 208: Calculate a first similarity between the target device and each of the center point devices according to the target coding vector and the device coding vector.

[0123] Step 210: Determine whether the first similarity is greater than or equal to a preset threshold.

[0124] If yes, execute step 212 ; if no, execute step 218 .

[0125] Specifically, when there are multiple center point devices, multiple first similarities can be calculated and obtained. Therefore, it can be determined whether there is at least one first similarity value greater than or equal to a preset threshold among the multiple first similarities.

[0126] Step 212: Determine the target device category to which the target device belongs according to the first similarity, and calculate a second similarity between the target device and historical devices included in the target device category based on the target encoding vector.

[0127] Step 214: Determine the evaluation index to be recommended corresponding to the target device according to the second similarity.

[0128] Step 216: Add the evaluation indicator to be recommended to the historical verification information set.

[0129] Step 218: Determine the maximum value among the first similarity values, and determine the target historical device corresponding to the maximum value.

[0130] Specifically, when each first similarity is smaller than a preset threshold, similarity correction is performed on the target device.

[0131] Step 220: Determine and output the evaluation index to be recommended corresponding to the target device according to the historical evaluation index corresponding to the target historical device.

[0132] After step 220 is completed, step 216 may be continued.

[0133] The embodiments of this specification achieve matching of recommendation results with actual conditions of the equipment by analyzing specific characteristics and working environment of the equipment, and provide corresponding evaluation indicators for different types of equipment under different working conditions, environments and other conditions; in addition, when the evaluation indicator recommendation for the target equipment is completed, the target equipment's recommended evaluation indicator information and equipment operation information can be used as historical verification information to further update and improve the historical verification information, and serve as a reference for verification of new equipment introduced in the future, which is conducive to improving the accuracy of recommendation results and verification results.

[0134] Corresponding to the above method embodiment, this specification also provides an embodiment of a fault detection model evaluation index recommendation device based on a clustering algorithm, Figure 3 FIG. 1 is a schematic diagram showing a structure of a fault detection model evaluation index recommendation device based on a clustering algorithm provided by an embodiment of the present specification. Figure 3 As shown, the device comprises:

[0135] An acquisition module 302 is configured to acquire device operation information and a plurality of historical evaluation indicators corresponding to each of a plurality of historical devices;

[0136] The encoding module 304 is configured to encode the device operation information and the multiple historical evaluation indicators to generate an encoding vector corresponding to each historical device;

[0137] A clustering module 306 is configured to cluster the plurality of historical devices based on the encoding vectors by using a target clustering algorithm to generate corresponding clustering results;

[0138] The determination module 308 is configured to determine the evaluation index to be recommended corresponding to each historical device according to the clustering result.

[0139] Optionally, the target clustering algorithm includes a DBSCAN clustering algorithm;

[0140] Accordingly, the clustering module 306 is further configured to:

[0141] Determine the cluster radius corresponding to each device category and the number of historical devices contained in each device category;

[0142] The plurality of historical devices are clustered by using the DBSCAN clustering algorithm and based on the clustering radius, the number and the encoding vector.

[0143] Optionally, the determining module 308 is further configured to:

[0144] Determine a first device category to which the first historical device belongs according to the clustering result, and determine second historical devices included in the first device category, wherein the first historical device is each of the plurality of historical devices, and the second historical device is each other historical device included in the first device category except the first historical device;

[0145] According to a plurality of historical evaluation indicators respectively corresponding to the first historical device and the second historical device, a to-be-recommended evaluation indicator corresponding to the first historical device is determined.

[0146] Optionally, the determining module 308 is further configured to:

[0147] Determine the overlapping historical evaluation indicators among the plurality of historical evaluation indicators respectively corresponding to the first historical device and the second historical device;

[0148] According to the multiple historical evaluation indicators corresponding to the first historical device and the overlapping historical evaluation indicators, the evaluation indicator to be recommended corresponding to the first historical device is determined.

[0149] Optionally, the fault detection model evaluation index recommendation device based on clustering algorithm further includes a processing module 310 configured to:

[0150] Acquire device operation information corresponding to the target device, and encode the device operation information corresponding to the target device to generate a target encoding vector corresponding to the target device;

[0151] Determining a second device category to which the target device belongs according to the target coding vector and the clustering result;

[0152] Determining the similarity between the target device and each historical device in the second device category according to the target coding vector and the coding vector corresponding to each historical device in the second device category;

[0153] Determine the evaluation index to be recommended corresponding to the target device according to the similarity.

[0154] Optionally, the processing module 310 is further configured to:

[0155] determining a maximum value among similarities between the target device and each historical device included in the second device category;

[0156] The target historical device corresponding to the maximum value is determined, and the evaluation index to be recommended corresponding to the target historical device is determined as the evaluation index to be recommended corresponding to the target device.

[0157] Optionally, the equipment operation information includes usage condition information, equipment failure information, data processing information, etc.

[0158] The above is a schematic scheme of a fault detection model evaluation index recommendation device based on a clustering algorithm of this embodiment. It should be noted that the technical scheme of the fault detection model evaluation index recommendation device based on a clustering algorithm and the technical scheme of the fault detection model evaluation index recommendation method based on a clustering algorithm belong to the same concept, and the details of the technical scheme of the fault detection model evaluation index recommendation device based on a clustering algorithm that are not described in detail can be found in the description of the technical scheme of the fault detection model evaluation index recommendation method based on a clustering algorithm.

[0159] Figure 4 The block diagram of a computing device 400 according to one embodiment of the present specification is shown. The components of the computing device 400 include but are not limited to a memory 410 and a processor 420. The processor 420 is connected to the memory 410 via a bus 430, and the database 450 is used to store data.

[0160] The computing device 400 also includes an access device 440 that enables the computing device 400 to communicate via one or more networks 460. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 440 may include one or more of any type of network interface (e.g., a network interface card (NIC)) wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a World Wide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, a near field communication (NFC) interface, and the like.

[0161] In one embodiment of the present specification, the above components of the computing device 400 and Figure 4 Other components not shown in the figure may also be connected to each other, for example, via a bus. It should be understood that Figure 4 The computing device structure block diagram shown is only for the purpose of illustration, and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0162] Computing device 400 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook, etc.), a mobile phone (e.g., a smart phone), a wearable computing device (e.g., a smart watch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or PC. Computing device 400 may also be a mobile or stationary server.

[0163] The processor 420 is used to execute the following computer executable instructions, which, when executed by the processor, implement the steps of the above-mentioned fault detection model evaluation index recommendation method based on clustering algorithm.

[0164] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the fault detection model evaluation index recommendation method based on the clustering algorithm are of the same concept, and the details not described in detail in the technical scheme of the computing device can be found in the description of the technical scheme of the fault detection model evaluation index recommendation method based on the clustering algorithm.

[0165] An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the above-mentioned method for recommending evaluation indicators of a fault detection model based on a clustering algorithm.

[0166] The above is a schematic scheme of a computer-readable storage medium of this embodiment. It should be noted that the technical scheme of the storage medium and the technical scheme of the fault detection model evaluation index recommendation method based on the clustering algorithm are of the same concept, and the details not described in detail in the technical scheme of the storage medium can be found in the description of the technical scheme of the fault detection model evaluation index recommendation method based on the clustering algorithm.

[0167] An embodiment of the present specification further provides a computer program, wherein when the computer program is executed in a computer, the computer is instructed to execute the steps of the above-mentioned method for recommending evaluation indicators of a fault detection model based on a clustering algorithm.

[0168] The above is a schematic scheme of a computer program of this embodiment. It should be noted that the technical scheme of the computer program and the technical scheme of the fault detection model evaluation index recommendation method based on the clustering algorithm are of the same concept, and the details not described in detail in the technical scheme of the computer program can be found in the description of the technical scheme of the fault detection model evaluation index recommendation method based on the clustering algorithm.

[0169] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0170] The computer instructions include computer program codes, which may be in source code form, object code form, executable files or some intermediate forms, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0171] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

[0172] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0173] The preferred embodiments of this specification disclosed above are only used to help explain this specification. The optional embodiments do not describe all the details in detail, nor do they limit the invention to only the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that technicians in the relevant technical field can well understand and use this specification. This specification is only limited by the claims and their full scope and equivalents.

Claims

1. A fault detection model evaluation index recommendation method based on clustering algorithm, comprising: Obtain device operation information and multiple historical evaluation indicators corresponding to each historical device among multiple historical devices; Encoding the device operation information and the multiple historical evaluation indicators to generate a coding vector corresponding to each historical device; Clustering the plurality of historical devices based on the encoding vectors by using a target clustering algorithm to generate corresponding clustering results; Determine the evaluation index to be recommended corresponding to each historical device according to the clustering result.

2. According to the clustering algorithm-based fault detection model evaluation index recommendation method of claim 1, the target clustering algorithm comprises a DBSCAN clustering algorithm; Accordingly, clustering the plurality of historical devices by using a target clustering algorithm and based on the encoding vectors includes: Determine the cluster radius corresponding to each device category and the number of historical devices contained in each device category; The plurality of historical devices are clustered by using the DBSCAN clustering algorithm and based on the clustering radius, the number and the encoding vector.

3. According to the method for recommending evaluation indicators of a fault detection model based on a clustering algorithm according to claim 1, the step of determining the evaluation indicator to be recommended corresponding to each historical device according to the clustering result comprises: Determine a first device category to which the first historical device belongs according to the clustering result, and determine second historical devices included in the first device category, wherein the first historical device is each of the plurality of historical devices, and the second historical device is each other historical device included in the first device category except the first historical device; According to a plurality of historical evaluation indicators respectively corresponding to the first historical device and the second historical device, a to-be-recommended evaluation indicator corresponding to the first historical device is determined.

4. According to the clustering algorithm-based fault detection model evaluation index recommendation method of claim 3, the step of determining the evaluation index to be recommended corresponding to the first historical device according to the plurality of historical evaluation indexes respectively corresponding to the first historical device and the second historical device comprises: Determine the overlapping historical evaluation indicators among the plurality of historical evaluation indicators respectively corresponding to the first historical device and the second historical device; According to the multiple historical evaluation indicators corresponding to the first historical device and the overlapping historical evaluation indicators, the to-be-recommended evaluation indicator corresponding to the first historical device is determined.

5. The method for recommending fault detection model evaluation indicators based on clustering algorithm according to claim 1, further comprising: Acquire device operation information corresponding to the target device, and encode the device operation information corresponding to the target device to generate a target encoding vector corresponding to the target device; Determining a second device category to which the target device belongs according to the target coding vector and the clustering result; Determining the similarity between the target device and each historical device in the second device category according to the target coding vector and the coding vector corresponding to each historical device in the second device category; Determine the evaluation index to be recommended corresponding to the target device according to the similarity.

6. According to the clustering algorithm-based fault detection model evaluation index recommendation method of claim 5, the step of determining the evaluation index to be recommended corresponding to the target device according to the similarity comprises: determining a maximum value among similarities between the target device and each historical device included in the second device category; The target historical device corresponding to the maximum value is determined, and the evaluation index to be recommended corresponding to the target historical device is determined as the evaluation index to be recommended corresponding to the target device.

7. According to the clustering algorithm-based fault detection model evaluation index recommendation method according to claim 1, the equipment operation information includes operating condition information, equipment fault information, data processing information, etc.

8. A fault detection model evaluation index recommendation device based on clustering algorithm, comprising: An acquisition module is configured to acquire device operation information and a plurality of historical evaluation indicators corresponding to each of a plurality of historical devices; An encoding module, configured to encode the device operation information and the multiple historical evaluation indicators to generate an encoding vector corresponding to each historical device; A clustering module, configured to cluster the plurality of historical devices based on the encoding vectors by using a target clustering algorithm, and generate corresponding clustering results; The determination module is configured to determine the evaluation index to be recommended corresponding to each historical device according to the clustering result.

9. A computing device comprising: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the fault detection model evaluation index recommendation method based on clustering algorithm according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for recommending evaluation indicators of a fault detection model based on a clustering algorithm as described in any one of claims 1 to 7.

Images