Fault positioning method and device, electronic equipment and storage medium

Abstract

The application provides a fault positioning method and device, electronic equipment and storage medium. The method comprises: acquiring a plurality of sensor parameters and alarm information of a target device; determining abnormal sensor information in the plurality of sensor parameters; performing splicing processing on the abnormal sensor information and the alarm information to obtain an alarm statement; inputting the alarm statement into a language model to obtain fault component information of the target device, wherein the language model is obtained by training alarm statements and corresponding component information. The language model in the above scheme can determine fault component information according to sensor parameters and alarm information of a target device, and the efficiency of positioning fault components is higher than relying on artificial experience.

Description

Technical Field

[0001] This application relates to the field of fault detection, and more particularly to a fault location method, apparatus, electronic device, and storage medium. Background Technology

[0002] In semiconductor manufacturing, etching equipment is used to create wafers for semiconductor production. During the manufacturing process, the operating status of the etching equipment has a significant impact on wafer fabrication. Etching equipment is costly; therefore, improving its maintenance efficiency and reducing downtime due to malfunctions are crucial for increasing semiconductor manufacturing efficiency and controlling costs.

[0003] The general etching process includes steps such as silicon wafer adsorption, gas flow and chamber pressure regulation, pre-etching, main etching, over-etching, and chamber cleaning. To ensure safe production and consistent product quality, quality monitoring and fault detection are required during the etching process. To monitor the production process, etching equipment is often equipped with numerous sensors to collect real-time data on monitored variables such as chamber pressure, gas flow, temperature, power, and spectral signals.

[0004] Currently, there are several methods for fault detection in etching equipment. One method marks wafers that trigger alarms as negative examples and wafers that do not trigger alarms as positive examples. Algorithms are used to classify wafers and calculate the weight of different sensors. Sensors with higher weights are considered abnormal. Another method uses algorithms to find outlier wafers and determines the abnormal sensor based on distance. Neither of these methods can pinpoint the component requiring repair; they only trace it back to the abnormal sensor. The next step requires engineers' experience to identify and repair the faulty component. The engineers' experience in identifying faulty components leads to low timeliness, impacting semiconductor production efficiency. Summary of the Invention

[0005] This application provides a fault location method, apparatus, electronic device, and storage medium to improve fault location efficiency.

[0006] In a first aspect, this application provides a fault location method, comprising: acquiring multiple sensor parameters and alarm information of a target device; identifying abnormal sensor information among the multiple sensor parameters; concatenating the abnormal sensor information with the alarm information to obtain an alarm statement; and inputting the alarm statement into a language model to obtain faulty component information of the target device, wherein the language model is obtained by training the alarm statement and the corresponding faulty component information.

[0007] In one possible implementation, identifying anomalous sensor information from the plurality of sensor parameters includes: processing the plurality of sensor parameters based on a random forest regression algorithm to obtain a weight for each sensor parameter; and identifying the anomalous sensor information based on the plurality of sensor parameters and the weight of each sensor parameter.

[0008] In one possible implementation, determining abnormal sensor information based on the plurality of sensor parameters and the weight of each sensor parameter includes: normalizing the weight of each sensor parameter to obtain a normalized weight for each sensor parameter; and sampling the plurality of sensor parameters with replacement using the plurality of sensor parameters as samples and the normalized weight of each sensor parameter as a sampling probability to obtain the abnormal sensor information.

[0009] In one possible implementation, inputting the alarm statement into a language model to obtain faulty component information of the target device includes: generating a word vector of the alarm statement based on the alarm statement; inputting the word vector of the alarm statement into the language model to obtain the encoding of the faulty component; and decoding the encoding of the faulty component to obtain the faulty component information of the target device.

[0010] In one possible implementation, the method further includes: acquiring the training alarm statement and the corresponding component information; generating word vectors for the training alarm statement based on the training alarm statement, performing one-hot vector encoding on the corresponding component information to obtain the encoding of the corresponding component; training the model using the word vectors of the training alarm statement and the encoding of the corresponding component until the model accuracy converges, thereby obtaining the language model.

[0011] In one possible implementation, obtaining multiple sensor parameters of a target device includes: obtaining multiple sensor parameters to be screened under the target device, wherein the multiple sensor parameters to be screened include multiple component features, each component feature having a corresponding proportion coefficient; and performing a screening process on the multiple sensor parameters to be screened based on a dimensionality reduction algorithm, deleting component features with a proportion coefficient less than a threshold, thereby obtaining the multiple sensor parameters.

[0012] In one possible implementation, the sensor parameters to be screened include statistical parameters, time period parameters, and cumulative parameters; obtaining multiple sensor parameters to be screened under the target device includes: determining multiple preset times, determining time period parameters based on the multiple preset times; obtaining multiple sensor data of the target device under the preset multiple times; performing statistical processing on the multiple sensor data to obtain the statistical parameters; obtaining the cumulative parameters based on the multiple sensor data and the time period parameters; and determining the statistical parameters, the time period parameters, and the cumulative parameters as the sensor parameters to be screened.

[0013] Secondly, this application provides a fault location device, comprising: an acquisition module for acquiring multiple sensor parameters and alarm information of a target device; a determination module for determining abnormal sensor information from the multiple sensor parameters; a splicing module for splicing the abnormal sensor information with the alarm information to obtain an alarm statement; and a processing module for inputting the alarm statement into a language model to obtain faulty component information of the target device, wherein the language model is obtained by training the alarm statement and the corresponding faulty component information.

[0014] In one possible implementation, the determining module is specifically used to process the plurality of sensor parameters based on a random forest regression algorithm to obtain the weight of each sensor parameter; the determining module is also specifically used to determine abnormal sensor information based on the plurality of sensor parameters and the weight of each sensor parameter.

[0015] In one possible implementation, the determining module is specifically used to normalize the weight of each sensor parameter to obtain a normalized weight for each sensor parameter; the determining module is also specifically used to sample the plurality of sensor parameters with replacement using the plurality of sensor parameters as samples and the normalized weight of each sensor parameter as a sampling probability to obtain the abnormal sensor information.

[0016] In one possible implementation, the device further includes: an execution module, configured to generate a word vector of an alarm statement based on the alarm statement; the execution module is further configured to input the word vector of the alarm statement into the language model to obtain the encoding of the faulty component; the execution module is further configured to decode the encoding of the faulty component to obtain faulty component information of the target device.

[0017] In one possible implementation, the device further includes: a training module, configured to acquire the training alarm statement and the corresponding component information; the training module is further configured to generate word vectors of the training alarm statement based on the training alarm statement, and perform one-hot vector encoding processing on the corresponding component information to obtain the encoding of the corresponding component; the training module is further configured to train the model using the word vectors of the training alarm statement and the encoding of the corresponding component until the model accuracy converges, thereby obtaining the language model.

[0018] In one possible implementation, the device further includes: a screening module, configured to acquire multiple sensor parameters to be screened under the target device, the multiple sensor parameters to be screened including multiple component features, each component feature having a corresponding proportion coefficient; the screening module is further configured to perform screening processing on the multiple sensor parameters to be screened based on a dimensionality reduction algorithm, deleting component features with proportion coefficients less than a threshold, to obtain the multiple sensor parameters.

[0019] In one possible implementation, the filtering module is specifically configured to determine multiple preset times and determine time period parameters based on the multiple preset times; the filtering module is further configured to acquire multiple sensor data of the target device at the preset multiple times; the filtering module is further configured to perform statistical processing on the multiple sensor data to obtain the statistical parameters; the filtering module is further configured to obtain the cumulative parameters based on the multiple sensor data and the time period parameters; the filtering module is further configured to determine the statistical parameters, the time period parameters, and the cumulative parameters as the sensor parameters to be filtered.

[0020] Thirdly, this application provides an electronic device, including: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method described in any one of the first aspects.

[0021] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which are executed by a processor as described in any one of the first aspects.

[0022] The fault location method, apparatus, electronic device, and storage medium provided in this application acquire multiple sensor parameters and alarm information of a target device; determine abnormal sensor information from the multiple sensor parameters; concatenate the abnormal sensor information with the alarm information to obtain an alarm statement; input the alarm statement into a language model to obtain faulty component information of the target device, wherein the language model is obtained by training the alarm statement and the corresponding faulty component information. The language model in the above scheme can determine faulty component information based on the sensor parameters and alarm information of the target device, which is more efficient than relying on human experience in locating faulty components. Attached Figure Description

[0023] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0024] Figure 1 This is a schematic diagram illustrating an application scenario of a fault location method provided in an embodiment of this application.

[0025] Figure 2 A flowchart illustrating a fault location method provided in an embodiment of this application;

[0026] Figure 3 This is a schematic diagram illustrating the determination of sensor parameters in an embodiment of this application.

[0027] Figure 4 A schematic diagram of the sensor for determining anomalies provided in an embodiment of this application;

[0028] Figure 5 A schematic diagram of language model training provided for an embodiment of this application;

[0029] Figure 6 This is a schematic diagram of the structure of a fault location device provided in an embodiment of this application;

[0030] Figure 7 This is a schematic diagram of the structure of a fault location device provided in an embodiment of this application;

[0031] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0032] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0033] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0034] Figure 1 This diagram illustrates an application scenario of a fault location method provided in this application. Taking the illustrated scenario as an example: an etching apparatus includes multiple working components for producing semiconductor wafers. These components work collaboratively to achieve the etching process. The etching apparatus also includes multiple sensors used to monitor the etching process and collect and record the operating parameters of the etching apparatus, such as pressure, gas flow rate, temperature, power, and spectral signals. Analysis of these operating parameters can identify abnormal parameters, and analysis of these abnormal parameters can pinpoint the faulty component within the etching apparatus.

[0035] The technical solutions of this application will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. In the description of this application, unless otherwise expressly specified and limited, the terms should be broadly understood within the art. The embodiments of this application will now be described with reference to the accompanying drawings.

[0036] Figure 2 This is a flowchart illustrating a fault location method provided in an embodiment of this application, as shown below. Figure 2 As shown, the method includes the following steps:

[0037] S201. Obtain multiple sensor parameters and alarm information of the target device.

[0038] As an example, the execution subject of this embodiment can be a detection device for locating faulty components, and the detection device can be implemented in various ways. For example, it can be program software, or a medium storing relevant computer programs, such as a USB flash drive; or, the device can also be a physical device that integrates or installs relevant computer programs, such as a chip, a smart terminal, a computer, a server, etc.

[0039] The target device package can collect sensor parameters from all sensors of the target device and monitor its operating status. If the target device's operating status is abnormal, it will issue an alarm. For example, if the target device's current operating temperature is 90℃, exceeding the temperature threshold of 85℃, an overheating alarm will be issued. At this point, it can only be determined that the target device is overheating, but not which specific component is malfunctioning and causing the overheating. Therefore, it is necessary to locate the faulty component to resolve the overheating problem.

[0040] Optionally, multiple sensor parameters and alarm information can be obtained from the target device's management center.

[0041] Alternatively, sensor parameters can be obtained directly from the sensor.

[0042] One feasible implementation method is to obtain multiple sensor parameters by: obtaining multiple sensor parameters to be screened under the target device, wherein the multiple sensor parameters to be screened include multiple component features, each component feature having a corresponding proportion coefficient; and performing a screening process on the multiple sensor parameters to be screened based on a dimensionality reduction algorithm, deleting component features with a proportion coefficient less than a threshold, thereby obtaining the multiple sensor parameters.

[0043] For example, the sensor parameters to be screened include statistical parameters, time-period parameters, and cumulative parameters. The components of statistical parameters include, but are not limited to, average, maximum, minimum, variance, skewness, kurtosis, median, and percentage. The components of time-period parameters include, but are not limited to, the number of abrupt change points, window mean, trend, period, and autocorrelation.

[0044] Optionally, while retaining a preset percentage of the sensor parameters to be screened, based on a dimensionality reduction algorithm, the constituent features with a proportion coefficient less than a threshold are deleted.

[0045] Optionally, principal component analysis (PCA) can be used for screening in dimensionality reduction algorithms. It should be noted that this application does not limit the specific type of dimensionality reduction algorithm.

[0046] It is understandable that compositional features with a proportion coefficient less than the threshold have a smaller effect on identifying and locating faulty components, which will affect the accuracy of locating faulty components.

[0047] In this feasible implementation, the accuracy of locating faulty components can be improved by deleting component features whose proportion coefficient is less than a threshold.

[0048] Furthermore, a feasible implementation method is to obtain multiple sensor parameters to be screened through the following steps: determining multiple preset times, determining time period parameters based on the multiple preset times; acquiring multiple sensor data of the target device at the preset multiple times; performing statistical processing on the multiple sensor data to obtain the statistical parameters; obtaining the cumulative parameters based on the multiple sensor data and the time period parameters; and determining the statistical parameters, the time period parameters, and the cumulative parameters as the sensor parameters to be screened.

[0049] Optionally, the engineer can specify multiple preset times or set periodic preset times, such as obtaining a preset time every 10 minutes.

[0050] Optionally, the cumulative value of some sensor data over a certain period of time has practical significance. The cumulative parameter can be obtained by multiplying the sensor data at each preset time with the time difference at the preset time.

[0051] Below, in conjunction with Figure 3 The above feasible implementation methods are explained.

[0052] Figure 3 This is a schematic diagram illustrating the determination of sensor parameters for an embodiment of this application. For example... Figure 3 As shown, the first step is to determine the preset time and acquire the sensor parameters. Statistics are then performed on the preset time to obtain time period parameters, and statistical parameters are obtained from the sensor data. The sensor data is then accumulated according to the time period parameters to obtain cumulative parameters. These statistical parameters, time period parameters, and cumulative parameters are then used as the sensor parameters to be screened.

[0053] In this feasible implementation, obtaining various types of sensor parameters to be screened allows for the location of faulty components from multiple angles, improving the accuracy of faulty component location.

[0054] S202, Determine abnormal sensor information from the multiple sensor parameters.

[0055] One feasible approach is to determine abnormal sensor information by: processing the multiple sensor parameters based on a random forest regression algorithm to obtain the weight of each sensor parameter; and determining the abnormal sensor information based on the multiple sensor parameters and the weight of each sensor parameter.

[0056] Optionally, the sensor parameters are used as input to the random forest regression algorithm, and whether an alarm is triggered is used as the output of the random forest regression algorithm. After multiple iterations, the iteration stops when the accuracy of the random forest regression algorithm is greater than the accuracy threshold, and the weights corresponding to the sensor parameters at this time are saved.

[0057] For example, an accuracy threshold of 97% can be set, with the goal of stopping the iteration when the accuracy of the random forest regression algorithm is greater than 97%.

[0058] Optionally, whether an alarm is triggered is output by the random forest regression algorithm, and the resulting weights can be used to evaluate the accuracy of sensor parameters in locating faulty components.

[0059] In this feasible implementation, prioritizing sensor parameters with higher weights can improve the accuracy of locating faulty components.

[0060] Furthermore, a feasible implementation method can be used to determine the abnormal sensor information by: normalizing the weight of each sensor parameter to obtain the normalized weight of each sensor parameter; using the multiple sensor parameters as samples and the normalized weight of each sensor parameter as the sampling probability, sampling the multiple sensor parameters with replacement to obtain the abnormal sensor information.

[0061] Optionally, a sampling number can be set to perform sampling with replacement until the sampling number is reached, thereby obtaining the number of abnormal sensor information samples.

[0062] For example, if the sampling count is set to 30, a total of 30 abnormal sensor information will be identified through sampling. Sensor parameters with high weights will be sampled multiple times. It can be understood that the number of samplings with replacement used in this application can be greater than the number of sensor parameters.

[0063] It should be noted that this application does not restrict the specific sampling method.

[0064] Optionally, the abnormal sensor information determined by sampling can be arranged in the order of sampling to obtain an abnormal sensor information queue, and the order in the queue reflects the importance of the abnormal sensor information.

[0065] In this feasible implementation, an abnormal sensor information queue is obtained by sampling. The importance of the abnormal sensor information can be reflected by the queue order and the frequency of occurrence of the abnormal sensor information, thereby improving the accuracy of locating faulty components.

[0066] Below, in conjunction with Figure 4 The above feasible implementation methods are explained.

[0067] Figure 4 This is a schematic diagram of a sensor for determining anomalies provided in an embodiment of this application. Figure 4As shown, the sensor parameters are processed using a random forest regression algorithm to obtain the weight of each sensor parameter. The weights of each sensor parameter are then normalized to obtain normalized weights. Based on the sensor parameters and normalized weights, sampling is performed to obtain abnormal sensor information.

[0068] S203. The abnormal sensor information and the alarm information are concatenated to obtain an alarm statement.

[0069] Optionally, multiple abnormal sensor information can be concatenated with commas and then combined with the alarm statement.

[0070] For example, alarm statements could be: "Target device temperature too high" and "Wafer chamber pressure too high".

[0071] S204. Input the alarm statement into the language model to obtain the faulty component information of the target device. The language model is obtained by training the alarm statement and the faulty component information.

[0072] One feasible implementation method is to train a language model by: acquiring the training alarm statement and the corresponding component information; generating the language of the training alarm statement based on the training alarm statement, performing one-hot vector encoding on the corresponding component information to obtain the encoding of the corresponding component; training the model using the word vectors of the training alarm statement and the encoding of the corresponding component until the model accuracy converges, thus obtaining the language model.

[0073] Optionally, training alarm statements and training faulty component information can be obtained from papers, operation manuals, target device log files, and device manuals.

[0074] Optionally, for a target device, the number of components is limited, so the components can be processed using one-hot vector encoding.

[0075] Below, in conjunction with Figure 5 This feasible implementation method will be explained.

[0076] Figure 5 This is a schematic diagram illustrating language model training as provided in an embodiment of this application. Figure 5 As shown, the training alarm statements and their corresponding component information are processed to obtain word vectors for the training alarm statements and codes for the corresponding components. The model is trained using the word vectors as input and the codes as output until the model accuracy converges. Training then stops, and the current model is used as the final language model. It can be understood that the training alarm statements and their corresponding component information obtained from the data have established a mapping relationship based on historical experience, and the language model trained using this mapping relationship can make predictions.

[0077] In this feasible implementation, the language model trained using existing data, training alarm statements, and corresponding component information yields higher accuracy.

[0078] One feasible implementation method is to obtain the faulty component information of the target device by: generating a word vector of the alarm statement based on the alarm statement; inputting the word vector of the alarm statement into the language model to obtain the encoding of the faulty component; and decoding the encoding of the faulty component to obtain the faulty component information of the target device.

[0079] Optionally, the alarm statement includes multiple sensor parameters and alarm information. The alarm statement is generated as a whole into a word vector. A word vector includes the entire content of the alarm statement. The code of the faulty component output by the language model based on a word vector is calculated by combining multiple sensor parameters and alarm information.

[0080] In this feasible implementation, word vectors containing the entire content of the alarm statement are used as input to the language model. This allows for the integration of multiple sensor parameters and alarm information, thereby improving the accuracy of locating faulty components.

[0081] Figure 6 This is a schematic diagram of a fault location device provided in an embodiment of this application. Figure 6 As shown, the fault location device 60 may include: an acquisition module 61, a determination module 62, a splicing module 63, and a processing module 64.

[0082] in,

[0083] The acquisition module 61 is used to acquire multiple sensor parameters and alarm information of the target device.

[0084] The determining module 62 is used to determine abnormal sensor information from the plurality of sensor parameters;

[0085] The splicing module 63 is used to splice the abnormal sensor information and the alarm information to obtain an alarm statement;

[0086] The processing module 64 is used to input the alarm statement into a language model to obtain the fault component information of the target device. The language model is obtained by training the alarm statement and the fault component information.

[0087] Optionally, module 61 can be executed. Figure 2 S201 in the embodiment.

[0088] Optionally, module 62 can be executed. Figure 2 S202 in the embodiment.

[0089] Optionally, the splicing module 63 can perform... Figure 2 S203 in the embodiment.

[0090] Optionally, processing module 63 can execute Figure 2 S204 in the embodiment.

[0091] It should be noted that the encoding device shown in the embodiments of this application can execute the technical solution shown in the above method embodiments, and its implementation principle and beneficial effects are similar, so they will not be described again here.

[0092] In one possible implementation, the determining module 62 is specifically used for:

[0093] Based on the random forest regression algorithm, the multiple sensor parameters are processed to obtain the weight of each sensor parameter;

[0094] Abnormal sensor information is determined based on the multiple sensor parameters and the weight of each sensor parameter.

[0095] In one possible implementation, the determining module is specifically used to: normalize the weights of each sensor parameter to obtain the normalized weights of each sensor parameter;

[0096] By using the multiple sensor parameters as samples and the normalized weight of each sensor parameter as the sampling probability, the multiple sensor parameters are sampled with replacement to obtain the abnormal sensor information.

[0097] Figure 7 This is a schematic diagram of the structure of a fault location device provided in an embodiment of this application. Figure 6 Based on the illustrated embodiments, as Figure 7 As shown, the fault location device 60 further includes: an execution module 65, a training module 66, and a screening module 67, wherein,

[0098] The execution module 65 is used to generate word vectors for the alarm statement based on the alarm statement;

[0099] The word vectors of the alarm statement are input into the language model to obtain the encoding of the faulty component;

[0100] The encoding of the faulty component is decoded to obtain the faulty component information of the target device.

[0101] The training module 66 is used to acquire the training alarm statement and the corresponding component information;

[0102] Based on the training alarm statement, word vectors of training alarm statements are generated, and one-hot vector encoding is performed on the corresponding component information to obtain the encoding of the training faulty component.

[0103] The language model is obtained by training the model using the word vectors of the training alarm statements and the encoding of the training faulty components until the model accuracy converges.

[0104] The filtering module 67 is used to acquire multiple sensor parameters to be filtered under the target device. The multiple sensor parameters to be filtered include multiple component features, and each component feature has a corresponding proportion coefficient.

[0105] Based on the dimensionality reduction algorithm, the multiple sensor parameters to be screened are processed by deleting the constituent features whose proportion coefficient is less than the threshold, and thus obtaining the multiple sensor parameters.

[0106] The filtering module 67 is specifically used to determine multiple preset times, determine time period parameters based on the multiple preset times, and acquire multiple sensor data of the target device at the preset multiple times.

[0107] The statistical parameters are obtained by performing statistical processing on the data from the multiple sensors.

[0108] The cumulative parameter is obtained based on the data from the multiple sensors and the time period parameter;

[0109] The statistical parameters, the time period parameters, and the cumulative parameters are determined as the sensor parameters to be screened.

[0110] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, such as... Figure 8 As shown, the electronic device includes:

[0111] The electronic device includes a processor 291 and a memory 292; it may also include a communication interface 293 and a bus 294. The processor 291, memory 292, and communication interface 293 can communicate with each other via the bus 294. The communication interface 293 can be used for information transmission. The processor 291 can invoke logical instructions stored in the memory 292 to execute the methods of the above embodiments.

[0112] Furthermore, the logic instructions in the aforementioned memory 292 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.

[0113] The memory 292, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of this application. The processor 291 executes functional applications and data processing by running the software programs, instructions, and modules stored in the memory 292, thereby implementing the methods in the above-described method embodiments.

[0114] The memory 292 may include a program storage area and a data storage area. The program storage area may store the operating system and application programs required for at least one function; the data storage area may store data created based on the use of the terminal device. Furthermore, the memory 292 may include high-speed random access memory and may also include non-volatile memory.

[0115] This application provides a non-transitory computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the methods described in the foregoing embodiments.

[0116] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0117] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0118] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A fault location method, characterized in that, include: Acquire multiple sensor parameters and alarm information of the target device, wherein the sensor parameters include statistical parameters, time period parameters, and cumulative parameters; Determining abnormal sensor information from the multiple sensor parameters includes: using the sensor parameters as input to a random forest regression algorithm and whether an alarm is triggered as the output of the random forest regression algorithm; iterating through multiple rounds until the accuracy of the random forest regression algorithm is greater than an accuracy threshold, then stopping the iteration; saving the weights corresponding to the sensor parameters at this point, whereby the weights are used to evaluate the accuracy of the sensor parameters in locating faulty components; normalizing the weights of each sensor parameter to obtain a normalized weight for each sensor parameter; and sampling the multiple sensor parameters with replacement using the multiple sensor parameters as samples and the normalized weights of each sensor parameter as sampling probabilities to obtain the abnormal sensor information. The abnormal sensor information determined through sampling is arranged in the order of sampling to form an abnormal sensor information queue, where the order in the queue reflects the importance of the abnormal sensor information. The abnormal sensor information and the alarm information are concatenated to obtain an alarm statement, and a word vector of the alarm statement is generated based on the alarm statement. The word vectors of the alarm statement are input into a language model to obtain the encoding of the faulty component of the target device. The encoding of the faulty component is then decoded to obtain the faulty component information of the target device. The language model is obtained by training the word vectors of the alarm statement and the encoding of the corresponding component.

2. The method according to claim 1, characterized in that, The method further includes: Obtain the training alarm statement and the corresponding component information; Based on the training alarm statement, word vectors of the training alarm statement are generated, and the corresponding component information is encoded to obtain the code of the corresponding component. The language model is obtained by training the model using the word vectors of the training alarm statements and the encoding of the corresponding components until the model accuracy converges.

3. The method according to claim 2, characterized in that, Acquire multiple sensor parameters of the target device, including: Acquire multiple sensor parameters to be screened under the target device. The multiple sensor parameters to be screened include multiple component features, and each component feature has a corresponding proportion coefficient. Based on the dimensionality reduction algorithm, the multiple sensor parameters to be screened are processed by deleting the constituent features whose proportion coefficient is less than the threshold, and thus obtaining the multiple sensor parameters.

4. The method according to claim 3, characterized in that, The sensor parameters to be screened include statistical parameters, time period parameters, and cumulative parameters; Obtain multiple sensor parameters to be screened under the target device, including: Determine multiple preset times, and determine time period parameters based on the multiple preset times; Acquire multiple sensor data of the target device at preset multiple time points; The statistical parameters are obtained by performing statistical processing on the data from the multiple sensors. The cumulative parameter is obtained based on the data from the multiple sensors and the time period parameter; The statistical parameters, the time period parameters, and the cumulative parameters are determined as the sensor parameters to be screened.

5. A fault location device, characterized in that, include: The acquisition module is used to acquire multiple sensor parameters and alarm information of the target device; The determination module is used to determine abnormal sensor information from the plurality of sensor parameters, including: using the sensor parameters as input to a random forest regression algorithm, using whether an alarm is triggered as the output of the random forest regression algorithm, iterating through multiple rounds until the accuracy of the random forest regression algorithm is greater than an accuracy threshold, then stopping the iteration, and saving the weights corresponding to the sensor parameters at this time, the weights being used to evaluate the accuracy of the sensor parameters in locating faulty components; normalizing the weights of each sensor parameter to obtain the normalized weights of each sensor parameter; using the plurality of sensor parameters as samples and the normalized weights of each sensor parameter as sampling probabilities, sampling the plurality of sensor parameters with replacement to obtain the abnormal sensor information, wherein the abnormal sensor information determined by sampling is arranged in the order of sampling to obtain an abnormal sensor information queue, the order in the queue reflecting the importance of the abnormal sensor information; The splicing module is used to splice the abnormal sensor information and the alarm information to obtain an alarm statement, and generate a word vector of the alarm statement based on the alarm statement; The processing module is used to input the word vector of the alarm statement into a language model to obtain the encoding of the faulty component of the target device, and to decode the encoding of the faulty component to obtain the faulty component information of the target device; wherein, the language model is obtained by training the word vector of the alarm statement and the encoding of the corresponding component.

6. The apparatus according to claim 5, characterized in that, The device further includes: The training module is used to acquire the training alarm statement and the corresponding component information; The training module is also used to generate word vectors for training alarm statements based on the training alarm statements, and to perform one-hot vector encoding on the corresponding component information to obtain the encoding of the corresponding component. The training module is further used to train the model using the word vectors of the training alarm statement and the encoding of the corresponding component until the model accuracy converges, thus obtaining the language model.

7. The apparatus according to claim 6, characterized in that, The device further includes: The filtering module is used to acquire multiple sensor parameters to be filtered under the target device. The multiple sensor parameters to be filtered include multiple component features, and each component feature has a corresponding proportion coefficient. The filtering module is also used to filter the multiple sensor parameters to be filtered based on a dimensionality reduction algorithm, and delete the constituent features with a proportion coefficient less than a threshold to obtain the multiple sensor parameters.

8. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1-4.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-4.

Images