Voiceprint vector library construction method and device, electronic equipment and storage medium

Abstract

The invention discloses a voiceprint vector library construction method and device, electronic equipment and a storage medium, and relates to the technical field of data processing. The voiceprint vector library construction method comprises the following steps: determining target voiceprint data information of power transmission, transformation and distribution equipment, determining target vector information according to the target voiceprint data information, and determining data exception information of the power transmission, transformation and distribution equipment according to the target vector information; constructing a data table according to a preset type based on a preset data model, and storing the target vector information and the data exception information to the corresponding data table; and determining a partition type of the data table, and creating a target vector index of each data table according to the partition type to obtain a voiceprint vector library. According to the method and the device, the target voiceprint data information and the equipment abnormal information are vectorized, so that discrete variables of the target voiceprint data information and the equipment abnormal information are converted into continuous vectors, the voiceprint vector library is established, and the searching efficiency of the target voiceprint data information is improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, device, electronic device and storage medium for constructing a voiceprint vector library. Background Art

[0002] The power transmission, transformation and distribution equipment such as substation main transformers, reactors, switch cabinets, overhead transmission lines, cables, distribution transformers, pole switches, switch cabinets, and ring network cabinets produce different sounds during operation. The voiceprint data information during the operation of the equipment can be collected to diagnose and monitor the operating status of the power transmission, transformation and distribution equipment through the voiceprint data information.

[0003] In order to facilitate the determination of the operation failure of the power transmission and distribution equipment through voiceprint data information, the voiceprint data information and the failure information of the power transmission and distribution equipment can be stored in the database. Most databases store data in a structured manner. The structured data has the same format and mode, and can generally be stored and quickly retrieved using the table form of a relational database. However, the amount of voiceprint data information is large, and its data scale greatly exceeds that of traditional relational databases. In addition, the data structure of voiceprint data information and the operation failure of power transmission and distribution equipment is irregular, and traditional relational databases are not convenient for the storage and search of voiceprint data information. Therefore, how to construct a voiceprint vector library to facilitate the search of voiceprint data information has become a problem that needs to be solved urgently. Summary of the invention

[0004] The present invention provides a method, device, electronic device and storage medium for constructing a voiceprint vector library to solve the problem in the prior art that it is inconvenient to store and search voiceprint data information through traditional relational databases. It is necessary to establish a voiceprint vector library to improve the efficiency and accuracy of searching target voiceprint data information.

[0005] According to one aspect of the present invention, a method for constructing a voiceprint vector library is provided, wherein the method comprises:

[0006] Determine target voiceprint data information of the power transmission, transformation and distribution equipment, determine target vector information according to the target voiceprint data information, and determine data anomaly information of the power transmission, transformation and distribution equipment according to the target vector information;

[0007] Building a data table according to a preset type based on a preset data model, and storing the target vector information and the data anomaly information in a corresponding data table;

[0008] The partition type of the data table is determined, and a target vector index of each data table is created according to the partition type to obtain a voiceprint vector library.

[0009] According to another aspect of the present invention, a device for constructing a voiceprint vector library is provided, wherein the device comprises:

[0010] An information determination module, used to determine target voiceprint data information of power transmission, transformation and distribution equipment, determine target vector information according to the target voiceprint data information, and determine data anomaly information of the power transmission, transformation and distribution equipment according to the target vector information;

[0011] A data table determination module, used to construct a data table according to a preset type based on a preset data model, and store the target vector information and the data anomaly information in a corresponding data table;

[0012] The vector library construction module is used to determine the partition type of the data table, and create a target vector index of each data table according to the partition type to obtain a voiceprint vector library.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the voiceprint vector library construction method described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the voiceprint vector library construction method described in any embodiment of the present invention when executed.

[0018] The technical solution of the embodiment of the present invention determines the target voiceprint data information of the power transmission and distribution equipment, determines the target vector information according to the target voiceprint data information, and determines the data anomaly information of the power transmission and distribution equipment according to the target vector information, so that the discrete variables of the target voiceprint data information and the equipment anomaly information are converted into continuous vectors; a data table is constructed according to a preset type based on a preset data model, and the target vector information and the data anomaly information are stored in the corresponding data table, the partition type of the data table is determined, and a target vector index of each data table is created according to the partition type to obtain a voiceprint vector library, thereby realizing the rapid establishment of the voiceprint vector library, avoiding the problem that the traditional relational database is inconvenient to store and search voiceprint data information, and improving the efficiency and accuracy of the search for the target voiceprint data information.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 is a flowchart of a method for constructing a voiceprint vector library provided according to Embodiment 1 of the present invention;

[0022] Figure 2 is a flowchart of a method for constructing a voiceprint vector library provided in accordance with Embodiment 2 of the present invention;

[0023] Figure 3 is a flowchart of a method for constructing a voiceprint vector library provided in accordance with Embodiment 3 of the present invention;

[0024] Figure 4 It is a structural schematic diagram of a voiceprint vector library construction device provided according to a fourth embodiment of the present invention;

[0025] Figure 5 It is a structural schematic diagram of an electronic device for implementing the voiceprint vector library construction method of an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] Embodiment 1

[0029] Figure 1 This is a flow chart of a method for constructing a voiceprint vector library according to the first embodiment of the present invention. This embodiment is applicable to the case where a voiceprint vector library is constructed for voiceprints of power transmission and distribution equipment. The method can be executed by a voiceprint vector library construction device. The voiceprint vector library construction device can be implemented in the form of hardware and / or software. The voiceprint vector library construction device can be configured in an electronic device. Figure 1 As shown, the method includes:

[0030] S110, determining target voiceprint data information of the power transmission, transformation and distribution equipment, determining target vector information according to the target voiceprint data information, and determining data anomaly information of the power transmission, transformation and distribution equipment according to the target vector information.

[0031] Among them, power transmission, transformation and distribution equipment refers to power equipment in the power transmission, transformation and distribution scenario, including power transmission equipment, power transformation equipment and power distribution equipment. Exemplarily, power transmission, transformation and distribution equipment may include but are not limited to transformers, reactors, switch cabinets, overhead transmission lines, cables, distribution transformers, pole switches, switch cabinets, ring network cabinets, etc. The target voiceprint data information can be understood as a data set obtained by collecting, analyzing and preprocessing the sounds generated during the operation of the power transmission, transformation and distribution equipment. In the actual operation process, the target voiceprint data information may include but is not limited to frequency characteristics, amplitude characteristics, harmonic characteristics, environmental parameters such as temperature and humidity, and the trend of sound changes over time. In one embodiment, the target voiceprint data information may also include the equipment type, equipment number and collection location of the power transmission, transformation and distribution equipment, so as to facilitate association with the power transmission, transformation and distribution equipment. During the application process, the voiceprint data of the power transmission and distribution equipment can be collected as the initial voiceprint data information, and the initial voiceprint data information can be cleaned, corrected, filtered and normalized to obtain the target voiceprint data information, eliminate the deviation, error, inconsistency and noise in the initial voiceprint data information, and improve the accuracy of the initial voiceprint data information.

[0032] The target vector information can be understood as the vector form obtained by vectorizing the target voiceprint data information and the device abnormality information. For example, the target vector information may include but is not limited to points, lines, and surfaces. In practical applications, the target voiceprint data information can be generated into the target vector information through manual drawing, point cloud processing, or data compression coding technology of preset cluster analysis.

[0033] Equipment abnormality information can be understood as abnormal information generated by the power transformation and distribution equipment during operation. For example, equipment abnormality information may include but is not limited to abnormal discharge, abnormal noise, mechanical failure, and no failure. In the actual operation process, equipment abnormality information can be determined based on the target vector information. For example, the frequency characteristics, amplitude characteristics, harmonic characteristics, environmental parameters such as temperature and humidity, and voiceprint characteristics such as target voiceprint data information contained in the target vector information can be determined. When there is an abnormality in the power transformation and distribution equipment, the voiceprint characteristics will change accordingly, and the target vector information will change accordingly. The abnormality in the target vector information can be determined to identify the equipment abnormality information of the power transmission, transformation and distribution equipment.

[0034] In an embodiment, the initial voiceprint data information of each power transmission, transformation and distribution equipment can be collected, the initial voiceprint data information can be preprocessed to obtain the target voiceprint data information, the target voiceprint data information can be vectorized to obtain the target vector information, and then the equipment abnormality information of the power transmission, transformation and distribution equipment can be determined by the target vector information. In the actual operation process, the power transmission, transformation and distribution equipment can be voiceprint sensed by a passive wireless voiceprint sensing module, the initial voiceprint data information can be collected, and the format of the initial voiceprint data information can be converted to a format suitable for vector data, and the noise in the initial voiceprint data information can be removed by wavelet analysis or Gaussian filtering as the target voiceprint data information. Then, the target voiceprint data information is vectorized to generate the target vector information by manual drawing, point cloud processing or preset cluster analysis data compression coding technology. The data abnormality information of the power transmission, transformation and distribution equipment is determined according to the target vector information by a preset vector machine and a preset model matching method.

[0035] S120 . Construct a data table according to a preset type based on a preset data model, and store the target vector information and data anomaly information in a corresponding data table.

[0036] Among them, the preset data model can be understood as a data model used to construct a data table. Exemplarily, the preset data model can include but is not limited to a vector model, a raster model, and an object-oriented data model. The preset type can be understood as a preset data table creation type. In the preset type, attribute information for creating a data table can be included, such as a data table structure, a data type, a data mode, and constraints. The number of data tables can be one or more. Exemplarily, the data table can include but is not limited to an original sample data table, a label sample data table, an original data detection record data table, a noise anomaly detection record data table, an abnormal discharge detection record data table, and a mechanical fault detection record data table.

[0037] In an embodiment, a preset type can be extracted, and attribute information such as a data table structure, a data type, a data mode, and a constraint condition of a data table can be determined according to the preset type, and a data table can be constructed according to the attribute information, and the field information of the target vector information and the data exception information associated with the field information can be determined, and the field information and the data exception information can be imported into the corresponding position of the corresponding data table. In the actual operation process, the field name of the data table can be determined, and the field information of the target vector information and the data exception information associated with the field information can be found according to the field name, and the field information of the target vector information and the data exception information can be imported into the data table according to the data type; or, the target vector information and the data exception information can be stored in the corresponding data table through a tool or scripting language provided by a database management system (DBMS). In an embodiment, there can be a preset correspondence between the preset type and the preset data model, and for different preset types, the corresponding preset data model can be selected to construct the data table.

[0038] S130: Determine the partition type of the data table, and create a target vector index for each data table according to the partition type to obtain a voiceprint vector library.

[0039] Among them, the partition type is understood as a pre-set method for dividing the data table into different sub-regions. Exemplarily, the partition type may include time partitioning and model partitioning, so as to improve query performance and facilitate data management. The target vector index is a data structure for storing and retrieving target vector information and data anomaly information. The target vector index can quickly find vector data similar to the vector to be queried, thereby improving the efficiency of data retrieval. Exemplarily, the target vector index may include IVF_FLAT index and Flat index. Among them, the IVF_FLAT index is an index structure based on an inverted file, which divides the vector into multiple small clusters (called buckets), and then uses Flat indexes in each bucket to speed up the search; Flat indexes store vectors (target vector information and data anomaly information) directly in memory or disk, and use the distance between vectors to calculate similarity.

[0040] For the partition type of the data table, the corresponding target vector index can be selected. Generally speaking, when the partition type is determined to be partitioned by time, the target vector index can use IVF_FLAT index; when the partition type is determined to be partitioned by model, the target vector index can use Flat index. The voiceprint vector library can be understood as a database for storing target vector information and data anomaly information, which is convenient for determining the corresponding data anomaly information according to the voiceprint information vector to be determined during the voiceprint recognition process. In one embodiment, the distance between vector information can be determined by Approximate Nearest Neighbor (ANN), thereby improving the retrieval speed of target vector information.

[0041] In an embodiment, the partition type of each data table can be determined. The partition type of each data table can be pre-set according to user needs; or it can be determined according to the data characteristics of the amount of data stored in the data table. Generally speaking, when the amount of data stored in the data table is large, the partition type can be partitioned by time; when the amount of data stored in the data table is small, the partition type can be partitioned by model. At the same time, the vector index corresponding to each partition type can be determined as the target vector index of the data table to obtain a voiceprint vector library.

[0042] The embodiment of the present invention determines the target voiceprint data information of the power transmission and distribution equipment, determines the target vector information according to the target voiceprint data information, and determines the data anomaly information of the power transmission and distribution equipment according to the target vector information, so that the discrete variables of the target voiceprint data information and the equipment anomaly information are converted into continuous vectors; a data table is constructed according to a preset type based on a preset data model, and the target vector information and the data anomaly information are stored in the corresponding data table, the partition type of the data table is determined, and a target vector index of each data table is created according to the partition type to obtain a voiceprint vector library, thereby realizing the rapid establishment of the voiceprint vector library, avoiding the problem that the traditional relational database is inconvenient to store and search voiceprint data information, and improving the efficiency and accuracy of the search for the target voiceprint data information.

[0043] In one embodiment, a method for constructing a voiceprint vector library further includes:

[0044] Determine the model parameters of the preset data model, and construct a model information data table associated with the model parameters in the voiceprint vector library.

[0045] Among them, model parameters can be understood as parameter information associated with a preset data model. For example, model parameters may include but are not limited to model name, model category, model version, model status, model capability vector data, model details description, model responsible person, update timestamp, and storage timestamp.

[0046] In an embodiment, model parameters such as model name, model category, model version, model status, model capability vector data, model detailed description, model responsible person, update timestamp, and storage timestamp of a preset data model can be determined, and a data table is constructed according to the model parameters as a model information data table associated with the model parameters. In an embodiment, since the amount of data of the model parameters is not large, the model information data table can be partitioned by model, and a Flat index can be used as the target vector index of the model to facilitate rapid query of the model parameters.

[0047] Embodiment 2

[0048] Figure 2 1 is a flow chart of a method for constructing a voiceprint vector library according to Embodiment 2 of the present invention. This embodiment is further optimized and expanded based on the above embodiment and can be combined with various optional technical solutions in the above embodiment. Figure 2 As shown, the method includes:

[0049] S210: Collect initial voiceprint data information of each power transmission, transformation and distribution equipment, perform data preprocessing on the initial voiceprint data information, and generate target voiceprint data information.

[0050] The initial voiceprint data information can be understood as a data set obtained by collecting and analyzing the sound generated during the operation of the power transmission and distribution equipment. In the actual operation process, the initial voiceprint data information may include but is not limited to frequency characteristics, amplitude characteristics, harmonic characteristics, environmental parameters such as temperature and humidity, and the trend of sound changes over time. In one embodiment, the initial voiceprint data information may also include the equipment type, equipment number and collection location of the power transmission and distribution equipment.

[0051] In the embodiment, the voiceprint sensing of power transmission and distribution equipment can be performed by voiceprint information collection devices such as passive wireless voiceprint sensing modules to collect initial voiceprint data information of the power transmission and distribution equipment. The abnormal information in the initial voiceprint data information is determined, and the abnormal information is deleted and replaced. Then, the initial voiceprint data information is reduced in dimension by normalization processing to obtain intermediate voiceprint data information, retaining the main information, while reducing the dimension of the data and reducing the complexity of the initial voiceprint data. The intermediate voiceprint data information is filtered by filtering methods such as preset wavelet analysis rules to reduce the noise of the intermediate voiceprint data information and obtain the target voiceprint data information.

[0052] In one embodiment, the initial voiceprint data information is preprocessed to generate target voiceprint data information, including:

[0053] Sort the initial voiceprint data information according to the time when the initial voiceprint data information was obtained;

[0054] Determine the abnormal information in the initial voiceprint data information through a preset linear regression algorithm and delete the abnormal information;

[0055] Determine the initial voiceprint data located on both sides of the abnormal information as the voiceprint data to be calculated, determine the average value of the voiceprint to be calculated as the new initial voiceprint data information and arrange it at the position of the deleted abnormal information;

[0056] Normalizing the rearranged initial voiceprint data information to obtain intermediate voiceprint data information;

[0057] The target voiceprint data information is obtained by filtering the intermediate voiceprint data information through preset wavelet analysis rules.

[0058] The preset linear regression algorithm can be understood as a preset algorithm for determining abnormal information in the initial voiceprint data information, and the abnormal information in the initial voiceprint data information can be determined based on the linear regression equation and the preset threshold. Exemplarily, the calculation formula of the preset linear regression algorithm can be Y i =α+βX i +ε i; Among them, α and β are constants, which are the parameters of the linear regression equation. α is the intercept of the linear regression line on the Y axis; β is the slope of the overall regression line; ε i The random disturbance term is a random variable that cannot be directly observed and can be set according to user needs; i represents the number of items of the i-th initial voiceprint data information. In the actual operation process, after the result value is determined by the linear regression equation, the distance between the result value and the linear regression line can be determined, and the distance is compared with the preset threshold. When it is determined that the distance is greater than the preset threshold, the initial voiceprint data information is determined to be abnormal information. The intermediate voiceprint data information can be understood as the initial voiceprint data information after normalization. The preset wavelet analysis rule is a preset rule for filtering through wavelet transform and wavelet reconstruction.

[0059] In an embodiment, the acquisition time of the initial voiceprint data information can be determined, and the initial voiceprint data information can be sorted according to the acquisition time of the initial voiceprint data information. The result value is determined by the linear regression equation in the preset linear regression algorithm, and the distance from the linear regression line is determined by the result value. When the distance is greater than the preset threshold, the initial voiceprint data information is determined to be abnormal information. The abnormal information is deleted, and the initial voiceprint data located on both sides of the abnormal information are determined as the voiceprint data to be calculated, and the average value of the voiceprint to be calculated is determined as the new initial voiceprint data information and arranged at the position of the deleted abnormal information. In the actual operation process, it is assumed that the data information to be eliminated is X j , then the initial voiceprint data on both sides are X j-1 and X j+1 , then the average value is calculated as follows: in, That is, the new initial voiceprint data information needs to be inserted into the abnormal information position. Then the rearranged initial voiceprint data information is normalized to obtain the intermediate voiceprint data information, and the weight of the intermediate voiceprint data information is mapped between the interval [0, 1]. The intermediate voiceprint data information is filtered by the preset wavelet analysis rule to obtain the target voiceprint data information.

[0060] In one embodiment, the target voiceprint data information is obtained by filtering the intermediate voiceprint data information through a preset wavelet analysis rule, including:

[0061] Determine a preset wavelet basis function, a preset number of decomposition layers, and a preset filtering threshold function in a preset wavelet analysis rule;

[0062] Based on a preset wavelet basis function and a preset number of decomposition layers, the intermediate voiceprint data information is subjected to wavelet decomposition to obtain sub-voiceprint data information, approximate coefficients, and detail coefficients;

[0063] The detail coefficient is threshold processed according to a preset filter threshold function to obtain a target detail coefficient;

[0064] The target voiceprint data information is obtained by performing wavelet reconstruction according to the sub-voiceprint data information, approximate coefficients and target detail coefficients.

[0065] Among them, the preset wavelet basis function can be a preset wavelet basis function for wavelet transform. Exemplarily, the preset wavelet basis function can include but is not limited to Haar wavelet, Daubechies wavelet, and Symlets wavelet. The preset decomposition layer number refers to the preset decomposition layer number of wavelet transform. The preset decomposition layer number can be set according to user needs. Generally speaking, the more the preset decomposition layers are, the more detailed the signal detail analysis is, but the amount of calculation will also increase accordingly. Exemplarily, the preset decomposition layer number can include but is not limited to 2, 3, and 4. The preset filter threshold function can be understood as a preset function for filtering. Exemplarily, the preset filter threshold function can include a hard threshold function and a soft threshold function. The preset wavelet basis function, the preset decomposition layer number, and the preset filter threshold function can be set in advance according to the user's filtering needs. The sub-voiceprint data information refers to the sub-bands of different frequencies of the intermediate voiceprint data information obtained after wavelet decomposition. The approximation coefficient represents the approximation of the intermediate voiceprint data information at a lower resolution, contains the main low-frequency components of the intermediate voiceprint data information, and reflects the general outline and trend of the signal. The detail coefficient represents the detail information of the intermediate voiceprint data information at a higher resolution. It contains the high-frequency components of the intermediate voiceprint data information, reflecting the rapid changes and local characteristics of the signal.

[0066] In an embodiment, a preset wavelet basis function, a preset decomposition layer number and a preset filtering threshold function in a preset wavelet analysis rule can be extracted, and the intermediate voiceprint data information is wavelet decomposed by the preset wavelet basis function and the preset decomposition layer number, and the sub-voiceprint data information, the approximate coefficient and the detail coefficient are determined, and then the detail coefficient threshold is processed according to the hard threshold function or the soft threshold function to obtain the target detail coefficient. Among them, the hard threshold function retains more peak information of the original signal (intermediate voiceprint data information); the soft threshold function shrinks the wavelet decomposition coefficient. Then, according to the sub-voiceprint data information, the approximate coefficient and the target detail coefficient, wavelet reconstruction is performed to obtain the target voiceprint data information. In the actual operation process, the inverse process of wavelet decomposition can be followed, and reconstruction can be performed from the lowest resolution level to the high resolution level step by step, and finally the filtered target voiceprint data information is obtained. In an embodiment, the intermediate voiceprint data information can be decomposed into sub-voiceprint data information according to the preset wavelet basis function and the preset decomposition layer number by the fast wavelet transform algorithm (Mallat algorithm), and the result is then reversely operated by the Mallat algorithm to obtain the target voiceprint data information.

[0067] S220. Determine the vector information of the target voiceprint data information as the target vector information based on the preset cluster analysis data compression coding technology.

[0068] Among them, the data compression coding technology of the preset cluster analysis is a technology that combines cluster analysis and data coding methods to achieve data compression. The vector information of the target voiceprint data information can be determined as the target vector information through the data compression coding technology of the preset cluster analysis.

[0069] S230. Determine data anomaly information of the power transmission, transformation and distribution equipment according to the target vector information.

[0070] S240. Determine attribute information of the data table according to a preset type, and construct the data table according to the attribute information through a preset data model; wherein the attribute information includes a data table structure, a data type, a data mode, and constraints.

[0071] Among them, the data table structure refers to the organization and layout of the data table, which defines the field information contained in the data table and the relationship between the field information. For example, the data table structure may include table name, field and record. Data type refers to the type of data stored in the field in the data table. For example, the data type may include but is not limited to Int64, Float, bool, String, Float Vector, etc. Data mode refers to the organization and storage structure of data in the data table. For example, the data mode may include but is not limited to hierarchical model, network mode, relational mode and object mode. Constraints are rules and conditions used to restrict data in the data table. For example, constraints may include but are not limited to primary key constraints, unique constraints, non-null constraints and foreign key constraints.

[0072] In an embodiment, attribute information of a data table, such as a data table structure, a data type, a data mode, and a constraint condition, contained in a preset type may be extracted.

[0073] S250: Determine the field information of the target vector information and the data exception information associated with the field information, and import the field information and the data exception information into the corresponding data table according to the attribute information of the data table.

[0074] In an embodiment, the field information of the target vector information and the data anomaly information associated with the field information can be extracted, and the field information and the data anomaly information can be imported into the corresponding data table according to the attribute information of the data table. In the actual operation process, the field name of the data table can be determined, the field information of the target vector information and the data anomaly information associated with the field information can be found according to the field name, and the field information of the target vector information and the data anomaly information can be imported into the data table according to the data type; or, the target vector information and the data anomaly information can be stored in the corresponding data table through tools or scripting languages ​​provided by a database management system (DBMS).

[0075] S260. Determine the partition type of the data table according to the amount of data in the data table; wherein the partition type includes time partition and model partition.

[0076] Partitioning by time refers to dividing the data table into different partitions according to the time attribute of the data; partitioning by model refers to dividing the data table into different partitions according to different business models or data models.

[0077] In an embodiment, the data size of a data table can be determined. When the data size of the data stored in the data table is large (greater than or equal to a preset data threshold), the partition type may be partitioning by time; when the data size of the data stored in the data table is large (less than a preset data threshold), the partition type may be partitioning by model.

[0078] S270: Determine the vector index corresponding to each partition type as the target vector index of the data table to obtain a voiceprint vector library.

[0079] In an embodiment, each partition type may have a corresponding vector index, and the vector index corresponding to each partition type may be used as a target vector index of a data table to obtain a voiceprint vector library.

[0080] The embodiment of the present invention collects initial voiceprint data information of each power transmission and distribution equipment, performs data preprocessing on the initial voiceprint data information, generates target voiceprint data information, removes duplicate, incomplete or erroneous data through data cleaning, and performs time-frequency analysis on the voiceprint data signal through wavelet analysis to eliminate deviations, errors and noise in the data, thereby improving the accuracy of the voiceprint data information; determines vector information of the target voiceprint data information as target vector information through a data compression coding technology based on a preset cluster analysis, determines data abnormality information of the power transmission and distribution equipment according to the target vector information, and determines attribute information of the data table according to a preset type, And through the preset data model, a data table is constructed according to the attribute information, the field information of the target vector information and the data anomaly information associated with the field information are determined, the field information and the data anomaly information are imported into the corresponding data table according to the attribute information of the data table, and the partition type of the data table is determined according to the data volume of the data table; wherein the partition types include time partitioning and model partitioning, and the vector index corresponding to each partition type is determined as the target vector index of the data table, and the voiceprint vector library is obtained, and the construction of the vector database is completed, so that the discrete variables of the target voiceprint data information and the device anomaly information are transformed into continuous vectors, thereby improving the efficiency of the target voiceprint data information search.

[0081] Embodiment 3

[0082] Figure 3 is a flow chart of a method for constructing a voiceprint vector library according to Embodiment 3 of the present invention. Figure 3 As shown, this embodiment is a further description of a method for constructing a voiceprint vector library based on the above embodiment. Figure 3 As shown, the method includes:

[0083] S310. Perform voiceprint sensing on the power transmission, transformation and distribution equipment through a passive wireless voiceprint sensing module to collect initial voiceprint data information of the power transmission, transformation and distribution equipment.

[0084] S320: pre-process the collected initial voiceprint data information to obtain target voiceprint data information.

[0085] In an embodiment, the collected voiceprint data information can be cleaned to remove duplicate, incomplete or erroneous data; data correction and alignment can be performed to eliminate deviations, errors or inconsistencies in the initial voiceprint data information; the initial voiceprint data information can be format converted to unify it into a format suitable for vector information.

[0086] In one embodiment, the abnormal information in the initial voiceprint data information can be determined based on the linear regression equation and the preset threshold. For example, the calculation formula of the preset linear regression algorithm can be Y i =α+βX i +εi ; Among them, α and β are constants, which are the parameters of the linear regression equation. α is the intercept of the linear regression line on the Y axis; β is the slope of the overall regression line; ε i The random disturbance term is a random variable that cannot be directly observed and can be set according to user needs; i represents the number of items in the i-th initial voiceprint data information. Assume that the data information to be eliminated is X j , then the initial voiceprint data on both sides are X j-1 and X j+1 , then the average value is calculated as follows: in, That is, the new initial voiceprint data information needs to be inserted into the abnormal information position. Then the rearranged initial voiceprint data information is normalized to obtain the intermediate voiceprint data information, and the weight of the intermediate voiceprint data information is mapped between the interval [0, 1]. The intermediate voiceprint data information is filtered by the interpolation zero-filling Mallat algorithm to obtain the target voiceprint data information.

[0087] S330, vectorizing the target voiceprint data information to obtain target vector information.

[0088] In an embodiment, the target voiceprint data information can be vectorized by manual drawing, automated vectorization algorithm or point cloud processing, and converted into vector elements, such as points, lines, and surfaces, to ensure that the vectorization result is accurate and complete, and retains key attribute information.

[0089] S340, constructing a voiceprint vector library based on the target vector information.

[0090] In an embodiment, an appropriate preset data model, such as a vector model, a raster model, or an object-oriented data model, can be selected according to the data type and requirements of the target vector information, and the data mode is determined, the data table, field, and constraint are defined, and the data table and the target vector index of the data table are established to support subsequent data storage and query.

[0091] In one embodiment, the data table may include an original sample data table, a model information data table, a label sample data table, an original data detection record data table, a noise anomaly detection record data table, an abnormal discharge detection record data table, and a mechanical fault detection record data table.

[0092] For example, the table name of the original sample data table can be set to tx_sw_ys_ybk, partitioned by time, and the target vector index adopts IVF_FLAT index; the field attributes are shown in Table 1:

[0093] Table 1

[0094]

[0095]

[0096] The table name of the model information data table can be set to tx_sw_mx_zyk, partitioned by model; the target vector index uses Flat index; the field attributes are shown in Table 2:

[0097] Table 2

[0098]

[0099]

[0100] The table name of the label sample data table can be set to tx_sw_bq_ybk, partitioned by model; the target vector index uses Flat index; the field attributes are shown in Table 3:

[0101] Table 3

[0102] Fields type Dimensions illustrate id Int64 - Sample label primary key id tag String - Label modelCode String - Model Number confidence Float - Confidence status Int8 - Sample label status tagVector FloatVector 128 Label vector data desc String - Label vector description source Int8 - source time Int64 - Label vector data timestamp createTime Int64 - Timestamp of label vector data entry

[0103] The table name of the raw data detection record data table can be set to tx_sw_ys_ybk, partitioned by time; the target vector index uses IVF_FLAT index; the field attributes are shown in Table 4:

[0104] Table 4

[0105]

[0106]

[0107] The table name of the noise anomaly detection record data table can be set to tx_sw_ys_ybk, partitioned by time; the target vector index uses IVF_FLAT index; the field attributes are shown in Table 5:

[0108] Table 5

[0109]

[0110]

[0111] The table name of the abnormal discharge detection record data table can be set to tx_sw_jc_ycfd_jlk, which is partitioned by time; the target vector index uses IVF_FLAT index; the field attributes are shown in Table 6:

[0112] Table 6

[0113]

[0114]

[0115] The table name of the mechanical fault detection record data table can be set to tx_sw_jc_jxgz_jlk, partitioned by time; the target vector index uses IVF_FLAT index; the field attributes are shown in Table 7:

[0116] Table 7

[0117] Fields type Dimensions illustrate id Int64 - Test record primary key id dataId Int64 - Voiceprint data id dataVector FloatVector 128 Fault detection vector data sampleId int64 - Sample id similarity Float - Sample matching similarity health Float - Health exceptionType String - Exception Type diagnosticSuggestion String - Diagnostic recommendations isValid bool - Is it effective? desc String - Test result description createTime Int64 - Test result data storage timestamp

[0118] In one embodiment, the voiceprint vector library can adopt Milvus. After the voiceprint vector library is constructed, the import tool or script provided by the database management system (DBMS) can be used to design a suitable data management strategy, including data backup, recovery, and version control, to ensure data integrity and consistency, and to perform data verification and repair regularly.

[0119] The query language and spatial analysis functions provided by DBMS can be used for data query and analysis to perform data query, filtering and spatial operations; design and execute appropriate spatial queries, buffer analysis, and spatial relationship operations; optimize query performance, use indexes, spatial indexes, and partitioning technologies to improve query efficiency, and visualize the voiceprint vector library.

[0120] Embodiment 4

[0121] Figure 4 Schematic diagram of a device for constructing a voiceprint vector library according to Embodiment 4 of the present invention. Figure 4 As shown, the device includes: an information determination module 41, a data table determination module 42 and a vector library construction module 43.

[0122] The information determination module 41 is used to determine the target voiceprint data information of the power transmission, transformation and distribution equipment, determine the target vector information according to the target voiceprint data information, and determine the data abnormality information of the power transmission, transformation and distribution equipment according to the target vector information;

[0123] A data table determination module 42 is used to construct a data table according to a preset type based on a preset data model, and store target vector information and data anomaly information in a corresponding data table;

[0124] The vector library construction module 43 is used to determine the partition type of the data table, and create a target vector index of each data table according to the partition type to obtain a voiceprint vector library.

[0125] In the embodiment of the present invention, the target voiceprint data information of the power transmission and distribution equipment is determined by the information determination module, the target vector information is determined according to the target voiceprint data information, and the data anomaly information of the power transmission and distribution equipment is determined according to the target vector information, so that the discrete variables of the target voiceprint data information and the equipment anomaly information are converted into continuous vectors; the data table determination module constructs a data table according to a preset type based on a preset data model, and stores the target vector information and the data anomaly information in the corresponding data table; the vector library construction module determines the partition type of the data table, and creates a target vector index for each data table according to the partition type, obtains a voiceprint vector library, and realizes the rapid establishment of the voiceprint vector library, avoids the problem that the traditional relational database is inconvenient to store and search for voiceprint data information, and improves the efficiency and accuracy of the search for the target voiceprint data information.

[0126] In one embodiment, the information determination module 41 includes:

[0127] A data acquisition unit is used to collect initial voiceprint data information of each power transmission, transformation and distribution equipment, perform data preprocessing on the initial voiceprint data information, and generate target voiceprint data information;

[0128] The information determination unit is used to determine the vector information of the target voiceprint data information as the target vector information based on the data compression coding technology of the preset cluster analysis.

[0129] In one embodiment, the data acquisition unit is specifically used for:

[0130] Sort the initial voiceprint data information according to the time when the initial voiceprint data information was obtained;

[0131] Determine the abnormal information in the initial voiceprint data information through a preset linear regression algorithm and delete the abnormal information;

[0132] Determine the initial voiceprint data located on both sides of the abnormal information as the voiceprint data to be calculated, determine the average value of the voiceprint to be calculated as the new initial voiceprint data information and arrange it at the position of the deleted abnormal information;

[0133] Normalizing the rearranged initial voiceprint data information to obtain intermediate voiceprint data information;

[0134] The target voiceprint data information is obtained by filtering the intermediate voiceprint data information through preset wavelet analysis rules.

[0135] In one embodiment, the data acquisition unit is further used for:

[0136] Determine a preset wavelet basis function, a preset number of decomposition layers, and a preset filtering threshold function in a preset wavelet analysis rule;

[0137] Based on a preset wavelet basis function and a preset number of decomposition layers, the intermediate voiceprint data information is subjected to wavelet decomposition to obtain sub-voiceprint data information, approximate coefficients, and detail coefficients;

[0138] The detail coefficient is threshold processed according to a preset filter threshold function to obtain a target detail coefficient;

[0139] The target voiceprint data information is obtained by performing wavelet reconstruction according to the sub-voiceprint data information, approximate coefficients and target detail coefficients.

[0140] In one embodiment, the data table determination module 42 includes:

[0141] A data table construction unit, used to determine the attribute information of the data table according to the preset type, and construct the data table according to the attribute information through the preset data model; wherein the attribute information includes the data table structure, data type, data mode and constraint conditions;

[0142] The data storage unit is used to determine the field information of the target vector information and the data anomaly information associated with the field information, and import the field information and the data anomaly information into the corresponding data table according to the attribute information of the data table.

[0143] In one embodiment, the vector library construction module 43 includes:

[0144] A partition determination unit, used to determine the partition type of the data table according to the data volume of the data table; wherein the partition types include time partition and model partition;

[0145] The index determination unit is used to determine the vector index corresponding to each partition type as the target vector index of the data table to obtain the voiceprint vector library.

[0146] In one embodiment, the voiceprint vector library construction device further includes:

[0147] The model data table construction module is used to determine the model parameters of the preset data model and construct a model information data table associated with the model parameters in the voiceprint vector library.

[0148] The voiceprint vector library construction device provided in the embodiment of the present invention can execute the voiceprint vector library construction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0149] Embodiment 5

[0150] Figure 5: is a schematic diagram of the structure of an electronic device that implements the voiceprint vector library construction method of an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0151] like Figure 5 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0152] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0153] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a voiceprint vector library construction method.

[0154] In some embodiments, the voiceprint vector library construction method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the voiceprint vector library construction method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the voiceprint vector library construction method in any other appropriate manner (for example, by means of firmware).

[0155] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0156] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0157] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0158] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0159] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0160] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0161] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0162] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for constructing a voiceprint vector library, It is characterized in that include: Determine target voiceprint data information of the power transmission, transformation and distribution equipment, determine target vector information according to the target voiceprint data information, and determine data anomaly information of the power transmission, transformation and distribution equipment according to the target vector information; Building a data table according to a preset type based on a preset data model, and storing the target vector information and the data anomaly information in a corresponding data table; The partition type of the data table is determined, and a target vector index of each data table is created according to the partition type to obtain a voiceprint vector library.

2. The method according to claim 1, It is characterized in that The step of determining target voiceprint data information of the power transmission, transformation and distribution equipment and determining target vector information according to the target voiceprint data information includes: Collecting initial voiceprint data information of each power transmission, transformation and distribution equipment, performing data preprocessing on the initial voiceprint data information, and generating target voiceprint data information; The data compression coding technology based on the preset cluster analysis determines the vector information of the target voiceprint data information as the target vector information.

3. The method according to claim 2, It is characterized in that The preprocessing of the initial voiceprint data to generate target voiceprint data includes: sorting the initial voiceprint data information according to the acquisition time of the initial voiceprint data information; Determine abnormal information in the initial voiceprint data information by using a preset linear regression algorithm, and delete the abnormal information; Determine the initial voiceprint data located on both sides of the abnormal information as the voiceprint data to be calculated, determine the average value of the voiceprints to be calculated as new initial voiceprint data information and arrange them at the position of the deleted abnormal information; Normalizing the rearranged initial voiceprint data information to obtain intermediate voiceprint data information; The target voiceprint data information is obtained by filtering the intermediate voiceprint data information using a preset wavelet analysis rule.

4. The method according to claim 3, It is characterized in that The filtering process of the intermediate voiceprint data information by using a preset wavelet analysis rule to obtain the target voiceprint data information includes: Determine a preset wavelet basis function, a preset number of decomposition layers, and a preset filtering threshold function in a preset wavelet analysis rule; Based on the preset wavelet basis function and the preset decomposition level, the intermediate voiceprint data information is subjected to wavelet decomposition to obtain sub-voiceprint data information, approximate coefficients and detail coefficients; Processing the detail coefficient threshold according to a preset filtering threshold function to obtain a target detail coefficient; The target voiceprint data information is obtained by performing wavelet reconstruction according to the sub-voiceprint data information, approximate coefficients and target detail coefficients.

5. The method according to claim 1, It is characterized in that The step of constructing a data table according to a preset type based on a preset data model, and storing the target vector information and the data anomaly information in a corresponding data table includes: Determine the attribute information of the data table according to the preset type, and construct the data table according to the attribute information through the preset data model; wherein the attribute information includes the data table structure, data type, data mode and constraint conditions; The field information of the target vector information and the data exception information associated with the field information are determined, and the field information and the data exception information are imported into the corresponding data table according to the attribute information of the data table.

6. The method according to claim 1, It is characterized in that The determining the partition type of the data table, and creating the target vector index of each data table according to the partition type to obtain the voiceprint vector library includes: Determining the partition type of the data table according to the amount of data in the data table; wherein the partition type includes time partition and model partition; The vector index corresponding to each partition type is determined as the target vector index of the data table to obtain a voiceprint vector library.

7. The method according to claim 1, It is characterized in that The method further comprises: Determine the model parameters of the preset data model, and construct a model information data table associated with the model parameters in the voiceprint vector library.

8. A device for constructing a voiceprint vector library, It is characterized in that include: An information determination module, used to determine target voiceprint data information of power transmission, transformation and distribution equipment, determine target vector information according to the target voiceprint data information, and determine data anomaly information of the power transmission, transformation and distribution equipment according to the target vector information; A data table determination module, used to construct a data table according to a preset type based on a preset data model, and store the target vector information and the data anomaly information in a corresponding data table; The vector library construction module is used to determine the partition type of the data table, and create a target vector index of each data table according to the partition type to obtain a voiceprint vector library.

9. An electronic device, It is characterized in that The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the voiceprint vector library construction method described in any one of claims 1-7.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the voiceprint vector library construction method according to any one of claims 1 to 7 when executed.

Images