Flexible DC support capacitor quality analysis method and system based on artificial intelligence

Abstract

The flexible DC support capacitor quality analysis method and system based on artificial intelligence provided by the present invention relate to the field of artificial intelligence technology. In the present invention, first, the target capacitor data corresponding to the flexible DC support capacitor is extracted, and the target capacitor data is semantically mined to output the target capacitor semantic vector; secondly, multiple associated capacitor data corresponding to the target capacitor data are determined based on the target capacitor semantic vector; then, based on the associated capacitor semantic vector and the corresponding capacitor quality data corresponding to each associated capacitor data, and in combination with the target capacitor semantic vector, a global capacitor semantic vector is determined; finally, based on the global capacitor semantic vector, the capacitor quality data corresponding to the target capacitor data is analyzed. Based on the above content, the problem of relatively low reliability of capacitor quality analysis existing in the prior art can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based flexible DC support capacitor quality analysis method and system. Background Art

[0002] In the power system, Flexible DC Transmission (FDC) system has become an important technology for power transmission. Among them, Flexible DC Capacitors play a vital role in these systems, especially in stabilizing voltage, improving power quality, and supporting system scheduling and control. Based on this, in order to ensure the stability and reliability of these capacitors in the system, effective quality analysis is essential. However, in the prior art, the operating data of the capacitor, such as current, voltage, power, and frequency, are generally detected, and then the detection results are compared with the threshold parameters set according to experience to obtain the quality of the capacitor. In this way, the reliability of the quality analysis is relatively low. Summary of the invention

[0003] In view of this, an object of the present invention is to provide a flexible DC support capacitor quality analysis method and system based on artificial intelligence to improve the problem of relatively low reliability of capacitor quality analysis in the prior art.

[0004] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0005] A quality analysis method of a flexible DC support capacitor based on artificial intelligence, comprising:

[0006] Extracting target capacitor data corresponding to the flexible DC support capacitor, and performing semantic mining on the target capacitor data to output a semantic vector of the target capacitor;

[0007] Determine a plurality of associated capacitor data corresponding to the target capacitor data based on the target capacitor semantic vector, wherein each of the associated capacitor data has capacitor quality data;

[0008] Determine a global capacitor semantic vector based on the associated capacitor semantic vector corresponding to each of the associated capacitor data and the capacitor quality data corresponding to each of the associated capacitor data, and in combination with the target capacitor semantic vector;

[0009] Based on the global capacitor semantic vector, capacitor quality data corresponding to the target capacitor data is analyzed, wherein the capacitor quality data corresponding to the target capacitor data is used to reflect the quality of the flexible DC support capacitor.

[0010] In some preferred embodiments, in the above-mentioned flexible DC support capacitor quality analysis method based on artificial intelligence, the target capacitor semantic vector includes a target capacitor test vector corresponding to the target capacitor test data in the target capacitor data and a target capacitor production vector corresponding to the target capacitor production data, and the associated capacitor semantic vector includes an associated capacitor test vector corresponding to the associated capacitor test data in the associated capacitor data and an associated capacitor production vector corresponding to the associated capacitor production data;

[0011] The test data is used to characterize the test process of the corresponding flexible DC support capacitor, and the production data is used to characterize the production process of the corresponding flexible DC support capacitor; the step of determining the corresponding global capacitor semantic vector based on the associated capacitor semantic vector corresponding to each of the associated capacitor data and the capacitor quality data corresponding to each of the associated capacitor data, and in combination with the target capacitor semantic vector, includes:

[0012] Performing a vector space mapping operation on the target capacitor test vector and outputting a corresponding target test mapping vector, and respectively performing a vector space mapping operation on the associated capacitor test vector corresponding to each of the associated capacitor data and outputting an associated test mapping vector corresponding to each of the associated capacitor data;

[0013] Performing a vector space mapping operation on the target capacitor production vector and outputting a corresponding target production mapping vector, and performing a vector space mapping operation on the associated capacitor production vector corresponding to each of the associated capacitor data and outputting an associated production mapping vector corresponding to each of the associated capacitor data;

[0014] Performing semantic mining on capacitor quality data corresponding to each of the associated capacitor data, and outputting a capacitor quality semantic vector corresponding to each of the associated capacitor data;

[0015] Based on the target test mapping vector, the associated test mapping vector, the target production mapping vector, the associated production mapping vector, and the capacitor quality semantic vector, a corresponding global capacitor semantic vector is determined.

[0016] In some preferred embodiments, in the above-mentioned flexible DC support capacitor quality analysis method based on artificial intelligence, the step of determining the corresponding global capacitor semantic vector based on the target test mapping vector, the associated test mapping vector, the target production mapping vector, the associated production mapping vector and the capacitor quality semantic vector includes:

[0017] For each of the associated capacitor data, based on the capacitor quality semantic vector corresponding to the associated capacitor data, focus mining is performed on the associated test mapping vector and the associated production mapping vector corresponding to the associated capacitor data, and a first test focus vector and a first production focus vector corresponding to the associated capacitor data are output;

[0018] For each of the associated capacitor data, based on the first test focus vector corresponding to the associated capacitor data, the target test mapping vector is focused and mined, and a second test focus vector corresponding to the associated capacitor data is output; and based on the first production focus vector corresponding to the associated capacitor data, the target production mapping vector is focused and mined, and a second production focus vector corresponding to the associated capacitor data is output;

[0019] For each of the associated capacitor data, based on the second production focusing vector corresponding to the associated capacitor data, focus mining is performed on the second test focusing vector corresponding to the associated capacitor data, and the capacitor focusing vector corresponding to the associated capacitor data is output;

[0020] Based on the capacitor focus vector corresponding to each of the associated capacitor data, a corresponding global capacitor semantic vector is determined.

[0021] In some preferred embodiments, in the above-mentioned flexible DC support capacitor quality analysis method based on artificial intelligence, for each of the associated capacitor data, based on the capacitor quality semantic vector corresponding to the associated capacitor data, the associated test mapping vector and the corresponding associated production mapping vector corresponding to the associated capacitor data are respectively focused and mined, and the first test focus vector and the first production focus vector corresponding to the associated capacitor data are output, including:

[0022] Linearly mapping the capacitor quality semantic vector corresponding to the associated capacitor data to obtain a capacitor quality linear vector, and performing two different linear mappings on the associated test mapping vector corresponding to the associated capacitor data to obtain a first associated test linear vector and a second associated test linear vector, and performing two different linear mappings on the associated production mapping vector corresponding to the associated capacitor data to obtain a first associated production linear vector and a second associated production linear vector;

[0023] Based on the correlation parameter between the capacitor quality linear vector and the first associated test linear vector, the second associated test linear vector is adjusted to obtain a first test focus vector corresponding to the associated capacitor data;

[0024] Based on the correlation parameter between the capacitor quality linear vector and the first associated production linear vector, the second associated production linear vector is adjusted to obtain a first production focus vector corresponding to the associated capacitor data.

[0025] In some preferred embodiments, in the above-mentioned flexible DC support capacitor quality analysis method based on artificial intelligence, for each of the associated capacitor data, based on the first test focus vector corresponding to the associated capacitor data, the target test mapping vector is focused and mined, and the second test focus vector corresponding to the associated capacitor data is output, and the target production mapping vector is focused and mined based on the first production focus vector corresponding to the associated capacitor data, and the step of outputting the second production focus vector corresponding to the associated capacitor data includes:

[0026] Performing a difference calculation on a first test focus vector corresponding to the associated capacitor data and the target test mapping vector to obtain a corresponding difference test vector, and generating a plurality of perturbation test vectors based on the difference test vector; and performing a difference calculation on a first production focus vector corresponding to the associated capacitor data and the target production mapping vector to obtain a corresponding difference production vector, and generating a plurality of perturbation production vectors based on the difference production vector;

[0027] In the process of focusing and mining the target test mapping vector based on the first test focusing vector corresponding to the associated capacitor data, applying the plurality of perturbation test vectors and outputting a second test focusing vector corresponding to the associated capacitor data;

[0028] In the process of focusing and mining the target production mapping vector based on the first production focusing vector corresponding to the associated capacitor data, the multiple perturbation production vectors are applied to output a second production focusing vector corresponding to the associated capacitor data.

[0029] In some preferred embodiments, in the above-mentioned flexible DC support capacitor quality analysis method based on artificial intelligence, the step of applying the multiple disturbance test vectors and outputting the second test focus vector corresponding to the associated capacitor data in the process of focusing and mining the target test mapping vector based on the first test focus vector corresponding to the associated capacitor data includes:

[0030] Superimposing each of the plurality of perturbation test vectors and the first test focus vector corresponding to the associated capacitor data respectively, outputting a plurality of corresponding perturbation test focus vectors, and serializing the plurality of perturbation test focus vectors to form a target sequence;

[0031] For the first perturbation test focusing vector in the target sequence, based on the perturbation test focusing vector, focusing mining is performed on the target test mapping vector, and a focusing mining result corresponding to the perturbation test focusing vector is output;

[0032] For each perturbation test focused vector other than the first perturbation test focused vector in the target sequence, based on the perturbation test focused vector, focused mining is performed on a focused mining result corresponding to a previous perturbation test focused vector of the perturbation test focused vector, and the focused mining result corresponding to the perturbation test focused vector is output;

[0033] A second test focus vector corresponding to the associated capacitor data is determined based at least on a focus mining result corresponding to a last perturbation test focus vector in the target sequence.

[0034] In some preferred embodiments, in the above-mentioned flexible DC support capacitor quality analysis method based on artificial intelligence, the step of analyzing the capacitor quality data corresponding to the target capacitor data based on the global capacitor semantic vector includes:

[0035] Performing semantic space conversion processing on the global capacitor semantic vector, and outputting a converted capacitor semantic vector corresponding to the global capacitor semantic vector, wherein the converted capacitor semantic vector is a semantic vector that can be processed by a capacitor quality analysis network;

[0036] The conversion capacitor semantic vector is loaded so that the capacitor quality analysis network acquires the conversion capacitor semantic vector, wherein the capacitor quality analysis network updates the network parameters of the second part of the candidate capacitor quality analysis network based on the training conversion capacitor semantic vector while maintaining the network parameters of the first part of the candidate capacitor quality analysis network, and the network parameters of the second part refer to the network parameters of the semantic adaptation unit in the candidate capacitor quality analysis network;

[0037] The converted capacitor semantic vector is analyzed through the capacitor quality analysis network to obtain capacitor quality data corresponding to the target capacitor data.

[0038] In some preferred embodiments, in the above-mentioned flexible DC support capacitor quality analysis method based on artificial intelligence, the step of updating and forming the capacitor quality analysis network includes:

[0039] Determine a first capacitor test vector, a first capacitor production vector, and a first capacitor actual quality corresponding to the first capacitor data, and determine a second capacitor test vector, a second capacitor production vector, and a second capacitor actual quality corresponding to the second capacitor data;

[0040] Using a global semantic aggregation unit in the candidate capacitor quality analysis network, globally semantically aggregate the first capacitor test vector, the first capacitor production vector, the second capacitor test vector, the second capacitor production vector, and the second capacitor actual quality to form a corresponding training global capacitor semantic vector;

[0041] Performing semantic space conversion processing on the training global capacitor semantic vector, and outputting a training conversion capacitor semantic vector corresponding to the training global capacitor semantic vector;

[0042] The semantic adaptation unit and the candidate capacitor quality analysis network are merged, so that the network parameters of the second part are merged in the candidate capacitor quality analysis network based on the semantic adaptation unit, wherein the network parameters of the first part refer to the network parameters of the global semantic aggregation unit;

[0043] The network parameters of the first part of the candidate capacitor quality analysis network are maintained, and the training conversion capacitor semantic vector is analyzed by the semantic adaptation unit to obtain a first capacitor analysis quality, and based on the error between the first capacitor analysis quality and the actual quality of the first capacitor, the network parameters of the second part of the candidate capacitor quality analysis network are updated to form a capacitor quality analysis network.

[0044] In some preferred embodiments, in the above-mentioned flexible DC support capacitor quality analysis method based on artificial intelligence, the target capacitor semantic vector includes a target capacitor test vector corresponding to the target capacitor test data in the target capacitor data and a target capacitor production vector corresponding to the target capacitor production data;

[0045] The step of determining a plurality of associated capacitor data corresponding to the target capacitor data based on the target capacitor semantic vector includes:

[0046] Determine a pre-built test vector set and a production vector set, wherein the test vector set includes each pending capacitor data and a test vector corresponding to the pending capacitor data, and the production vector set includes each pending capacitor data and a production vector corresponding to the pending capacitor data;

[0047] Based on the target capacitor test vector, performing a comparative analysis in the test vector set and the production vector set, and based on the target capacitor production vector, performing a comparative analysis in the test vector set and the production vector set to determine a plurality of pending capacitor data matching the target capacitor data;

[0048] Based on a test data matching parameter between a plurality of candidate capacitor data matching the target capacitor data and the target capacitor data, a plurality of associated capacitor data corresponding to the target capacitor data is determined from among the plurality of candidate capacitor data matching the target capacitor data.

[0049] An embodiment of the present invention also provides an artificial intelligence-based flexible DC support capacitor quality analysis system, comprising a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to implement the above-mentioned artificial intelligence-based flexible DC support capacitor quality analysis method.

[0050] The flexible DC support capacitor quality analysis method and system based on artificial intelligence provided by the embodiment of the present invention firstly extracts the target capacitor data corresponding to the flexible DC support capacitor, and performs semantic mining on the target capacitor data to output the target capacitor semantic vector; secondly, determines multiple associated capacitor data corresponding to the target capacitor data based on the target capacitor semantic vector; then, determines the global capacitor semantic vector based on the associated capacitor semantic vector and the corresponding capacitor quality data corresponding to each associated capacitor data, and combines the target capacitor semantic vector; finally, analyzes the capacitor quality data corresponding to the target capacitor data based on the global capacitor semantic vector. Based on the above content, since the global capacitor semantic vector is formed based on the associated capacitor semantic vector and the corresponding capacitor quality data corresponding to each associated capacitor data, and combines the target capacitor semantic vector, the global capacitor semantic vector not only carries the relevant semantic information of the flexible DC support capacitor, but also carries the semantic information of the associated associated capacitor data and the capacitor quality data, so that the semantic representation ability of the global capacitor semantic vector is better. Therefore, when performing quality analysis based on the global capacitor semantic vector, it can have a higher reliability, thereby improving the problem of relatively low reliability of capacitor quality analysis existing in the prior art.

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A structural block diagram of an artificial intelligence-based flexible DC support capacitor quality analysis system provided in an embodiment of the present invention.

[0053] Figure 2 A schematic flow chart of the steps included in the flexible DC support capacitor quality analysis method based on artificial intelligence provided in an embodiment of the present invention.

[0054] Figure 3A first schematic diagram of forming a second test focus vector based on focused mining of a first test focus vector provided by an embodiment of the present invention.

[0055] Figure 4 A second schematic diagram of forming a second test focus vector by focusing and mining based on a first test focus vector provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions 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. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0057] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] like Figure 1 As shown, an embodiment of the present invention provides a flexible DC support capacitor quality analysis system based on artificial intelligence. The flexible DC support capacitor quality analysis system based on artificial intelligence may include a memory and a processor.

[0059] In detail, the memory and the processor are electrically connected directly or indirectly to achieve data transmission or interaction. For example, they can be electrically connected to each other through one or more communication buses or signal lines. The memory may store at least one software function module (computer program) that can exist in the form of software or firmware. The processor can be used to execute the executable computer program stored in the memory, thereby realizing the flexible DC support capacitor quality analysis method based on artificial intelligence provided by an embodiment of the present invention.

[0060] Optionally, the memory may be, but not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc. The processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0061] and, Figure 1 The structure shown is for illustration only. The flexible DC support capacitor quality analysis system based on artificial intelligence may also include a comparison Figure 1 More or fewer components as shown, or with Figure 1 The various configurations shown, for example, include a communication unit for interacting with other devices.

[0062] Among them, in an alternative example, the flexible DC support capacitor quality analysis system based on artificial intelligence can be a server with data processing capabilities.

[0063] Combination Figure 2 The embodiment of the present invention also provides a flexible DC support capacitor quality analysis method based on artificial intelligence, which can be applied to the flexible DC support capacitor quality analysis system based on artificial intelligence. The method steps defined in the process related to the flexible DC support capacitor quality analysis method based on artificial intelligence can be implemented by the flexible DC support capacitor quality analysis system based on artificial intelligence (hereinafter referred to as capacitor quality analysis system).

[0064] The following will Figure 2 The specific process shown is explained in detail.

[0065] Step S110 , extracting target capacitor data corresponding to the flexible DC support capacitor, and performing semantic mining on the target capacitor data to output a semantic vector of the target capacitor.

[0066] In an embodiment of the present invention, the capacitor quality analysis system can extract the target capacitor data corresponding to the flexible DC support capacitor (i.e., the flexible DC support capacitor to be quality analyzed), and perform semantic mining on the target capacitor data to output the target capacitor semantic vector. Exemplarily, semantic mining can refer to vectorizing the target capacitor data, i.e., mapping it into a vector space to obtain a corresponding semantic vector for subsequent processing. For example, when the target capacitor data belongs to image data (such as a monitoring image of a flexible DC support capacitor during production), the image data can be convolved by a convolution unit, and the convolution vector obtained can be used as the corresponding semantic vector. When the target capacitor data belongs to text data (such as a description text of a performance test of a flexible DC support capacitor), the text data can be embedded by a word embedding model, and the obtained embedding vector can be used as the corresponding semantic vector. For example, the target capacitor data belongs to the text describing the test "the capacitor is 100µF, and the impedance is 159 ohms at 1kHz". The text is segmented and embedded, and the following word embedding vector can be obtained:

[0067] Capacitors: [0.25, -0.1, 0.02, 0.3, -0.4, 0.1, 0.5, -0.3, 0.4, -0.2, 0.15,0.3, 0.05, 0.7, -0.6, 0.1,......];

[0068] is: [0.1, -0.2, 0.3, 0.05, -0.1, 0.15, -0.3, 0.25, -0.1, 0.1, 0.2, -0.3, 0.05, 0.3, -0.4, 0.1,......];

[0069] 100: [0.5, -0.1, 0.1, -0.2, 0.15, 0.1, 0.3, 0.2, -0.4, 0.1, 0.05, -0.2, 0.1, 0.3, -0.15, 0.2, ...];

[0070] µF: [0.3, -0.2, 0.1, 0.4, -0.3, 0.2, 0.05, 0.1, -0.15, 0.3, -0.25,0.2, 0.1, 0.05, 0.1, -0.4,...];

[0071] in: [0.7, -0.8, 0.1, 0.3, -0.2, 0.1, 0.02, 0.3, -0.16, 0.6, -0.25,0.23, 0.11, 0.04, 0.1, -0.8,......];

[0072] 1kHz: [0.4, 0.2, -0.1, 0.3, -0.3, 0.15, 0.05, -0.25, 0.15, -0.1, 0.1,0.2, -0.3, 0.25, 0.2, 0.05,...];

[0073] When: [0.4, -0.4, 0.2, 0.3, -0.1, 0.4, 0.06, 0.8, -0.11, 0.7, -0.25,0.24, 0.11, 0.03, 0.2, -0.9,......];

[0074] Impedance: [0.6, 0.4, 0.3, -0.2, 0.15, 0.5, 0.25, -0.1, 0.1, 0.3, 0.4, -0.3, 0.2, -0.15, 0.1, 0.35,......];

[0075] 159: [0.25, -0.15, 0.2, 0.1, 0.4, -0.25, 0.35, 0.2, -0.1, -0.2, 0.15,0.25, 0.3, -0.05, 0.1, -0.4,......];

[0076] Ohm: [0.05, 0.15, -0.1, 0.3, 0.2, -0.05, 0.2, 0.25, -0.3, 0.1, 0.05,0.15, -0.25, 0.1, 0.2, -0.3,......].

[0077] Step S120: determining a plurality of associated capacitor data corresponding to the target capacitor data based on the target capacitor semantic vector.

[0078] In an embodiment of the present invention, after obtaining the target capacitor semantic vector, the capacitor quality analysis system can determine a plurality of associated capacitor data corresponding to the target capacitor data (i.e., capacitor data having an associated relationship or high correlation with the target capacitor semantic vector) based on the target capacitor semantic vector. Each of the associated capacitor data has capacitor quality data (based on actual quality obtained by evaluation in other ways, such as expert analysis, etc.).

[0079] Step S130: determining a global capacitor semantic vector based on the associated capacitor semantic vector corresponding to each of the associated capacitor data and the capacitor quality data corresponding to each of the associated capacitor data, and in combination with the target capacitor semantic vector.

[0080] In an embodiment of the present invention, after determining the associated capacitor data corresponding to the target capacitor semantic vector, the capacitor quality analysis system can determine a global capacitor semantic vector based on the associated capacitor semantic vector corresponding to each of the associated capacitor data and the capacitor quality data corresponding to each of the associated capacitor data, and in combination with the target capacitor semantic vector. Based on this, the global capacitor semantic vector can carry semantic information of three dimensions: the target capacitor semantic vector, the associated capacitor semantic vector, and the capacitor quality data.

[0081] Step S140: Analyze capacitor quality data corresponding to the target capacitor data based on the global capacitor semantic vector.

[0082] In an embodiment of the present invention, after determining the global capacitor semantic vector, the capacitor quality analysis system can analyze the capacitor quality data corresponding to the target capacitor data based on the global capacitor semantic vector. The capacitor quality data corresponding to the target capacitor data is used to reflect the quality of the flexible DC support capacitor. Exemplarily, the global capacitor semantic vector can be fully connected to obtain the corresponding fully connected capacitor semantic vector, and then the corresponding quality score can be output through a linear regression function, such as through a Sigmoid function to output a quality score of 0-1, where the higher the value, the better the quality. As mentioned above, when the target capacitor data includes high-frequency response (impedance) data, the capacitor quality data is used to reflect the high-frequency response quality. When the target capacitor data includes other data, it can be used to reflect other qualities; or, when multiple data are included, it can reflect the comprehensive quality of multiple dimensions.

[0083] Based on the above content, since the global capacitor semantic vector is formed based on the associated capacitor semantic vector and the corresponding capacitor quality data corresponding to each associated capacitor data, and combined with the target capacitor semantic vector, the global capacitor semantic vector not only carries the relevant semantic information of the flexible DC support capacitor, but also carries the semantic information of the associated associated capacitor data and capacitor quality data, so that the semantic representation ability of the global capacitor semantic vector is better. Therefore, when performing quality analysis based on the global capacitor semantic vector, it can have a higher reliability, thereby improving the problem of relatively low reliability of capacitor quality analysis in the prior art.

[0084] In the above step S120, the specific method of determining the plurality of associated capacitor data is not limited and can be selected according to actual needs.

[0085] For example, in a specific implementation, in order to ensure that the determined multiple associated capacitor data have high reliability, the semantic information of the target capacitor data can be assisted, and when the target capacitor semantic vector includes a target capacitor test vector corresponding to the target capacitor test data in the target capacitor data and a target capacitor production vector corresponding to the target capacitor production data, the above step S120 may include the following contents:

[0086] First, a pre-constructed test vector set and a production vector set can be determined, wherein the test vector set includes each pending capacitor data and a test vector corresponding to the pending capacitor data (such as obtained by performing word embedding processing on the test data in the pending capacitor data), and the production vector set includes each pending capacitor data and a production vector corresponding to the pending capacitor data (such as obtained by performing convolution processing on the production data in the pending capacitor data); in addition, each pending capacitor data has corresponding capacitor quality data;

[0087] Secondly, based on the target capacitor test vector, a comparative analysis can be performed in the test vector set and the production vector set, and based on the target capacitor production vector, a comparative analysis can be performed in the test vector set and the production vector set to determine a plurality of pending capacitor data that matches the target capacitor data; illustratively, each pending capacitor data whose vector cosine distance between the test vector and the target capacitor test vector is less than a first threshold, and whose vector cosine distance between the production vector and the target capacitor production vector is less than a second threshold can be screened out, and the first threshold and the second threshold can be the same or different, and can be selected accordingly according to actual needs;

[0088] Then, based on the test data matching parameters between the multiple pending capacitor data matching the target capacitor data and the target capacitor data (such as the larger the edit distance between the texts, the smaller the test data matching parameter), multiple associated capacitor data corresponding to the target capacitor data can be determined from the multiple pending capacitor data matching the target capacitor data, such as each associated capacitor data having a test data matching parameter greater than a third threshold or a specified number of associated capacitor data having the largest test data matching parameter.

[0089] In the above step S130, the specific method of determining the corresponding global capacitor semantic vector is not limited and can be selected according to actual needs.

[0090] For example, in a specific implementation, in order to ensure that the determined global capacitor semantic vector has better semantic representation ability, the target capacitor semantic vector includes the target capacitor test vector corresponding to the target capacitor test data in the target capacitor data and the target capacitor production vector corresponding to the target capacitor production data, and the associated capacitor semantic vector includes the associated capacitor test vector corresponding to the associated capacitor test data in the associated capacitor data and the associated capacitor production vector corresponding to the associated capacitor production data, wherein the test data is used to characterize the test process of the corresponding flexible DC support capacitor, and the production data is used to characterize the production process of the corresponding flexible DC support capacitor, as described above. Based on this, the above-mentioned step S130 can further include step S131, step S132, step S133 and step S134, and the specific contents of each step are described as follows.

[0091] Step S131, performing a vector space mapping operation on the target capacitor test vector, outputting a corresponding target test mapping vector, and performing a vector space mapping operation on the associated capacitor test vector corresponding to each of the associated capacitor data, outputting an associated test mapping vector corresponding to each of the associated capacitor data.

[0092] In an embodiment of the present invention, the target capacitor test vector may be subjected to a vector space mapping operation to output a corresponding target test mapping vector, and the associated capacitor test vector corresponding to each of the associated capacitor data may be subjected to a vector space mapping operation to output an associated test mapping vector corresponding to each of the associated capacitor data. Exemplarily, the first matrix in the corresponding neural network (such as the global semantic aggregation unit included in the capacitor quality analysis network described later) may be multiplied with the target capacitor test vector to obtain a corresponding target test mapping vector, and the second matrix may be multiplied with the associated capacitor test vector to obtain a corresponding associated test mapping vector. In this way, the mapping of the vector space may be realized so as to facilitate the subsequent coordinated processing of the target test mapping vector and the associated test mapping vector.

[0093] Step S132, performing a vector space mapping operation on the target capacitor production vector, outputting a corresponding target production mapping vector, and performing a vector space mapping operation on the associated capacitor production vector corresponding to each of the associated capacitor data, respectively, outputting an associated production mapping vector corresponding to each of the associated capacitor data.

[0094] In an embodiment of the present invention, the target capacitor production vector may be subjected to a vector space mapping operation to output a corresponding target production mapping vector, and the associated capacitor production vector corresponding to each of the associated capacitor data may be subjected to a vector space mapping operation to output an associated production mapping vector corresponding to each of the associated capacitor data. Exemplarily, the third matrix in the corresponding neural network may be multiplied by the target capacitor production vector to obtain a corresponding target production mapping vector, and the fourth matrix may be multiplied by the associated capacitor production vector to obtain a corresponding associated production mapping vector. In this way, the mapping of the vector space may be realized so as to facilitate the subsequent coordinated processing of the target production mapping vector and the associated production mapping vector.

[0095] Step S133 , performing semantic mining on the capacitor quality data corresponding to each of the associated capacitor data, and outputting a capacitor quality semantic vector corresponding to each of the associated capacitor data.

[0096] In an embodiment of the present invention, semantic mining can be performed on the capacitor quality data corresponding to each of the associated capacitor data, and a capacitor quality semantic vector corresponding to each of the associated capacitor data can be output. Exemplarily, word embedding processing can be performed on the capacitor quality data.

[0097] Step S134: determining a corresponding global capacitor semantic vector based on the target test mapping vector, the associated test mapping vector, the target production mapping vector, the associated production mapping vector and the capacitor quality semantic vector.

[0098] In an embodiment of the present invention, after obtaining the target test mapping vector, the associated test mapping vector, the target production mapping vector, the associated production mapping vector and the capacitor quality semantic vector, a corresponding global capacitor semantic vector can be determined based on the target test mapping vector, the associated test mapping vector, the target production mapping vector, the associated production mapping vector and the capacitor quality semantic vector. Based on this, the global capacitor semantic vector can carry the semantic information of the aforementioned five vectors and has a high semantic richness.

[0099] In the above-mentioned step S134, the specific method of aggregating the vectors of five dimensions to form the corresponding global capacitor semantic vector is not limited and can be selected according to actual needs. For example, in a specific implementation, in order to achieve sufficient aggregation of vectors of multiple dimensions and ensure that the obtained global capacitor semantic vector can be reliably semantically represented, the above-mentioned step S134 can further include step S134a, step S134b, step S134c and step S134d, and the specific content of each step is described as follows.

[0100] Step S134a, for each of the associated capacitor data, based on the capacitor quality semantic vector corresponding to the associated capacitor data, focus mining is performed on the associated test mapping vector and the corresponding associated production mapping vector corresponding to the associated capacitor data, and the first test focus vector and the first production focus vector corresponding to the associated capacitor data are output.

[0101] In an embodiment of the present invention, for each of the associated capacitor data, based on the capacitor quality semantic vector corresponding to the associated capacitor data, the associated test mapping vector and the corresponding associated production mapping vector corresponding to the associated capacitor data are respectively focused and mined, and the first test focus vector and the first production focus vector corresponding to the associated capacitor data are output. In other words, the associated test mapping vector is focused and mined based on the capacitor quality semantic vector, so that the semantic information in the capacitor quality semantic vector is integrated into the associated test mapping vector, and a first test focus vector with richer semantics is obtained. In addition, the associated production mapping vector is focused and mined based on the capacitor quality semantic vector, so that the semantic information in the capacitor quality semantic vector is integrated into the associated production mapping vector, and a first production focus vector with richer semantics is obtained.

[0102] Step S134b, for each of the associated capacitor data, based on the first test focus vector corresponding to the associated capacitor data, focus mining is performed on the target test mapping vector, and the second test focus vector corresponding to the associated capacitor data is output; and based on the first production focus vector corresponding to the associated capacitor data, focus mining is performed on the target production mapping vector, and the second production focus vector corresponding to the associated capacitor data is output.

[0103] In an embodiment of the present invention, for each of the associated capacitor data, based on the first test focus vector corresponding to the associated capacitor data, the target test mapping vector is focused and mined, and the second test focus vector corresponding to the associated capacitor data is output; and, based on the first production focus vector corresponding to the associated capacitor data, the target production mapping vector is focused and mined, and the second production focus vector corresponding to the associated capacitor data is output. In other words, the semantic information of the first test focus vector can be fused into the target test mapping vector to obtain a second test focus vector with richer semantics. Furthermore, the semantic information of the first production focus vector can be fused into the target production mapping vector to obtain a second production focus vector with richer semantics. In this way, the semantic information of the vectors of the test dimension and the vectors of the production dimension can be fused separately, thereby ensuring the reliability of the fusion of each semantic information.

[0104] Step S134c, for each of the associated capacitor data, based on the second production focusing vector corresponding to the associated capacitor data, focus mining is performed on the second test focusing vector corresponding to the associated capacitor data, and the capacitor focusing vector corresponding to the associated capacitor data is output.

[0105] In an embodiment of the present invention, after the second production focus vector and the second test focus vector are respectively fused to form, for each of the associated capacitor data, based on the second production focus vector corresponding to the associated capacitor data, the second test focus vector corresponding to the associated capacitor data can be focused and mined, and the capacitor focus vector corresponding to the associated capacitor data can be output. In other words, the semantic information in the second production focus vector is fused into the semantic information of the second test focus vector to obtain a capacitor focus vector with richer semantics. Among them, since the required quality is more strongly correlated with the test data, it is necessary to take the second test focus vector as the main one and the second production focus vector as the auxiliary one, that is, to fuse the second production focus vector into the second test focus vector, so that the formed capacitor focus vector can be guaranteed to have a better semantic representation effect in the test dimension.

[0106] Step S134d: Determine a corresponding global capacitor semantic vector based on the capacitor focus vector corresponding to each of the associated capacitor data.

[0107] In an embodiment of the present application, after obtaining the capacitor focus vector, the corresponding global capacitor semantic vector can be determined based on the capacitor focus vector corresponding to each of the associated capacitor data. For example, the capacitor focus vector corresponding to each of the associated capacitor data can be concatenated, averaged or weighted averaged to obtain the global capacitor semantic vector. Alternatively, in some specific implementations, the capacitor focus vector corresponding to each of the associated capacitor data can also be serialized to obtain a corresponding vector sequence (the condition for serialization is that the mean of the vector cosine similarity between each two adjacent capacitor focus vectors is the largest), and then, based on the focus mining result corresponding to the previous capacitor focus vector, the next capacitor focus vector can be focused and mined, and the focus mining result corresponding to the next capacitor focus vector can be output, wherein the focus mining result corresponding to the first capacitor focus vector can be the first capacitor focus vector itself, and the focus mining result corresponding to the last capacitor focus vector can be used as the corresponding global capacitor semantic vector, or the focus mining result corresponding to the last capacitor focus vector can be superimposed with the target test mapping vector to obtain the corresponding global capacitor semantic vector, so that the problem of semantic distortion caused by deep focus mining can be avoided.

[0108] In the above step S134a, the specific manner of focusing on mining the associated test mapping vector and the associated production mapping vector corresponding to the associated capacitor data is not limited and can be selected according to actual needs. For example, in a specific implementation, the above step S134a can further include the following contents:

[0109] First, the capacitor quality semantic vector corresponding to the associated capacitor data may be linearly mapped (such as multiplied by the first mapping matrix in the corresponding neural network) to obtain a capacitor quality linear vector, and the associated test mapping vector corresponding to the associated capacitor data may be subjected to two different linear mappings (such as multiplied by the second mapping matrix and the third mapping matrix in the corresponding neural network, respectively) to obtain a first associated test linear vector and a second associated test linear vector, and the associated production mapping vector corresponding to the associated capacitor data may be subjected to two different linear mappings (such as multiplied by the fourth mapping matrix and the fifth mapping matrix in the corresponding neural network, respectively) to obtain a first associated production linear vector and a second associated production linear vector;

[0110] Secondly, based on the correlation parameter between the capacitor quality linear vector and the first associated test linear vector (such as the multiplication result of the capacitor quality linear vector and the transposed vector of the first associated test linear vector), the second associated test linear vector is adjusted (such as multiplied) to obtain a first test focus vector corresponding to the associated capacitor data;

[0111] Then, based on the correlation parameter between the capacitor quality linear vector and the first associated production linear vector (such as the multiplication result of the capacitor quality linear vector and the transposed vector of the first associated production linear vector), the second associated production linear vector can be adjusted (such as multiplied) to obtain the first production focus vector corresponding to the associated capacitor data.

[0112] In the above step S134b, the specific manner of performing focused mining on the target test mapping vector and the specific manner of performing focused mining on the target production mapping vector is not limited and can be selected according to actual needs. For example, in an alternative implementation, in order to avoid excessive focus on some universal features during the focused mining process, so that the reliability of the obtained second test focused vector and the second production focused vector can be higher, the above step S134b can further include step b1, step b2 and step b3. The specific contents of each step are as follows (combined with Figure 3 ).

[0113] Step b1, performing difference calculation on the first test focus vector corresponding to the associated capacitor data and the target test mapping vector to obtain the corresponding difference test vector, and generating multiple perturbation test vectors based on the difference test vector, and performing difference calculation on the first production focus vector corresponding to the associated capacitor data and the target production mapping vector to obtain the corresponding difference production vector, and generating multiple perturbation production vectors based on the difference production vector.

[0114] In an embodiment of the present invention, the first test focus vector corresponding to the associated capacitor data and the target test mapping vector can be differenced (i.e., the vector parameters at corresponding positions are differenced) to obtain a corresponding difference test vector, and based on the difference test vector, a plurality of perturbation test vectors can be generated, and the first production focus vector corresponding to the associated capacitor data and the target production mapping vector can be differenced to obtain a corresponding difference production vector, and based on the difference production vector, a plurality of perturbation production vectors can be generated. Exemplarily, since the difference test vector is obtained based on difference calculation, in the difference test vector, the closer the vector parameter is to zero, the more similar the first test focus vector and the target test mapping vector are at corresponding positions. Thus, in order to avoid paying too much attention to the similarity features, the vector position where the vector parameter is equal to zero in the difference test vector can be determined first, and then, in the target test mapping vector, a disturbance is applied once to the vector parameter corresponding to the vector position (which can be a small number of vector parameters to avoid introducing too much disturbance and causing complete semantic distortion), thereby forming a disturbance test vector, wherein the disturbance applied twice is not exactly the same, that is, the position and / or amplitude of the disturbance applied are not exactly the same, and the amplitude can refer to the proportion of increase or decrease.

[0115] Step b2, in the process of focusing and mining the target test mapping vector based on the first test focusing vector corresponding to the associated capacitor data, applying the multiple perturbation test vectors and outputting the second test focusing vector corresponding to the associated capacitor data.

[0116] In an embodiment of the present application, after obtaining a plurality of perturbation test vectors, the plurality of perturbation test vectors can be applied in the process of focusing and mining the target test mapping vector based on the first test focus vector corresponding to the associated capacitor data, and a second test focus vector corresponding to the associated capacitor data can be output, so that the semantics of the second test focus vector is richer.

[0117] Step b3, in the process of focusing and mining the target production mapping vector based on the first production focusing vector corresponding to the associated capacitor data, applying the multiple perturbation production vectors and outputting the second production focusing vector corresponding to the associated capacitor data.

[0118] In an embodiment of the present application, after obtaining a plurality of perturbation production vectors, the plurality of perturbation production vectors can be applied in the process of focusing mining the target production mapping vector based on the first production focusing vector corresponding to the associated capacitor data, and a second production focusing vector corresponding to the associated capacitor data can be output, so that the semantics of the second production focusing vector is richer.

[0119] In the above step b2, the specific method of applying the multiple perturbation test vectors is not limited and can be selected according to actual needs. For example, in an alternative implementation, in order to achieve deep mining of features in the process of applying perturbations, the above step b2 can further include the following specific contents (combined with Figure 4 ):

[0120] First, each of the multiple perturbation test vectors and the first test focus vector corresponding to the associated capacitor data can be superimposed respectively, and the corresponding multiple perturbation test focus vectors are output, and the multiple perturbation test focus vectors are serialized to form a target sequence; illustratively, they can be sorted arbitrarily;

[0121] Secondly, for the first perturbation test focus vector in the target sequence, based on the perturbation test focus vector, focus mining can be performed on the target test mapping vector (refer to the description of step S134a in the previous text), and a focus mining result corresponding to the perturbation test focus vector can be output;

[0122] Then, for each perturbation test focusing vector other than the first perturbation test focusing vector in the target sequence, based on the perturbation test focusing vector, focused mining can be performed on the focused mining result corresponding to the previous perturbation test focusing vector of the perturbation test focusing vector, and the focused mining result corresponding to the perturbation test focusing vector can be output;

[0123] Finally, based on at least the focus mining result corresponding to the last perturbation test focus vector in the target sequence, the second test focus vector corresponding to the associated capacitor data is determined; exemplarily, the focus mining result corresponding to the last perturbation test focus vector can be used as the second test focus vector corresponding to the associated capacitor data, or the focus mining result corresponding to the last perturbation test focus vector can be superimposed with the target test mapping vector to obtain the second test focus vector corresponding to the associated capacitor data.

[0124] In the above step S140, the specific method of analyzing the capacitor quality data corresponding to the target capacitor data is not limited and can be selected according to actual needs.

[0125] For example, in an alternative implementation, in order to ensure that the analyzed capacitor quality data has a high reliability, the above step S140 may further include the following contents:

[0126] First, the global capacitor semantic vector may be subjected to semantic space conversion processing (for example, it may be multiplied by a conversion parameter first, and then the multiplication result may be added to a bias vector), and a converted capacitor semantic vector corresponding to the global capacitor semantic vector may be output, wherein the converted capacitor semantic vector is a semantic vector that can be processed by a capacitor quality analysis network;

[0127] Secondly, the conversion capacitor semantic vector can be loaded so that the capacitor quality analysis network acquires the conversion capacitor semantic vector, wherein the capacitor quality analysis network updates the network parameters of the second part of the candidate capacitor quality analysis network based on the training conversion capacitor semantic vector while maintaining the network parameters of the first part of the candidate capacitor quality analysis network, and the network parameters of the second part refer to the network parameters of the semantic adaptation unit in the candidate capacitor quality analysis network; that is, in the candidate capacitor quality analysis network, including the first part and the second part, the two parts can be trained separately, so that the training accuracy can be relatively higher;

[0128] Then, the converted capacitor semantic vector can be analyzed through the capacitor quality analysis network (the semantic adaptation unit therein) to obtain the capacitor quality data corresponding to the target capacitor data. For example, the converted capacitor semantic vector is processed through the fully connected subunit in the semantic adaptation unit to obtain a fully connected vector, and then the fully connected vector is processed through the output function in the semantic adaptation unit to obtain the capacitor quality data corresponding to the target capacitor data.

[0129] In the above steps, it is mentioned that the capacitor quality analysis network is formed by updating the candidate capacitor quality analysis network, wherein the step of updating and forming the capacitor quality analysis network may include the following specific contents:

[0130] First, the first capacitor test vector, the first capacitor production vector and the first capacitor actual quality corresponding to the first capacitor data may be determined, and the second capacitor test vector, the second capacitor production vector and the second capacitor actual quality corresponding to the second capacitor data may be determined, and reference may be made to the relevant explanations of step S110 hereinabove;

[0131] Secondly, the global semantic aggregation unit in the candidate capacitor quality analysis network may be used to perform global semantic aggregation on the first capacitor test vector, the first capacitor production vector, the second capacitor test vector, the second capacitor production vector, and the second capacitor actual quality to form a corresponding training global capacitor semantic vector, and reference may be made to the relevant explanation of step S130 in the previous text;

[0132] Then, the training global capacitor semantic vector may be subjected to semantic space conversion processing, and a training conversion capacitor semantic vector corresponding to the training global capacitor semantic vector may be output;

[0133] Further, the semantic adaptation unit and the candidate capacitor quality analysis network may be fused (i.e., a candidate capacitor quality analysis network including a global semantic aggregation unit and a semantic adaptation unit is formed), so that the network parameters of the second part are fused in the candidate capacitor quality analysis network based on the semantic adaptation unit, wherein the network parameters of the first part refer to the network parameters of the global semantic aggregation unit;

[0134] Finally, the network parameters of the first part of the candidate capacitor quality analysis network can be maintained, and the training conversion capacitor semantic vector can be analyzed by the semantic adaptation unit to obtain the first capacitor analysis quality, and, based on the error between the first capacitor analysis quality and the actual quality of the first capacitor, the network parameters of the second part of the candidate capacitor quality analysis network can be updated to form a capacitor quality analysis network; exemplarily, the network parameters of the second part can be updated in the direction of reducing the error until the error converges, thereby obtaining a capacitor quality analysis network.

[0135] In addition, it should be noted that for the global semantic aggregation unit, during the individual training process, sample capacitor test vectors, sample capacitor production vectors, and actual capacitor quality corresponding to multiple sample capacitor data can be obtained. Then, two sample capacitor data with the same actual capacitor quality can be formed into a sample positive combination, and two sample capacitor data with different actual capacitor quality can be formed into a sample negative combination.

[0136] For the sample forward combination, the first sample capacitor data can be used as the target capacitor data, and the second sample capacitor data can be used as the associated capacitor data, and then, based on the processing method of step S130, processing can be performed to obtain the corresponding first sample global capacitor semantic vector; and the first sample capacitor data can be used as the associated capacitor data, and the second sample capacitor data can be used as the target capacitor data, and then, based on the processing method of step S130, processing can be performed to obtain the corresponding second sample global capacitor semantic vector, and then, the network parameters of the global semantic aggregation unit can be updated in the direction of reducing the difference between the first sample global capacitor semantic vector and the second sample global capacitor semantic vector.

[0137] For the sample negative combination, the first sample capacitor data can be used as the target capacitor data, and the second sample capacitor data can be used as the associated capacitor data, and then, based on the processing method of step S130, processing can be performed to obtain the corresponding first sample global capacitor semantic vector; and the first sample capacitor data can be used as the associated capacitor data, and the second sample capacitor data can be used as the target capacitor data, and then, based on the processing method of step S130, processing can be performed to obtain the corresponding second sample global capacitor semantic vector, and then, the network parameters of the global semantic aggregation unit can be updated in the direction of increasing the difference between the first sample global capacitor semantic vector and the second sample global capacitor semantic vector.

[0138] In summary, the flexible DC support capacitor quality analysis method and system based on artificial intelligence provided by the present invention firstly extracts the target capacitor data corresponding to the flexible DC support capacitor, and semantically mines the target capacitor data to output the target capacitor semantic vector; secondly, determines multiple associated capacitor data corresponding to the target capacitor data based on the target capacitor semantic vector; then, based on the associated capacitor semantic vector and the corresponding capacitor quality data corresponding to each associated capacitor data, and in combination with the target capacitor semantic vector, determines the global capacitor semantic vector; finally, based on the global capacitor semantic vector, analyzes the capacitor quality data corresponding to the target capacitor data. Based on the above content, since the global capacitor semantic vector is formed based on the associated capacitor semantic vector and the corresponding capacitor quality data corresponding to each associated capacitor data, and in combination with the target capacitor semantic vector, the global capacitor semantic vector not only carries the relevant semantic information of the flexible DC support capacitor, but also carries the semantic information of the associated associated capacitor data and the capacitor quality data, so that the semantic representation ability of the global capacitor semantic vector is better. Therefore, when performing quality analysis based on the global capacitor semantic vector, it can have a higher reliability, thereby improving the problem of relatively low reliability of capacitor quality analysis existing in the prior art.

[0139] In several embodiments provided in the embodiments of the present invention, it should be understood that the disclosed apparatus and method can also be implemented in other ways. The apparatus and method embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the apparatus, method and computer program product according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0140] In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0141] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, an electronic device, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code. It should be noted that in this article, the term "include", "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such process, method, article or device. Without more constraints, an element defined by the phrase "comprising a..." does not exclude the existence of other identical elements in the process, method, article or apparatus comprising the element.

[0142] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A quality analysis method for flexible DC support capacitors based on artificial intelligence, characterized in that: The flexible DC support capacitor quality analysis method based on artificial intelligence includes: Extracting target capacitor data corresponding to the flexible DC support capacitor, and performing semantic mining on the target capacitor data to output a semantic vector of the target capacitor; Determine a plurality of associated capacitor data corresponding to the target capacitor data based on the target capacitor semantic vector, wherein each of the associated capacitor data has capacitor quality data; Determine a global capacitor semantic vector based on the associated capacitor semantic vector corresponding to each of the associated capacitor data and the capacitor quality data corresponding to each of the associated capacitor data, and in combination with the target capacitor semantic vector; Analyzing capacitor quality data corresponding to the target capacitor data based on the global capacitor semantic vector, wherein the capacitor quality data corresponding to the target capacitor data is used to reflect the quality of the flexible DC support capacitor; The target capacitor semantic vector includes a target capacitor test vector corresponding to the target capacitor test data in the target capacitor data and a target capacitor production vector corresponding to the target capacitor production data, and the associated capacitor semantic vector includes an associated capacitor test vector corresponding to the associated capacitor test data in the associated capacitor data and an associated capacitor production vector corresponding to the associated capacitor production data; The test data is used to characterize the test process of the corresponding flexible DC support capacitor, including a description text of the performance test of the flexible DC support capacitor, and the production data is used to characterize the production process of the corresponding flexible DC support capacitor, including a monitoring image of the flexible DC support capacitor during the production process; the step of determining the corresponding global capacitor semantic vector based on the associated capacitor semantic vector corresponding to each of the associated capacitor data and the capacitor quality data corresponding to each of the associated capacitor data, and in combination with the target capacitor semantic vector, includes: Performing a vector space mapping operation on the target capacitor test vector and outputting a corresponding target test mapping vector, and respectively performing a vector space mapping operation on the associated capacitor test vector corresponding to each of the associated capacitor data and outputting an associated test mapping vector corresponding to each of the associated capacitor data; Performing a vector space mapping operation on the target capacitor production vector and outputting a corresponding target production mapping vector, and performing a vector space mapping operation on the associated capacitor production vector corresponding to each of the associated capacitor data and outputting an associated production mapping vector corresponding to each of the associated capacitor data; Performing semantic mining on capacitor quality data corresponding to each of the associated capacitor data, and outputting a capacitor quality semantic vector corresponding to each of the associated capacitor data; Determining a corresponding global capacitor semantic vector based on the target test mapping vector, the associated test mapping vector, the target production mapping vector, the associated production mapping vector, and the capacitor quality semantic vector; The step of determining a corresponding global capacitor semantic vector based on the target test mapping vector, the associated test mapping vector, the target production mapping vector, the associated production mapping vector and the capacitor quality semantic vector comprises: For each of the associated capacitor data, based on the capacitor quality semantic vector corresponding to the associated capacitor data, focus mining is performed on the associated test mapping vector and the associated production mapping vector corresponding to the associated capacitor data, and a first test focus vector and a first production focus vector corresponding to the associated capacitor data are output; For each of the associated capacitor data, based on the first test focus vector corresponding to the associated capacitor data, the target test mapping vector is focused and mined, and a second test focus vector corresponding to the associated capacitor data is output; and based on the first production focus vector corresponding to the associated capacitor data, the target production mapping vector is focused and mined, and a second production focus vector corresponding to the associated capacitor data is output; For each of the associated capacitor data, based on the second production focusing vector corresponding to the associated capacitor data, focus mining is performed on the second test focusing vector corresponding to the associated capacitor data, and the capacitor focusing vector corresponding to the associated capacitor data is output; Determining a corresponding global capacitor semantic vector based on a capacitor focus vector corresponding to each of the associated capacitor data; The step of focusing and mining the associated test mapping vector and the associated production mapping vector corresponding to the associated capacitor data based on the capacitor quality semantic vector corresponding to the associated capacitor data for each of the associated capacitor data, and outputting the first test focus vector and the first production focus vector corresponding to the associated capacitor data comprises: Linearly mapping the capacitor quality semantic vector corresponding to the associated capacitor data to obtain a capacitor quality linear vector, and performing two different linear mappings on the associated test mapping vector corresponding to the associated capacitor data to obtain a first associated test linear vector and a second associated test linear vector, and performing two different linear mappings on the associated production mapping vector corresponding to the associated capacitor data to obtain a first associated production linear vector and a second associated production linear vector; Based on the correlation parameter between the capacitor quality linear vector and the first associated test linear vector, the second associated test linear vector is adjusted to obtain a first test focus vector corresponding to the associated capacitor data; Based on the correlation parameter between the capacitor quality linear vector and the first associated production linear vector, the second associated production linear vector is adjusted to obtain a first production focus vector corresponding to the associated capacitor data; The target capacitor semantic vector includes a target capacitor test vector corresponding to the target capacitor test data and a target capacitor production vector corresponding to the target capacitor production data in the target capacitor data; The step of determining a plurality of associated capacitor data corresponding to the target capacitor data based on the target capacitor semantic vector includes: Determine a pre-built test vector set and a production vector set, wherein the test vector set includes each pending capacitor data and a test vector corresponding to the pending capacitor data, and the production vector set includes each pending capacitor data and a production vector corresponding to the pending capacitor data; Based on the target capacitor test vector, performing a comparative analysis in the test vector set and the production vector set, and based on the target capacitor production vector, performing a comparative analysis in the test vector set and the production vector set to determine a plurality of pending capacitor data matching the target capacitor data; Based on a test data matching parameter between a plurality of candidate capacitor data matching the target capacitor data and the target capacitor data, a plurality of associated capacitor data corresponding to the target capacitor data is determined from among the plurality of candidate capacitor data matching the target capacitor data.

2. The method for quality analysis of flexible DC support capacitors based on artificial intelligence according to claim 1, characterized in that: The step of focusing and mining the target test mapping vector based on the first test focusing vector corresponding to the associated capacitor data for each of the associated capacitor data, and outputting the second test focusing vector corresponding to the associated capacitor data, and focusing and mining the target production mapping vector based on the first production focusing vector corresponding to the associated capacitor data, and outputting the second production focusing vector corresponding to the associated capacitor data, comprises: Performing a difference calculation on a first test focus vector corresponding to the associated capacitor data and the target test mapping vector to obtain a corresponding difference test vector, and generating a plurality of perturbation test vectors based on the difference test vector; and performing a difference calculation on a first production focus vector corresponding to the associated capacitor data and the target production mapping vector to obtain a corresponding difference production vector, and generating a plurality of perturbation production vectors based on the difference production vector; In the process of focusing and mining the target test mapping vector based on the first test focusing vector corresponding to the associated capacitor data, applying the plurality of perturbation test vectors and outputting a second test focusing vector corresponding to the associated capacitor data; In the process of focusing and mining the target production mapping vector based on the first production focusing vector corresponding to the associated capacitor data, the multiple perturbation production vectors are applied to output a second production focusing vector corresponding to the associated capacitor data.

3. The method for quality analysis of flexible DC support capacitors based on artificial intelligence according to claim 2, characterized in that: The step of applying the plurality of perturbation test vectors and outputting a second test focus vector corresponding to the associated capacitor data in the process of focusing and mining the target test mapping vector based on the first test focus vector corresponding to the associated capacitor data comprises: Superimposing each of the plurality of perturbation test vectors and the first test focus vector corresponding to the associated capacitor data respectively, outputting a plurality of corresponding perturbation test focus vectors, and serializing the plurality of perturbation test focus vectors to form a target sequence; For the first perturbation test focusing vector in the target sequence, based on the perturbation test focusing vector, focusing mining is performed on the target test mapping vector, and a focusing mining result corresponding to the perturbation test focusing vector is output; For each perturbation test focused vector other than the first perturbation test focused vector in the target sequence, based on the perturbation test focused vector, focused mining is performed on a focused mining result corresponding to a previous perturbation test focused vector of the perturbation test focused vector, and the focused mining result corresponding to the perturbation test focused vector is output; A second test focus vector corresponding to the associated capacitor data is determined based at least on a focus mining result corresponding to a last perturbation test focus vector in the target sequence.

4. The method for analyzing quality of flexible DC support capacitors based on artificial intelligence according to any one of claims 1 to 3, characterized in that: The step of analyzing capacitor quality data corresponding to the target capacitor data based on the global capacitor semantic vector includes: Performing semantic space conversion processing on the global capacitor semantic vector, and outputting a converted capacitor semantic vector corresponding to the global capacitor semantic vector, wherein the converted capacitor semantic vector is a semantic vector that can be processed by a capacitor quality analysis network; The conversion capacitor semantic vector is loaded so that the capacitor quality analysis network acquires the conversion capacitor semantic vector, wherein the capacitor quality analysis network updates the network parameters of the second part of the candidate capacitor quality analysis network based on the training conversion capacitor semantic vector while maintaining the network parameters of the first part of the candidate capacitor quality analysis network, and the network parameters of the second part refer to the network parameters of the semantic adaptation unit in the candidate capacitor quality analysis network; The converted capacitor semantic vector is analyzed through the capacitor quality analysis network to obtain capacitor quality data corresponding to the target capacitor data.

5. The method for quality analysis of flexible DC support capacitors based on artificial intelligence according to claim 4, characterized in that: The step of updating and forming the capacitor quality analysis network comprises: Determine a first capacitor test vector, a first capacitor production vector, and a first capacitor actual quality corresponding to the first capacitor data, and determine a second capacitor test vector, a second capacitor production vector, and a second capacitor actual quality corresponding to the second capacitor data; Using a global semantic aggregation unit in the candidate capacitor quality analysis network, globally semantically aggregate the first capacitor test vector, the first capacitor production vector, the second capacitor test vector, the second capacitor production vector, and the second capacitor actual quality to form a corresponding training global capacitor semantic vector; Performing semantic space conversion processing on the training global capacitor semantic vector, and outputting a training conversion capacitor semantic vector corresponding to the training global capacitor semantic vector; The semantic adaptation unit and the candidate capacitor quality analysis network are merged, so that the network parameters of the second part are merged in the candidate capacitor quality analysis network based on the semantic adaptation unit, wherein the network parameters of the first part refer to the network parameters of the global semantic aggregation unit; The network parameters of the first part of the candidate capacitor quality analysis network are maintained, and the training conversion capacitor semantic vector is analyzed by the semantic adaptation unit to obtain a first capacitor analysis quality, and based on the error between the first capacitor analysis quality and the actual quality of the first capacitor, the network parameters of the second part of the candidate capacitor quality analysis network are updated to form a capacitor quality analysis network.

6. A flexible DC support capacitor quality analysis system based on artificial intelligence, characterized in that: It comprises a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to implement the flexible DC support capacitor quality analysis method based on artificial intelligence as described in any one of claims 1 to 5.

Images