CN109815855B - Electronic equipment automatic test method and system based on machine learning - Google Patents
Electronic equipment automatic test method and system based on machine learning Download PDFInfo
- Publication number
- CN109815855B CN109815855B CN201910012589.XA CN201910012589A CN109815855B CN 109815855 B CN109815855 B CN 109815855B CN 201910012589 A CN201910012589 A CN 201910012589A CN 109815855 B CN109815855 B CN 109815855B
- Authority
- CN
- China
- Prior art keywords
- test
- data
- feature
- characteristic
- machine learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The disclosure provides an electronic device automatic test method and system based on machine learning. The automatic test method for the electronic equipment based on machine learning comprises the following steps: acquiring test data of the electronic equipment, and extracting characteristic components of the test data; the test data includes serialized waveform data and non-serialized video and image data; performing principal component analysis on the test data, determining the correlation of different feature components, and further generating a feature vector; and inputting the characteristic vector into a preset fault diagnosis model, and outputting a prediction result of the electronic test fault.
Description
Technical Field
The disclosure belongs to the field of electronic testing, and particularly relates to an electronic device automatic testing method and system based on machine learning.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The automatic test is the mainstream technology in the field of current electronic test, and is a fault diagnosis technology which collects and inputs data in a client-side form, realizes data display, integrates and analyzes data at a server-side, and takes threshold judgment or an expert system as a decision basis. The automatic test integrates a large amount of test resources, including test instruments, test personnel and test data, and meets the centralized realization and decision judgment of test requirements, wherein the realization of the test requirements is mainly based on systematic platformization data acquisition, and the decision judgment is mainly based on artificial analysis of the data and combination of a physical model of a specific tested object. The automatic test is mainly realized by C/S from the aspect of architecture, namely a client/server architecture, which is determined by the characteristic of the centralization of test resources of the automatic test, and the decision judgment is mainly based on threshold judgment and expert knowledge, which is determined by the test data of the automatic test and the characteristic of the centralization of testers. Compared with the traditional manual test, the automatic test realizes the improvement of the test capability, and the acquisition and sharing capability of a large amount of data is improved to a certain degree by combining the development and introduction of a networking technology; the method combines a statistical method and fault analysis modeling, has certain data analysis and processing capacity, and although the automatic test is widely applied in the field of electronic test, the automatic test can not meet the requirements of modern electronic intelligent test along with the promotion of test requirements of electronic test intellectualization, sharing, accuracy, high efficiency and rapidness.
The traditional automatic test emphasizes reducing the manpower consumption and improving the automation degree in the test link, and has the following defects: firstly, an automatic test technology often adopts a centralized test mode, all links in the test technology are easily limited by regions, and although the automatic test technology can realize data sharing to a certain degree, the intelligent data processing capability is still limited, and the data processing efficiency is low; secondly, the automatic test usually uses expert knowledge, threshold judgment and simple statistical analysis methods, and the deep mining capability of the data is insufficient; thirdly, the current automatic test field is not closely combined with machine learning, and in the execution process of the automatic test, after data are collected according to test resource configuration and a flow, a diagnosis prediction result is obtained through simple preprocessing and a machine learning algorithm. In the process, the data acquisition based on the test and the decision acquisition based on the machine learning are always operated independently, and the intelligent test integration is not realized; fourthly, the intelligent degree of the current automatic test is low, the automatic test process is developed by depending on human experience, and then the test behavior is developed, so that the automatic test cannot update the test resource configuration and process according to the real-time test result and the decision of intelligent inference, and cannot accurately improve the consumption of the test resource according to the data acquired by the specific test.
Disclosure of Invention
According to an aspect of one or more embodiments of the present disclosure, there is provided a machine learning-based electronic device automatic test method capable of achieving sharing of test data and efficiency of test analysis.
The disclosed automatic testing method for electronic equipment based on machine learning comprises the following steps:
acquiring test data of the electronic equipment, and extracting characteristic components of the test data; the test data includes serialized waveform data and non-serialized video and image data;
performing principal component analysis on the test data, determining the correlation of different feature components, and further generating a feature vector;
and inputting the characteristic vector into a preset fault diagnosis model, and outputting a prediction result of the electronic test fault.
In one or more embodiments, the method for automatically testing an electronic device based on machine learning further includes:
and while the prediction result of the electronic test fault is output, the analysis result of the feature vector is returned to the automatic test front end to guide the front end to test resource allocation and the updating and improvement of the process.
In one or more embodiments, the analysis results of the feature vectors include redundant feature information.
In one or more embodiments, in extracting feature components of the test data, for feature extraction of the serialized data:
solving the sum of squares of the amplitudes of each component array of a group of data vectors obtained after test sampling to obtain the total energy value characteristic;
denoising, and then obtaining the denoised peak-to-peak value characteristic by taking the difference between the maximum value and the minimum value in the component array;
giving a certain amplitude threshold value, and then solving the average value of all the amplitude values which are higher than the threshold value in the component array to obtain the amplitude characteristic calculated based on the threshold value;
and giving a certain step length, and solving the sum of squares of the amplitude values at intervals of the certain step length in the component array to obtain the energy value characteristic based on the certain step length.
In one or more embodiments, in extracting feature components of the test data, for feature extraction of the nonserialized image data:
the input feature vector is a pixel point matrix, a two-dimensional matrix is obtained through reconstruction, then the pixel point matrix is further compressed by utilizing convolution mathematical operation, and feature extraction of non-serialized image data is obtained.
According to another aspect of one or more embodiments of the present disclosure, there is provided a machine learning-based electronic device automatic test system capable of realizing sharing of test data and efficiency of test analysis.
The utility model discloses an electronic equipment automatic test system based on machine learning, including automatic testing front end and server end, the server end includes:
the characteristic classification extraction module is used for acquiring test data of the electronic equipment and extracting characteristic components of the test data; the test data includes serialized waveform data and non-serialized video and image data;
the feature vector generation module is used for performing principal component analysis on the test data, determining the correlation of different feature components and further generating feature vectors;
and the prediction result output module is used for inputting the characteristic vector into a preset fault diagnosis model and outputting the prediction result of the electronic test fault.
In one or more embodiments, the server further includes:
and the characteristic vector analysis result feedback module is used for outputting the prediction result of the electronic test fault and returning the analysis result of the characteristic vector to the automatic test front end to guide the front end to test resource allocation and the updating and improvement of the process.
In one or more embodiments, the analysis results of the feature vectors include redundant feature information.
In one or more embodiments, the feature classification extraction module extracts features for the serialized data:
solving the sum of squares of the amplitudes of each component array of a group of data vectors obtained after test sampling to obtain the total energy value characteristic;
denoising, and then obtaining the denoised peak-to-peak value characteristic by taking the difference between the maximum value and the minimum value in the component array;
giving a certain amplitude threshold value, and then solving the average value of all the amplitude values which are higher than the threshold value in the component array to obtain the amplitude characteristic calculated based on the threshold value;
and giving a certain step length, and solving the sum of squares of the amplitude values at intervals of the certain step length in the component array to obtain the energy value characteristic based on the certain step length.
In one or more embodiments, the feature classification extraction module extracts features for the nonserialized image data:
the input feature vector is a pixel point matrix, a two-dimensional matrix is obtained through reconstruction, then the pixel point matrix is further compressed by utilizing convolution mathematical operation, and feature extraction of non-serialized image data is obtained.
The beneficial effects of this disclosure are:
(1) the intelligent test platform architecture for electronic test realizes sharing of test data and high efficiency of test analysis;
(2) based on the feature extraction and preprocessing technology of different test signals, data analysis and input are provided for the application of expert knowledge and the realization of machine learning;
(3) the machine learning analyzes the correlation of the automatic test acquisition characteristics, and feeds the correlation back to the automatic test front end for optimizing data acquisition characteristic selection and optimizing test resource configuration and flow, thereby meeting the intelligent requirements of modern electronic automatic tests.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure and are not to limit the disclosure.
FIG. 1 is a diagram of a machine learning based architecture for an automatic test system for electronic devices.
Fig. 2 is a flow chart of feature extraction.
Fig. 3 is a schematic diagram of a machine learning based automatic test system for electronic devices according to the present disclosure.
Detailed Description
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
Machine learning is a product of artificial intelligence development to a certain stage, and is a subject across multiple fields such as statistics, optimization theory, neural network, computer science, biological science and the like. Machine learning is based on data, learns and trains a special model suitable for a specific data set, and provides high-precision, high-efficiency, intelligent and personalized decision support. The current mainstream automatic test technology solves the centralization and automation of test flow and data acquisition, and the machine learning algorithm realizes the supplement and deepening of the automatic test technology, namely, the logic relation between the internal complex structure of the electronic equipment and fault diagnosis and prediction is constructed by using the test data acquired by automatic test. However, the current machine learning is not tightly combined in the automatic test, and the intelligence degree is low.
Fig. 1 is a flowchart of an automatic testing method for electronic devices based on machine learning according to the present disclosure.
As shown in fig. 1, an automatic testing method for electronic devices based on machine learning according to the present disclosure includes:
s101: acquiring test data of the electronic equipment, and extracting characteristic components of the test data; the test data includes serialized waveform data and nonserialized video and image data.
In the disclosure, different feature acquisition methods are specifically proposed according to different data types. Different sensors or instrument equipment are required to be adopted for acquiring test data of the electronic equipment, and the acquired data types are various and include but are not limited to serialized time domain waveform data, non-serialized image data and the like.
In order to meet the input requirements of machine learning, different features of input data need to be extracted to form feature vectors, the dimensions of the feature vectors are the dimensions of the machine learning input, and the components of the feature vectors are different features of the machine learning input.
The input to machine learning, whether for serialized or non-serialized data, is a feature vector, i.e., a set of vectors in the mathematical sense, the dimensions of the vector representing the number of input features, and each component of the vector representing a particular feature. For example, for feature extraction of serialized data, the input feature vector is { current amplitude, voltage amplitude, temperature value, total energy of vibration signal }, the machine learning input dimension is 4, and each component of the feature vector represents a corresponding physical index.
As shown in fig. 2, for feature extraction of non-serialized data, taking image data as an example, an input feature vector is a pixel matrix, a two-dimensional matrix is obtained through reconstruction, and then further compression of the pixel matrix, that is, feature extraction of non-serialized data, can be realized by using convolution mathematical operation.
Most of the results after electronic test sampling are serialized data in the form of a set of data vectors, each component of the set of vectors being a multi-dimensional array of samples taken at one time. For such data, the following feature component determination methods are proposed, as shown in fig. 2:
solving the sum of squares of the amplitudes of each component array of a group of data vectors obtained after test sampling to obtain the total energy value characteristic;
denoising, and then obtaining the denoised peak-to-peak value characteristic by taking the difference between the maximum value and the minimum value in the component array;
giving a certain amplitude threshold value, and then solving the average value of all the amplitude values which are higher than the threshold value in the component array to obtain the amplitude characteristic calculated based on the threshold value;
and giving a certain step length, and solving the sum of squares of the amplitude values at intervals of the certain step length in the component array to obtain the energy value characteristic based on the certain step length.
S102: and performing principal component analysis on the test data, determining the correlation of different characteristic components, and further generating a characteristic vector.
For automatic test of electronic equipment, firstly test data is obtained through data acquisition, then each characteristic component is obtained after data processing such as cleaning, denoising, calculation of various characteristic components and the like is carried out on the test data, then principal component analysis is carried out on each characteristic component to determine the correlation of different characteristic components, so that redundant and weak correlation characteristics are obtained, and finally the redundant and weak correlation characteristics are removed from a characteristic vector, so that the composition of the characteristic vector is determined.
S103: and inputting the characteristic vector into a preset fault diagnosis model, and outputting a prediction result of the electronic test fault.
Specifically, the predetermined fault diagnosis model may be a convolutional neural network, or other existing neural network model.
The preset fault diagnosis model can be trained by adopting the historical characteristic vectors until the accuracy of the output prediction result of the electronic test fault meets the preset accuracy requirement, and then the training of the preset fault diagnosis model is completed.
In another embodiment, the method for automatically testing an electronic device based on machine learning further includes:
s104: and while the prediction result of the electronic test fault is output, the analysis result of the feature vector is returned to the automatic test front end to guide the front end to test resource allocation and the updating and improvement of the process.
S1041: firstly, determining automatic test resource allocation and flow, and acquiring data characteristics as complete as possible according to previous experience or current requirements.
Specifically, according to the collected data features, feature vectors are obtained through the above-mentioned feature extraction method, and then a principal component analysis method is applied to determine redundant and weakly correlated features. For example, setting the cumulative feature contribution rate to reach 80% in advance, and after giving a feature component set { x1, x2, x3, x4, x5, x6} (where x1 to x6 may respectively represent various features such as voltage amplitude, current amplitude, temperature, total vibration energy value, current peak-to-peak value, voltage peak-to-peak value, etc.), applying a principal component analysis method to obtain a feature component ranking as: x3, x4, x2, x5, x1, x 6; if the cumulative contribution rate of the first four principal components x3, x4, x2 and x5 reaches 80%, the final feature vector is determined to be { x2, x3, x4 and x5}, the redundancy and weak correlation features are the fifth principal component and the sixth principal component { x1 and x6}, the feature which has a large contribution effect on the final machine learning model is { x2, x3, x4 and x5}, namely the current amplitude, the temperature, the total vibration energy value and the current peak-to-peak value, and the feature which has a small contribution effect is { x1, x6}, namely the voltage amplitude and the voltage peak-to-peak value.
S1042: and then converting the acquired data information into a feature vector, applying the feature vector to a background machine learning algorithm, and determining whether the model effect and the precision under the current feature set meet the test requirements or not by a principal component analysis method and a model result hypothesis test method, namely which features are redundant and weakly related features.
Training the extracted feature vectors { x2, x3, x4 and x5} by applying a machine learning algorithm to obtain a training result, carrying out model accuracy hypothesis test on the training result, setting a confidence coefficient and an accuracy rate in advance, and giving an original hypothesis (for example, "under 85% of confidence coefficient, the model accuracy rate is more than or equal to 80%"). If the model test does not pass the original hypothesis, the model effect is considered to be poor, a fifth principal component x1 is added into the eigenvector { x2, x3, x4, x5} again, the eigenvector is updated to { x2, x3, x4, x5, x1}, and the redundant and weakly correlated characteristic is determined to be x6, namely the voltage peak value; and if the model passes the original hypothesis through the model inspection, the model effect is considered to meet the requirement, the characteristic vectors are determined to be { x2, x3, x4 and x5}, the characteristic vectors do not need to be updated, and the redundant and weakly correlated characteristics are determined to be x1 and x6, namely the voltage amplitude and the voltage peak-to-peak value.
S1043: and finally, feeding back the information of the redundant features to an automatic test front end, optimizing resource configuration in the automatic test resource configuration and process, and eliminating test steps related to the redundant features, so that the test resource configuration and process updating and improvement are completed, and the subsequent test is more accurate and efficient.
Finally, the information of the redundant features is fed back to the automatic testing front end, namely, the testing front end is informed of which features do not need to be acquired or which features do not need to be extracted, the resource configuration is optimized in the automatic testing resource configuration and process, and the testing steps related to the redundant features are eliminated, namely, for the features which do not need to be acquired, the automatic testing can eliminate the acquisition process and the testing resources (for example, if the voltage features do not need to be acquired, the testing front end does not need to use voltage acquisition equipment, and does not need to design and arrange the testing process of the voltage, and if the voltage peak-to-peak features are not needed, the voltage peak-to-peak value does not need to be calculated during feature extraction, so that the updating and the improvement of the testing resource configuration and.
As shown in fig. 3, an automatic testing system for electronic devices based on machine learning according to the present disclosure includes an automatic testing front end and a server end.
The automatic test front end comprises a data visualization and man-machine interaction module.
The server side is provided with a database, and the test data is stored in the database.
The server side includes:
(1) the characteristic classification extraction module is used for acquiring test data of the electronic equipment and extracting characteristic components of the test data; the test data includes serialized waveform data and nonserialized video and image data.
In the disclosure, different feature acquisition methods are specifically proposed according to different data types. Different sensors or instrument equipment are required to be adopted for acquiring test data of the electronic equipment, and the acquired data types are various and include but are not limited to serialized time domain waveform data, non-serialized image data and the like.
In order to meet the input requirements of machine learning, different features of input data need to be extracted to form feature vectors, the dimensions of the feature vectors are the dimensions of the machine learning input, and the components of the feature vectors are different features of the machine learning input.
The input to machine learning, whether for serialized or non-serialized data, is a feature vector, i.e., a set of vectors in the mathematical sense, the dimensions of the vector representing the number of input features, and each component of the vector representing a particular feature. For example, for feature extraction of serialized data, the input feature vector is { current amplitude, voltage amplitude, temperature value, total energy of vibration signal }, the machine learning input dimension is 4, and each component of the feature vector represents a corresponding physical index.
For the feature extraction of the non-serialized data, taking image data as an example, the input feature vector is a pixel matrix, a two-dimensional matrix is obtained through reconstruction, and then the further compression of the pixel matrix can be realized by utilizing convolution mathematical operation, namely the feature extraction of the non-serialized data.
Most of the results after electronic test sampling are serialized data in the form of a set of data vectors, each component of the set of vectors being a multi-dimensional array of samples taken at one time. For such data, the following feature component determination methods are proposed:
solving the sum of squares of the amplitudes of each component array of a group of data vectors obtained after test sampling to obtain the total energy value characteristic;
denoising, and then obtaining the denoised peak-to-peak value characteristic by taking the difference between the maximum value and the minimum value in the component array;
giving a certain amplitude threshold value, and then solving the average value of all the amplitude values which are higher than the threshold value in the component array to obtain the amplitude characteristic calculated based on the threshold value;
and giving a certain step length, and solving the sum of squares of the amplitude values at intervals of the certain step length in the component array to obtain the energy value characteristic based on the certain step length.
(2) And the feature vector generation module is used for performing principal component analysis on the test data, determining the correlation of different feature components and further generating feature vectors.
For automatic test of electronic equipment, firstly test data is obtained through data acquisition, then each characteristic component is obtained after data processing such as cleaning, denoising, calculation of various characteristic components and the like is carried out on the test data, then principal component analysis is carried out on each characteristic component to determine the correlation of different characteristic components, so that redundant and weak correlation characteristics are obtained, and finally the redundant and weak correlation characteristics are removed from a characteristic vector, so that the composition of the characteristic vector is determined.
(3) And the prediction result output module is used for inputting the characteristic vector into a preset fault diagnosis model and outputting the prediction result of the electronic test fault.
Specifically, a machine learning algorithm in the fault diagnosis model is preset, wherein the machine learning algorithm can be a convolutional neural network or other existing neural network algorithms.
The preset fault diagnosis model can be trained by adopting the historical characteristic vectors until the accuracy of the output prediction result of the electronic test fault meets the preset accuracy requirement, and then the training of the preset fault diagnosis model is completed.
In another embodiment, the server further includes:
and the characteristic vector analysis result feedback module is used for outputting the prediction result of the electronic test fault and returning the analysis result of the characteristic vector to the automatic test front end to guide the front end to test resource allocation and the updating and improvement of the process.
Specifically, automatic test resource configuration and flow are determined first, and data features as complete as possible are collected according to previous experience or current requirements.
Specifically, according to the collected data features, feature vectors are obtained through the above-mentioned feature extraction method, and then a principal component analysis method is applied to determine redundant and weakly correlated features. For example, setting the cumulative feature contribution rate to reach 80% in advance, and after giving a feature component set { x1, x2, x3, x4, x5, x6} (where x1 to x6 may respectively represent various features such as voltage amplitude, current amplitude, temperature, total vibration energy value, current peak-to-peak value, voltage peak-to-peak value, etc.), applying a principal component analysis method to obtain a feature component ranking as: x3, x4, x2, x5, x1, x 6; if the cumulative contribution rate of the first four principal components x3, x4, x2 and x5 reaches 80%, the final feature vector is determined to be { x2, x3, x4 and x5}, the redundancy and weak correlation features are the fifth principal component and the sixth principal component { x1 and x6}, the feature which has a large contribution effect on the final machine learning model is { x2, x3, x4 and x5}, namely the current amplitude, the temperature, the total vibration energy value and the current peak-to-peak value, and the feature which has a small contribution effect is { x1, x6}, namely the voltage amplitude and the voltage peak-to-peak value.
And then converting the acquired data information into a feature vector, applying the feature vector to a background machine learning algorithm, and determining whether the model effect and the precision under the current feature set meet the test requirements or not by a principal component analysis method and a model result hypothesis test method, namely which features are redundant and weakly related features.
Training the extracted feature vectors { x2, x3, x4 and x5} by applying a machine learning algorithm to obtain a training result, carrying out model accuracy hypothesis test on the training result, setting a confidence coefficient and an accuracy rate in advance, and giving an original hypothesis (for example, "under 85% of confidence coefficient, the model accuracy rate is more than or equal to 80%"). If the model test does not pass the original hypothesis, the model effect is considered to be poor, a fifth principal component x1 is added into the eigenvector { x2, x3, x4, x5} again, the eigenvector is updated to { x2, x3, x4, x5, x1}, and the redundant and weakly correlated characteristic is determined to be x6, namely the voltage peak value; and if the model passes the original hypothesis through the model inspection, the model effect is considered to meet the requirement, the characteristic vectors are determined to be { x2, x3, x4 and x5}, the characteristic vectors do not need to be updated, and the redundant and weakly correlated characteristics are determined to be x1 and x6, namely the voltage amplitude and the voltage peak-to-peak value.
And finally, feeding back the information of the redundant features to an automatic test front end, optimizing resource configuration in the automatic test resource configuration and process, and eliminating test steps related to the redundant features, so that the test resource configuration and process updating and improvement are completed, and the subsequent test is more accurate and efficient.
Finally, the information of the redundant features is fed back to the automatic testing front end, namely, the testing front end is informed of which features do not need to be acquired or which features do not need to be extracted, the resource configuration is optimized in the automatic testing resource configuration and process, and the testing steps related to the redundant features are eliminated, namely, for the features which do not need to be acquired, the automatic testing can eliminate the acquisition process and the testing resources (for example, if the voltage features do not need to be acquired, the testing front end does not need to use voltage acquisition equipment, and does not need to design and arrange the testing process of the voltage, and if the voltage peak-to-peak features are not needed, the voltage peak-to-peak value does not need to be calculated during feature extraction, so that the updating and the improvement of the testing resource configuration and.
As will be appreciated by one skilled in the art, embodiments of the present disclosure may be provided as a method, system, or computer program product. Accordingly, the present disclosure may take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present disclosure may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, and the like) having computer-usable program code embodied therein.
The present disclosure is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a server side of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the server side of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.
Although the present disclosure has been described with reference to specific embodiments, it should be understood that the scope of the present disclosure is not limited thereto, and those skilled in the art will appreciate that various modifications and changes can be made without departing from the spirit and scope of the present disclosure.
Claims (7)
1. An electronic device automatic testing method based on machine learning is characterized by comprising the following steps:
acquiring test data of the electronic equipment, and extracting characteristic components of the test data; the test data includes serialized waveform data and non-serialized video and image data;
performing principal component analysis on the test data, determining the correlation of different feature components, and further generating a feature vector;
inputting the characteristic vector into a preset fault diagnosis model, and outputting a prediction result of the electronic test fault;
in the process of extracting the feature components of the test data, for feature extraction of the serialized data:
solving the sum of squares of the amplitudes of each component array of a group of data vectors obtained after test sampling to obtain the total energy value characteristic; denoising, and then obtaining the denoised peak-to-peak value characteristic by taking the difference between the maximum value and the minimum value in the component array; giving a certain amplitude threshold value, and then solving the average value of all the amplitude values which are higher than the threshold value in the component array to obtain the amplitude characteristic calculated based on the threshold value; a certain step length is given, and the square sum of the amplitude values at intervals of the certain step length in the component array is obtained and is the energy value characteristic based on the certain step length;
the automatic test method of the electronic equipment based on the machine learning further comprises the following steps:
while outputting the prediction result of the electronic test fault, returning the analysis result of the feature vector to the automatic test front end to guide the front end to test the resource allocation and the updating and improvement of the process;
firstly, determining automatic test resource allocation and flow, and acquiring data characteristics as complete as possible according to previous experience or current requirements;
then converting the acquired data information into a feature vector, applying the feature vector to a background machine learning algorithm, and determining whether the model effect and precision under the current feature set meet the test requirements or not by a principal component analysis method and a model result hypothesis test method, namely which features are redundant and weakly related features;
and finally, feeding back the information of the redundant features to an automatic test front end, optimizing resource configuration in the automatic test resource configuration and process, and eliminating test steps related to the redundant features, so that the test resource configuration and process updating and improvement are completed, and the subsequent test is more accurate and efficient.
2. The method of claim 1, wherein the analysis result of the feature vector comprises redundant feature information.
3. The method for automatically testing electronic equipment based on machine learning of claim 1, wherein in the process of extracting the feature components of the test data, for the feature extraction of the nonserialized image data:
the input feature vector is a pixel point matrix, a two-dimensional matrix is obtained through reconstruction, then the pixel point matrix is further compressed by utilizing convolution mathematical operation, and feature extraction of non-serialized image data is obtained.
4. An automatic test system of electronic equipment based on machine learning is characterized by comprising an automatic test front end and a server end, wherein the server end comprises:
the characteristic classification extraction module is used for acquiring test data of the electronic equipment and extracting characteristic components of the test data; the test data includes serialized waveform data and non-serialized video and image data;
the feature vector generation module is used for performing principal component analysis on the test data, determining the correlation of different feature components and further generating feature vectors;
the prediction result output module is used for inputting the characteristic vector into a preset fault diagnosis model and outputting a prediction result of the electronic test fault;
the server side further comprises:
the characteristic vector analysis result feedback module is used for outputting the prediction result of the electronic test fault and simultaneously returning the analysis result of the characteristic vector to the automatic test front end to guide the front end to test the resource allocation and the updating and the improvement of the process, particularly,
firstly, determining automatic test resource allocation and flow, and acquiring data characteristics as complete as possible according to previous experience or current requirements;
then converting the acquired data information into a feature vector, applying the feature vector to a background machine learning algorithm, and determining whether the model effect and precision under the current feature set meet the test requirements or not by a principal component analysis method and a model result hypothesis test method, namely which features are redundant and weakly related features;
and finally, feeding back the information of the redundant features to an automatic test front end, optimizing resource configuration in the automatic test resource configuration and process, and eliminating test steps related to the redundant features, so that the test resource configuration and process updating and improvement are completed, and the subsequent test is more accurate and efficient.
5. The system of claim 4, wherein the analysis of the feature vectors includes redundant feature information.
6. The system of claim 4, wherein the feature classification extraction module extracts, for the feature extraction of the serialized data:
solving the sum of squares of the amplitudes of each component array of a group of data vectors obtained after test sampling to obtain the total energy value characteristic;
denoising, and then obtaining the denoised peak-to-peak value characteristic by taking the difference between the maximum value and the minimum value in the component array;
giving a certain amplitude threshold value, and then solving the average value of all the amplitude values which are higher than the threshold value in the component array to obtain the amplitude characteristic calculated based on the threshold value;
and giving a certain step length, and solving the sum of squares of the amplitude values at intervals of the certain step length in the component array to obtain the energy value characteristic based on the certain step length.
7. The system of claim 4, wherein the feature classification extraction module extracts features of the nonserialized image data by:
the input feature vector is a pixel point matrix, a two-dimensional matrix is obtained through reconstruction, then the pixel point matrix is further compressed by utilizing convolution mathematical operation, and feature extraction of non-serialized image data is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910012589.XA CN109815855B (en) | 2019-01-07 | 2019-01-07 | Electronic equipment automatic test method and system based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910012589.XA CN109815855B (en) | 2019-01-07 | 2019-01-07 | Electronic equipment automatic test method and system based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109815855A CN109815855A (en) | 2019-05-28 |
| CN109815855B true CN109815855B (en) | 2021-04-02 |
Family
ID=66604031
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910012589.XA Active CN109815855B (en) | 2019-01-07 | 2019-01-07 | Electronic equipment automatic test method and system based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109815855B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111507393B (en) * | 2020-04-14 | 2021-11-09 | 艾瑞思检测技术(苏州)有限公司 | Display card interface machine testing method based on Laplace feature mapping learning |
| CN113963456A (en) * | 2020-07-20 | 2022-01-21 | 核工业理化工程研究院 | Method and system for analyzing operation data of multiple high-speed rotating devices |
| CN113029592A (en) * | 2021-03-11 | 2021-06-25 | 奇瑞新能源汽车股份有限公司 | Electric automobile testing method and device |
| CN114689975B (en) * | 2022-04-01 | 2023-03-24 | 深圳市华检测试技术服务有限公司 | Harmonic current-based product testing method, device, equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106650942A (en) * | 2015-11-02 | 2017-05-10 | 中国船舶工业系统工程研究院 | Fault diagnosis method based on electronic equipment testability model |
| CN108073158A (en) * | 2017-12-05 | 2018-05-25 | 上海电机学院 | Based on PCA and KNN density algorithm Wind turbines Method for Bearing Fault Diagnosis |
| CN108320040A (en) * | 2017-01-17 | 2018-07-24 | 国网重庆市电力公司 | Acquisition terminal failure prediction method and system based on Bayesian network optimization algorithm |
| CN108985279A (en) * | 2018-08-28 | 2018-12-11 | 上海仁童电子科技有限公司 | The method for diagnosing faults and device of double-unit traction controller waveform |
| JP2018200677A (en) * | 2017-05-25 | 2018-12-20 | 株式会社日立製作所 | Management method, non-temporary computer-readable medium, and management apparatus |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103606221B (en) * | 2013-12-04 | 2016-01-20 | 广州广电运通金融电子股份有限公司 | Fault automatic diagnostic method of counter and device |
-
2019
- 2019-01-07 CN CN201910012589.XA patent/CN109815855B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106650942A (en) * | 2015-11-02 | 2017-05-10 | 中国船舶工业系统工程研究院 | Fault diagnosis method based on electronic equipment testability model |
| CN108320040A (en) * | 2017-01-17 | 2018-07-24 | 国网重庆市电力公司 | Acquisition terminal failure prediction method and system based on Bayesian network optimization algorithm |
| JP2018200677A (en) * | 2017-05-25 | 2018-12-20 | 株式会社日立製作所 | Management method, non-temporary computer-readable medium, and management apparatus |
| CN108073158A (en) * | 2017-12-05 | 2018-05-25 | 上海电机学院 | Based on PCA and KNN density algorithm Wind turbines Method for Bearing Fault Diagnosis |
| CN108985279A (en) * | 2018-08-28 | 2018-12-11 | 上海仁童电子科技有限公司 | The method for diagnosing faults and device of double-unit traction controller waveform |
Non-Patent Citations (1)
| Title |
|---|
| 新形势下电子测试技术的需求与;吴乐 等;《科技创新与应用》;20130731(第13期);116 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109815855A (en) | 2019-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109815855B (en) | Electronic equipment automatic test method and system based on machine learning | |
| CN110352660B (en) | Method and device for processing information of rapid nondestructive testing of vitality of delinted cotton seeds | |
| CN112084974B (en) | Multi-label rolling bearing fault diagnosis method based on meta-learning | |
| CN109685314B (en) | Non-intrusive load decomposition method and system based on long-term and short-term memory network | |
| CN108491931B (en) | Method for improving nondestructive testing precision based on machine learning | |
| CN112529234A (en) | Surface water quality prediction method based on deep learning | |
| CN111650453A (en) | Power equipment diagnosis method and system based on windowing characteristic Hilbert imaging | |
| CN116089870A (en) | Industrial equipment fault prediction method and device based on meta-learning under small sample condition | |
| CN112541510A (en) | Intelligent fault diagnosis method based on multi-channel time series data | |
| CN115438726A (en) | Device life and fault type prediction method and system based on digital twin technology | |
| CN115730947A (en) | Bank customer loss prediction method and device | |
| CN112947080B (en) | Scene parameter transformation-based intelligent decision model performance evaluation system | |
| CN110320802B (en) | Complex system signal time sequence identification method based on data visualization | |
| CN115184193B (en) | Automatic testing method, device, equipment and storage medium for cable | |
| CN111415326A (en) | Method and system for detecting abnormal state of railway contact net bolt | |
| CN111241629A (en) | Intelligent prediction method for performance change trend of airplane hydraulic pump based on data driving | |
| Schierwagen | Neuronal Morphology: Shape Characteristics and Models: Proceedings of the International School “Problems of Experimental, Clinical, and Theoretical Neurosciences”(Dnepropetrovsk, Ukraine, May 2–4, 2008) | |
| CN114742122A (en) | Equipment fault diagnosis method and device, electronic equipment and storage medium | |
| CN108074240A (en) | Recognition methods, identification device, computer readable storage medium and program product | |
| CN111241725B (en) | Structure response reconstruction method for generating countermeasure network based on conditions | |
| CN113139332A (en) | Automatic model construction method, device and equipment | |
| CN114295967A (en) | Analog circuit fault diagnosis method based on migration neural network | |
| CN112215351A (en) | Enhanced multi-scale convolution neural network soft measurement method | |
| CN111310907A (en) | Microwave assembly fault diagnosis method, device and equipment | |
| CN112834194B (en) | Fault intelligent detection method based on soft target measurement under fault-free sample |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |