Analog circuit switch control method with fault self-checking function

Abstract

The invention discloses an analog circuit switch control method with a fault self-checking function, and relates to the technical field of electric switches, and the method comprises the steps: dynamically monitoring and obtaining circuit state information through a built-in circuit monitor in an analog circuit. And calling a circuit feature extraction strategy, performing feature extraction analysis on the circuit state information, and generating a circuit feature vector. And reading a predetermined feature vector, comparing the circuit feature vector with the predetermined feature vector, and calculating a triangular distance value. And performing normalization processing on the triangular distance value to obtain a degradation state index, and sending a first fault early warning signal when the degradation state index does not meet a predetermined index threshold. And executing action control on a circuit switch of the analog circuit based on the first fault early warning signal so as to realize early warning and response of the circuit fault. Therefore, the technical effects of reducing the calculation amount, reducing the data dependence and providing a clear and explainable diagnosis result are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical switches, and particularly to a control method for an analog circuit switch with a fault self-checking function. Background Art

[0002] In the prior art, the fault detection of traditional analog circuits usually relies on static monitoring or regular detection methods. By collecting parameters such as voltage and current of the circuit for simple threshold judgment, the calculation amount is large when facing complex circuits, and it is affected by component tolerances, and the diagnosis result is fuzzy. And the intelligent method based on the neural network method requires a large amount of fault sample data, and the diagnosis process has the "black box" property and is difficult to explain. Summary of the Invention

[0003] The present invention provides a control method for an analog circuit switch with a fault self-checking function to solve the technical problems of large diagnosis calculation amount, high data requirement, and unclear diagnosis result in the prior art, and achieve the technical effects of reducing the calculation amount, reducing data dependence, and providing a clear and interpretable diagnosis result.

[0004] The control method for an analog circuit switch with a fault self-checking function provided by the present invention includes: Dynamically monitoring the circuit state information through a built-in circuit monitor in the analog circuit.

[0005] Invoking a circuit feature extraction strategy to perform feature extraction and analysis on the circuit state information to obtain a circuit feature vector.

[0006] Reading a predetermined feature vector, and comparing the circuit feature vector with the predetermined feature vector to obtain a triangular distance value.

[0007] Obtaining a degradation state index by normalizing the triangular distance value, and sending a first fault warning signal when the degradation state index does not meet a predetermined index threshold.

[0008] Based on the first fault warning signal, controlling the operation of the circuit switch of the analog circuit.

[0009] In a feasible implementation manner, invoking a circuit feature extraction strategy to perform feature extraction and analysis on the circuit state information to obtain a circuit feature vector includes: Extracting the kernel principal component strategy in the circuit feature extraction strategy.

[0010] According to the kernel principal component strategy, extracting the global feature vector of the circuit state information.

[0011] Extracting the locally linear embedding strategy in the circuit feature extraction strategy.

[0012] Extract the local feature vector of the circuit state information according to the local linear embedding strategy.

[0013] Extract the wavelet packet decomposition strategy in the circuit feature extraction strategy.

[0014] Extract the time-frequency feature vector of the circuit state information according to the wavelet packet decomposition strategy.

[0015] Construct the circuit feature vector based on the global feature vector, the local feature vector and the time-frequency feature vector.

[0016] In a feasible implementation manner, extracting the global feature vector of the circuit state information according to the kernel principal component strategy includes: Obtain the circuit state normalization information of the circuit state information through normalization processing according to the kernel principal component strategy.

[0017] Read a predetermined kernel function, and obtain the kernel matrix of the circuit state normalization information based on the predetermined kernel function.

[0018] Obtain the decomposition result of the kernel matrix, where the decomposition result includes multiple feature vectors with eigenvalues.

[0019] Arrange the feature vectors based on the descending order of the eigenvalues to obtain a feature vector list.

[0020] Read a predetermined ranking threshold, and take the feature vectors of the predetermined ranking threshold in the feature vector list to form the global feature vector.

[0021] In a feasible implementation manner, extracting the local feature vector of the circuit state information according to the local linear embedding strategy includes: Arbitrarily extract the first information in the circuit state information.

[0022] Construct the first neighborhood of the first information according to the local linear embedding strategy, where the first neighborhood includes multiple nearest neighbor information.

[0023] Read the weight matrix, and obtain the first point linear representation of the first information in combination with the multiple nearest neighbor information.

[0024] Construct the local feature vector based on the first point linear representation.

[0025] In a feasible implementation manner, extracting the time-frequency feature vector of the circuit state information according to the wavelet packet decomposition strategy includes: Perform hierarchical decomposition on the circuit state information according to the wavelet packet decomposition strategy to obtain a hierarchical decomposition result.

[0026] Extract the first sub-signal corresponding to the first frequency band in the hierarchical decomposition result, and obtain the first energy spectrum of the first sub-signal.

[0027] Form the time-frequency feature vector based on the correspondence between the first frequency band and the first energy spectrum.

[0028] In a feasible implementation manner, reading a predetermined feature vector and comparing the circuit feature vector with the predetermined feature vector to obtain a triangular distance value includes: Read a distance calculation function, and perform a comparison calculation on the circuit feature vector and the predetermined feature vector according to the distance calculation function to obtain the triangular distance value.

[0029] The expression of the distance calculation function is: .

[0030] Where, refers to the triangular distance value, refers to the weight coefficient of the th feature component in the circuit feature vector and the predetermined feature vector at time , refers to the component value of the th feature component in the circuit feature vector at time , refers to the component value of the th feature component in the predetermined feature vector at time .

[0031] In a feasible implementation manner, after invoking a circuit feature extraction strategy to perform feature extraction and analysis on the circuit state information to obtain a circuit feature vector, it further includes: Construct a fault database for the same type of circuits of the analog circuit, where the fault database includes multiple circuit feature vectors with fault type identifiers.

[0032] Match the target feature vector corresponding to the circuit feature vector among the multiple circuit feature vectors with fault type identifiers.

[0033] Use the target fault type corresponding to the target feature vector as the real-time predicted fault of the analog circuit.

[0034] Send out a second fault warning signal, and perform a fault warning for the real-time predicted fault on the analog circuit based on the second fault warning signal.

[0035] In a feasible implementation, after sending a second fault warning signal and performing fault warning of real-time predicted faults on the analog circuit based on the second fault warning signal, the method further includes: activating a backup channel based on the second fault warning signal and supporting the operation of the analog circuit through the backup channel.

[0036] The present invention discloses a method for controlling an analog circuit switch with a fault self-checking function, including: dynamically monitoring circuit state information by using a circuit monitor built in the analog circuit, and calling a circuit feature extraction strategy to perform feature extraction and analysis on the circuit state information to generate a circuit feature vector; reading a predetermined feature vector and calculating a triangular distance value by comparing it with the circuit feature vector; performing normalization processing on the triangular distance value to obtain a degradation state index, and if the index does not reach a predetermined threshold, triggering a first fault warning signal; based on the warning signal, performing corresponding action control on the circuit switch of the analog circuit. The method for controlling an analog circuit switch with a fault self-checking function disclosed by the present invention solves the technical problems of large diagnostic calculation amount, high data requirement, and unclear diagnostic result, and realizes the technical effects of reducing the calculation amount, reducing data dependence, and providing a clear and interpretable diagnostic result. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic flow chart of the method for controlling an analog circuit switch with a fault self-checking function according to the present invention.

[0038] Figure 2 It is a schematic flow chart of performing fault warning in the method for controlling an analog circuit switch with a fault self-checking function according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] The above technical solutions will be described in detail below in combination with the accompanying drawings of the specification and specific embodiments to better understand the above technical solutions. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments used to explain the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts belong to the scope of protection of the present invention. In addition, it should be noted that only the parts related to the present invention are shown in the drawings for the convenience of description, rather than all of them.

[0040] Embodiment Figure 1 It is a schematic flow chart of the method for controlling an analog circuit switch with a fault self-checking function according to the present invention, wherein the method for controlling an analog circuit switch with a fault self-checking function includes: S100: Dynamically monitor circuit state information through a circuit monitor built in the analog circuit.

[0041] Specifically, first, dynamic monitoring of the power state is performed based on the circuit monitor embedded in the target analog circuit, and the corresponding circuit state information is collected. Among them, the power monitor may include a variety of sensors, an analog-to-digital converter (ADC), and a data processing unit. The various sensors are used to obtain parameters such as the voltage and current of the circuit respectively, and convert them into digital signals through the analog-to-digital converter for subsequent processing.

[0042] Exemplarily, in the analog circuit, the built-in circuit monitor collects parameters such as the voltage and current of the circuit in real time according to a preset time interval or sampling resolution through sensors, converts them into digital signals by the analog-to-digital converter, and stores them in the buffer after basic data processing (such as data filtering, sampling, etc.) by the data processing unit.

[0043] In the whole solution, the above-mentioned dynamic monitoring is the basis of the entire fault detection process, ensuring that the subsequent steps can be analyzed based on the latest circuit state information. Through dynamic monitoring, abnormal changes in the circuit state can be captured in a timely manner, providing reliable data support for real-time fault detection and early warning.

[0044] S200: Invoke the circuit feature extraction strategy to perform feature extraction and analysis on the circuit state information to obtain a circuit feature vector.

[0045] Specifically, the circuit feature extraction strategy defines the way of performing feature extraction on the circuit state information. In other words, the circuit feature extraction strategy is used to determine which feature index items to extract from the circuit state information and defines the extraction method for each feature index item.

[0046] Exemplarily, the strategy usually includes techniques such as kernel principal component analysis (KPCA), local linear embedding (LLE), and wavelet packet decomposition (WPD); key features that can reflect the circuit health state are extracted from the original circuit state information through the above strategy and combined into a vector, which is called a circuit feature vector.

[0047] Through feature extraction, the original high-dimensional circuit state information can be transformed into a low-dimensional feature vector, providing efficient and accurate data support for subsequent fault detection.

[0048] In some embodiments, invoking the circuit feature extraction strategy to perform feature extraction and analysis on the circuit state information to obtain a circuit feature vector includes: Extract the kernel principal component strategy in the circuit feature extraction strategy; extract the global feature vector of the circuit state information according to the kernel principal component strategy; extract the locally linear embedding strategy in the circuit feature extraction strategy; extract the local feature vector of the circuit state information according to the locally linear embedding strategy; extract the wavelet packet decomposition strategy in the circuit feature extraction strategy; extract the time-frequency feature vector of the circuit state information according to the wavelet packet decomposition strategy; based on the global feature vector, the local feature vector and the time-frequency feature vector, construct the circuit feature vector.

[0049] Specifically, the kernel principal component strategy is a non-linear feature extraction method that maps the original data to a high-dimensional space through a kernel function to extract global features. Among them, the kernel function is used to calculate the similarity of data points in the high-dimensional space, and the calculation result of the kernel function is represented in the form of a matrix called the kernel matrix. Eigenvalue decomposition is the process of decomposing the kernel matrix into eigenvalues and eigenvectors for extracting the most important features.

[0050] Specifically, the locally linear embedding strategy is a non-linear dimensionality reduction method that extracts local features by constructing the local neighborhood relationship of data points. Among them, the neighborhood relationship refers to the local similarity between data points, and the weight matrix represents the linear representation of data points in the local neighborhood.

[0051] Specifically, wavelet packet decomposition is a time-frequency analysis method that decomposes a signal into sub-signals of different frequency bands by performing hierarchical decomposition on the signal to extract time-frequency features, and the time-frequency features characterize the energy features of different frequency bands.

[0052] Specifically, to obtain the global feature vector of the circuit state information, first, standardize the circuit state information to eliminate the influence of dimension, and obtain the standardized circuit state information; read the predetermined kernel function, and calculate the kernel matrix of the standardized circuit state information based on the kernel function; perform eigenvalue decomposition on the kernel matrix to obtain multiple eigenvectors and their corresponding eigenvalues; arrange the eigenvectors in descending order of eigenvalues, and take the first several eigenvectors to form the global feature vector.

[0053] Specifically, to obtain the local feature vector of the circuit state information, first, extract a data point from the circuit state information and construct the neighborhood of this data point, and the neighborhood includes multiple nearest neighbor data points; then, read the weight matrix, calculate the linear representation (low-dimensional embedding representation) of this data point in combination with the nearest neighbor data points, and finally extract the features representing the local pattern changes based on the linear representation to form the local feature vector.

[0054] Specifically, to obtain the time-frequency feature vector of the circuit state information, first, select an appropriate wavelet basis function (such as the Daubechies wavelet) to perform multi-layer wavelet packet decomposition on the circuit signal to obtain sub-signals in different frequency bands; then, calculate the energy spectra of the sub-signals in different frequency bands, and form a time-frequency feature vector based on the correspondence between the frequency bands and the energy spectra.

[0055] Finally, through feature splicing, weighted fusion or dimensionality reduction methods, combine the global feature vector, local feature vector and time-frequency feature vector into a comprehensive circuit feature vector. In the whole scheme, the role of the above steps is to transform the original high-dimensional circuit state information into a low-dimensional feature vector, so as to provide efficient and accurate data support for subsequent fault detection.

[0056] In some implementation manners, obtaining the global feature vector of the circuit state information according to the kernel principal component strategy includes: Obtaining the circuit state normalization information of the circuit state information through normalization processing according to the kernel principal component strategy; reading a predetermined kernel function, and obtaining a kernel matrix of the circuit state normalization information based on the predetermined kernel function; decomposing to obtain a decomposition result of the kernel matrix, where the decomposition result includes multiple eigenvectors with eigenvalue identifiers; arranging the eigenvectors in descending order of the eigenvalues to obtain an eigenvector list; reading a predetermined ranking threshold, and taking the eigenvectors of the predetermined ranking threshold of the eigenvector list to form the global feature vector.

[0057] Specifically, a kernel function is a mathematical function used to map original data into a high-dimensional space for linear processing in the high-dimensional space. Common kernel functions include Gaussian kernels, polynomial kernels, etc. The kernel matrix is the matrix form of the calculation result of the kernel function, representing the similarity of data points in the high-dimensional space. The elements of the kernel matrix are the values obtained by calculating two data points through the kernel function.

[0058] Specifically, first, the circuit state information is standardized to obtain standardized circuit state information. This step ensures that the data has zero mean and unit variance by subtracting the mean and dividing by the standard deviation. Then, a predetermined kernel function is read, and the kernel matrix of the standardized circuit state information is calculated based on this kernel function. Next, the kernel matrix is subjected to eigenvalue decomposition to obtain multiple eigenvectors and their corresponding eigenvalues. Among them, the eigenvalue represents the scaling factor of the matrix in the direction of the corresponding eigenvector, that is, the importance of the eigenvector, and the eigenvector represents the main change direction of the matrix. Then, the eigenvectors are arranged in descending order of eigenvalues to obtain a list of eigenvectors. Among them, the larger the eigenvalue of an eigenvector, the greater the change in the data in the corresponding direction (the larger the eigenvalue, the more important it is). This step is used to ensure that the most important eigenvectors are ranked first. Further, a predetermined ranking threshold is read, and the eigenvectors ranked within the predetermined ranking threshold in the list of eigenvectors are extracted to form a global eigenvector. This global eigenvector retains the main global features of the circuit state information while achieving information reduction.

[0059] The above steps extract the global eigenvector through the kernel principal component strategy, which can significantly reduce the data dimension and computational complexity. Among them, the standardization process ensures the consistency of the data, the kernel matrix calculation and eigenvalue decomposition extract the main change directions of the data, and the eigenvector sorting and global eigenvector formation ensure that the extracted eigenvectors have the highest information content. This method not only reduces the dependence on a large amount of fault sample data but also provides clear and interpretable diagnostic results, providing efficient and accurate data support for subsequent fault detection and early warning, and ensuring the real-time performance and reliability of the entire fault detection scheme.

[0060] In some implementation manners, extracting the local eigenvector of the circuit state information according to the local linear embedding strategy includes: Arbitrarily extract the first information from the circuit state information; construct the first neighborhood of the first information according to the local linear embedding strategy, where the first neighborhood includes multiple nearest neighbor information; read the weight matrix and combine the multiple nearest neighbor information to obtain the first point linear representation of the first information; and form the local eigenvector based on the first point linear representation.

[0061] Specifically, in the local linear embedding strategy, the first information refers to an arbitrarily extracted data point from the circuit state information, which is used to construct the neighborhood relationship. The first neighborhood refers to multiple nearest neighbor data points around the first information (data point). These nearest neighbor data points are used to construct the local linear representation.

[0062] Specifically, the weight matrix is used to represent the linear representation relationship of data points in the local neighborhood. The weight matrix of each data point reflects the contribution degree of other data points in its neighborhood to this data point. The first linear representation is calculated through the weight matrix and the nearest neighbor information, and is used to reflect the linear representation relationship of the first information in the local neighborhood.

[0063] Specifically, first, arbitrarily extract a data point from the circuit state information as the starting point for analysis; according to the local linear embedding strategy, find multiple nearest neighbor data points of this data point to construct the first neighborhood, where the size of the neighborhood can be adjusted according to the specific application scenario and data characteristics; then, read the pre-computed weight matrix, which represents the linear representation relationship of data points in the local neighborhood. Furthermore, combine the nearest neighbor information and the weight matrix, and determine the linear representation coefficients (i.e., linear representations) of the data points in the neighborhood through mathematical operations (such as the least squares method); finally, combine the linear representations of all data points into a local feature vector, which reflects the local features of the circuit state information and is used for subsequent fault detection and analysis.

[0064] Extracting the local feature vector through the local linear embedding strategy can significantly improve the accuracy of fault detection. Among them, the local feature vector captures the local changes of the circuit state, and combined with the global feature vector and the time-frequency feature vector, provides a more comprehensive description of the circuit state, not only reducing the computational amount, but also reducing the dependence on a large number of fault sample data.

[0065] In some implementation manners, extracting the time-frequency feature vector of the circuit state information according to the wavelet packet decomposition strategy includes: Performing hierarchical decomposition on the circuit state information according to the wavelet packet decomposition strategy to obtain a hierarchical decomposition result; extracting the first sub-signal corresponding to the first frequency band in the hierarchical decomposition result, and obtaining the first energy spectrum of the first sub-signal; forming the time-frequency feature vector based on the corresponding relationship between the first frequency band and the first energy spectrum.

[0066] Specifically, first, select a suitable wavelet basis function (such as the Daubechies wavelet db4), and set the decomposition layer number L, and perform L-layer wavelet packet decomposition on the circuit state signal. Exemplarily, assuming that the sampling frequency of the circuit state signal X(t) is 8 kHz, perform 3-layer wavelet packet decomposition on it to obtain 8 sub-signals: ; Among them, each sub-signal corresponds to a frequency band: ; Then, select the first frequency band and obtain the time-domain sequence S30 of this sub-signal. Exemplarily: ; Calculate the energy of the selected sub-signal: ; Wherein, is the energy of the sub-signal corresponding to the first frequency band (i.e., 0, 500 Hz); n is the sequence depth of the time-domain sequence, that is, the number of data included. For the time-domain sequence S30 of the first frequency band, n = 6; are the n sequence samples in the time-domain sequence S30 of the first frequency band, which characterize the energy distribution of the signal within the first frequency band range.

[0067] Finally, based on the correspondence between the first frequency band and the first energy spectrum, a time-frequency feature vector is formed. The time-frequency feature vector includes energy and the corresponding frequency band information. Exemplarily, is the time-frequency feature vector corresponding to the frequency band information [0, 500] Hz and the energy 0.051.

[0068] The above method steps perform multi-layer decomposition on the circuit state information based on wavelet packet decomposition, extract sub-signals of specific frequency bands, calculate their energy spectra, and finally form a time-frequency feature vector. This feature vector can accurately describe the circuit state information and provide efficient data support for intelligent circuit monitoring and fault diagnosis.

[0069] S300: Read a predetermined feature vector, and compare the circuit feature vector with the predetermined feature vector to obtain a triangular distance value.

[0070] Specifically, the predetermined feature vector is a previously stored reference feature vector, which is usually used as a comparison and analysis standard to judge the deviation of the current circuit state. The triangular distance value is used to quantitatively measure the similarity and can be obtained by calculating the weighted Euclidean distance between the circuit feature vector and the predetermined feature vector.

[0071] In some embodiments, reading a predetermined feature vector and comparing the circuit feature vector with the predetermined feature vector to obtain a triangular distance value includes: Read a distance calculation function, and perform a comparison calculation on the circuit feature vector and the predetermined feature vector according to the distance calculation function to obtain the triangular distance value; the expression of the distance calculation function is: ; Wherein, refers to the triangular distance value, refers to the weight coefficient of the th feature component in the circuit feature vector and the predetermined feature vector at time , refers to the th feature component of the circuit feature vector at time The component value, refers to the th eigen-component of the said predetermined eigen-vector at time The component value.

[0072] Specifically, the weight coefficient is used to adjust the contribution of different eigen-components to the final distance value, usually dynamically adjusted according to time changes. The distance calculation function is used to calculate the weighted distance between two vectors.

[0073] Specifically, multiple eigen-vectors can be preset as reference vectors, such as eigen-vectors for normal state, circuit aging, short circuit, overload, etc., and the corresponding predetermined eigen-vector is selected as the comparison reference when analyzing the current circuit state. Then, using the weighted Euclidean distance formula as the distance calculation function, calculate the distance between two eigen-vectors. The calculation of the obtained triangular distance value takes into account the weight coefficient, component value, and time factor of each eigen-component; among them, a smaller triangular distance value indicates that the current circuit state is similar to the predetermined state; if it is larger, it means that the circuit state may be abnormal.

[0074] In the above method steps, through dynamic weighted calculation, the contribution of different features to the final judgment is made more reasonable. By setting different predetermined eigen-vectors, the detection of states such as short circuit, aging, and harmonic interference can be realized. By adopting a time-related weight adjustment mechanism, it can dynamically adapt to the changes in circuit operation, improve the detection stability, and thus achieve more accurate circuit state analysis.

[0075] S400: Obtain the degradation state index by normalizing the said triangular distance value, and when the degradation state index does not meet the predetermined index threshold, issue a first fault warning signal.

[0076] Specifically, the degradation state index is a scalar value representing the circuit health state obtained through normalization processing. The closer the value is to 0, the closer the circuit state is to the normal state; the closer the value is to 1, the closer the circuit state is to the fault state. The predetermined index threshold is a set critical value used to judge whether the circuit is in a fault state. When the degradation state index exceeds this threshold, it can be considered that there is a fault and a first fault warning signal is issued. This first fault warning signal is used to trigger the fault handling mechanism.

[0077] S500: Based on the said first fault warning signal, perform action control on the circuit switch of the analog circuit.

[0078] Specifically, based on the first fault warning signal, perform action control on the circuit switch of the analog circuit, including disconnecting the faulty circuit, switching to the standby circuit, or adjusting the circuit parameters. The action control based on the fault warning signal can timely isolate the faulty circuit or activate the standby channel, ensure the continuous operation of the circuit, and reduce the downtime and maintenance cost.

[0079] In some embodiments, as Figure 2 shown, after retrieving the circuit feature extraction strategy to perform feature extraction and analysis on the circuit state information to obtain a circuit feature vector, it further includes: Constructing a fault database for circuits of the same type as the analog circuit, the fault database including a plurality of circuit feature vectors with fault type identifiers; matching the target feature vector corresponding to the circuit feature vector among the plurality of circuit feature vectors with fault type identifiers; taking the target fault type corresponding to the target feature vector as the real-time predicted fault of the analog circuit; sending out a second fault warning signal, and performing fault warning of the real-time predicted fault on the analog circuit based on the second fault warning signal.

[0080] Specifically, the fault database is a database storing various fault types and their corresponding feature vectors, which is used as a benchmark for subsequent fault matching and prediction; the target feature vector refers to the feature vector most similar to the current circuit feature vector in the fault database, which is used to determine the fault type corresponding to the current circuit state. The second fault warning signal is a signal sent out after a fault is predicted in real time, which is used to trigger a fault warning mechanism to remind the user or the system to perform corresponding processing.

[0081] Specifically, first, collect historical fault data of circuits of the same type, extract the circuit feature vector in each fault state, and store it in the fault database, and each feature vector is marked with the corresponding fault type; then, match the extracted circuit feature vector with the feature vectors in the fault database, and use a similarity calculation method (such as Euclidean distance, cosine similarity, etc.) to determine the most similar target feature vector; then, according to the matching result, extract the fault type corresponding to the target feature vector as the real-time predicted fault of the current circuit, and send out a second fault warning signal to remind the user or trigger an automatic processing mechanism.

[0082] In the above method steps, the establishment of the fault database and the matching of the current feature vector with the feature vectors in the fault database help to more quickly and accurately predict the fault type of the circuit and send out a warning signal in time.

[0083] In some implementation manners, after sending out the second fault warning signal and performing fault warning of the real-time predicted fault on the analog circuit based on the second fault warning signal, it further includes: activating a standby channel based on the second fault warning signal, and supporting the operation of the analog circuit through the standby channel.

[0084] Specifically, the standby channel refers to a standby circuit used to replace the main circuit and continue running when the main circuit fails. The standby channel is usually in a standby state and will only be activated when the main circuit fails.

[0085] Furthermore, after the second fault warning signal is issued, the backup channel can be further activated to ensure the continuous operation of the circuit. The activation process of the backup channel includes switching circuit connections, initializing backup circuit parameters, etc. The activation of the backup channel ensures the continuous operation of the circuit after a fault occurs, improving the reliability and availability of the circuit system.

[0086] In summary, the analog circuit switch control method with a fault self-checking function provided by the present invention has the following technical effects: The circuit state information is dynamically monitored by the circuit monitor built in the analog circuit, and the circuit feature extraction strategy is called to perform feature extraction and analysis on it to generate a circuit feature vector; the predetermined feature vector is read and the triangular distance value is calculated by comparing it with the circuit feature vector; the triangular distance value is normalized to obtain a degradation state index. If the index does not reach the predetermined threshold, the first fault warning signal is triggered; based on this warning signal, corresponding action control is performed on the circuit switch of the analog circuit, thereby achieving the technical effects of reducing the calculation amount, reducing data dependence, and providing clear and interpretable diagnostic results.

[0087] It should be understood that the disclosed embodiments of the present invention and the above descriptions enable those skilled in the art to implement the present invention using the present invention. At the same time, the present invention is not limited to the above-mentioned part of the embodiments. It should be understood that those of ordinary skill in the art can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for controlling an analog circuit switch with a fault self-checking function, characterized in that: include: The circuit status information is obtained by dynamically monitoring the circuit monitor built into the analog circuit; Retrieving a circuit feature extraction strategy to perform feature extraction analysis on the circuit state information to obtain a circuit feature vector; Reading a predetermined feature vector, and comparing the circuit feature vector with the predetermined feature vector to obtain a triangular distance value; obtaining a degradation state index by normalizing the triangular distance value, and issuing a first fault warning signal when the degradation state index does not meet a predetermined index threshold; The circuit switch of the analog circuit is controlled based on the first fault warning signal.

2. The analog circuit switch control method with fault self-detection function as claimed in claim 1, characterized in that: The circuit feature extraction strategy is called to perform feature extraction and analysis on the circuit state information to obtain a circuit feature vector, including: Extracting the core principal component strategy in the circuit feature extraction strategy; Extracting a global feature vector of the circuit state information according to the core principal component strategy; Extracting a local linear embedding strategy in the circuit feature extraction strategy; Extracting a local feature vector of the circuit state information according to the local linear embedding strategy; Extracting the wavelet packet decomposition strategy in the circuit feature extraction strategy; Extracting the time-frequency feature vector of the circuit state information according to the wavelet packet decomposition strategy; The circuit feature vector is constructed based on the global feature vector, the local feature vector and the time-frequency feature vector.

3. The analog circuit switch control method with fault self-detection function as claimed in claim 2, characterized in that: Extracting a global feature vector of the circuit state information according to the core principal component strategy includes: The circuit state normalization information of the circuit state information is obtained by performing normalization processing according to the core principal component strategy; Reading a predetermined kernel function, and obtaining a kernel matrix of the circuit state normalization information based on the predetermined kernel function; Decomposing to obtain a decomposition result of the kernel matrix, wherein the decomposition result includes a plurality of eigenvectors having eigenvalue identifiers; Arranging the eigenvectors based on the descending order of the eigenvalues ​​to obtain a eigenvector list; A predetermined ranking threshold is read, and a feature vector of the predetermined ranking threshold of the feature vector list is taken to form the global feature vector.

4. The analog circuit switch control method with fault self-detection function as claimed in claim 2, characterized in that: Extracting a local feature vector of the circuit state information according to the local linear embedding strategy includes: arbitrarily extracting first information from the circuit state information; Constructing a first neighborhood of the first information according to the local linear embedding strategy, wherein the first neighborhood includes a plurality of nearest neighbor information; Reading a weight matrix, and combining the plurality of nearest neighbor information to obtain a first point linear representation of the first information; The local feature vector is constructed based on the first point linear representation.

5. The analog circuit switch control method with fault self-detection function as claimed in claim 2, characterized in that: Extracting the time-frequency feature vector of the circuit state information according to the wavelet packet decomposition strategy includes: Decomposing the circuit state information in layers according to the wavelet packet decomposition strategy to obtain a layered decomposition result; Extracting a first sub-signal corresponding to a first frequency band in the hierarchical decomposition result, and acquiring a first energy spectrum of the first sub-signal; The time-frequency feature vector is formed based on the corresponding relationship between the first frequency band and the first energy spectrum.

6. The analog circuit switch control method with fault self-detection function as claimed in claim 1, characterized in that: Reading a predetermined feature vector, and comparing the circuit feature vector with the predetermined feature vector to obtain a triangular distance value, including: Reading a distance calculation function, and performing a comparison calculation on the circuit characteristic vector and the predetermined characteristic vector according to the distance calculation function to obtain the triangular distance value; The expression of the distance calculation function is: ； in, refers to the trigonometric distance value, It refers to the first The characteristic components in time The weight coefficient of It refers to the first characteristic vector of the circuit The characteristic components in time The component value of It refers to the first The characteristic components in time Component value.

7. The analog circuit switch control method with fault self-detection function as claimed in claim 1, characterized in that: After calling the circuit feature extraction strategy to perform feature extraction and analysis on the circuit state information to obtain the circuit feature vector, the method further includes: Building a fault database of circuits of the same type as the simulated circuit, the fault database comprising a plurality of circuit feature vectors with identifications of fault types; Matching a target feature vector corresponding to the circuit feature vector among the plurality of circuit feature vectors having fault type identifiers; Using the target fault type corresponding to the target feature vector as the real-time predicted fault of the analog circuit; A second fault warning signal is issued, and based on the second fault warning signal, a fault warning of the real-time predicted fault is performed on the analog circuit.

8. The analog circuit switch control method with fault self-detection function as claimed in claim 7, characterized in that: After issuing a second fault warning signal and performing a fault warning of the real-time predicted fault on the analog circuit based on the second fault warning signal, it also includes: activating a backup channel based on the second fault warning signal, and supporting the operation of the analog circuit through the backup channel.

Images