Magnet quench detection method based on individual group independent optimization immune algorithm

Abstract

The application relates to the technical field of magnet quench detection, and discloses a magnet quench detection method based on individual group independent optimization immune algorithm. The method comprises the following steps: respectively pre-processing existing samples and to-be-detected samples; respectively extracting features of existing sample signal frames and to-be-detected sample signal frames; respectively standardizing feature vectors of the existing sample signal frames and the to-be-detected sample signal frames; initializing an antibody group to generate an initialized antibody group; iteratively evolving the initialized antibody group to obtain an iteratively evolved antibody group; group-optimizing and screening the iteratively evolved antibody group to obtain a screening result; repeating the steps of iteratively evolving and group-optimizing and screening until the antibody group no longer changes, and generating an optimized discrimination area according to the antibody group that no longer changes; and calculating distances between the standardized feature vector of the to-be-detected sample signal frame and all antibodies in the antibody group that no longer changes, and judging whether the magnet quenches or not according to the calculated distances.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of magnet quench detection, and particularly relates to a magnet quench detection method based on individual group independent optimization immune algorithm. BACKGROUND

[0002] The superconducting magnet is one of the core components of the power module of the maglev train, the electromagnetic sled and the like. The main principle is that the superconducting magnet after refrigeration and excitation enters the superconducting state and generates a strong magnetic field with little power loss. The system interacts with the ground magnetic field to form thrust, suspension force and the like to drive the body to move at high speed, which is one of the high-efficiency ways to replace the traditional internal combustion engine and motor drive.

[0003] The stability of the superconducting magnet in the superconducting state is one of the basic conditions for normal work. The maintenance of the superconducting state is related to many factors. The fluctuation of the related conditions in a large range will affect the working state of the superconducting magnet, thereby causing the superconducting magnet to lose the superconducting state, that is, the magnet quench.

[0004] The current magnet quench detection methods mainly include threshold method and data analysis method. The main process of the threshold method is to summarize the quench threshold according to the critical parameters or historical data of the superconducting magnet, and to judge whether the magnet quench occurs by using the threshold to judge the parameters of the superconducting magnet. The data analysis method mainly uses the time series data of the superconducting magnet, such as voltage signal, temperature signal and field strength signal, to analyze the correlation and change of the composition in the signal, thereby realizing the detection of the quench fault.

[0005] The threshold-based quench detection method is susceptible to noise. When the signal is disturbed by environmental noise, false detection is prone to occur. At the same time, due to the difference in the structure of the superconducting magnet and the difference in the installation of the sensor, the threshold setting is different, which causes the quench determination to have a certain hysteresis, thereby affecting the operation safety.

[0006] The quench detection method based on the analysis of the correlation and change of the composition of the superconducting magnet sequence data has a certain specialization. For different types and models of superconducting magnets, the corresponding specific parameters are different. Due to the specificity of the data distribution, the fault detection criterion cannot take into account the data distribution, so the generalization ability of the method is limited, and it is difficult to improve the robustness and detection ability of the method as a whole. SUMMARY

[0007] The present application provides a magnet quench detection method based on individual group independent optimization immune algorithm, which can solve the problems in the prior art.

[0008] The present application provides a magnet quench detection method based on individual group independent optimization immune algorithm, wherein the method comprises:

[0009] preprocessing the existing sample and the to-be-tested sample respectively to obtain an existing sample signal frame and a to-be-tested sample signal frame, the existing sample including existing superconducting magnet voltage time series data, and the to-be-tested sample including to-be-tested superconducting magnet voltage time series data;

[0010] extracting features from the existing sample signal frame and the to-be-tested sample signal frame respectively to obtain a feature vector of the existing sample signal frame and a feature vector of the to-be-tested sample signal frame;

[0011] standardizing the feature vector of the existing sample signal frame and the feature vector of the to-be-tested sample signal frame respectively to obtain a standardized feature vector of the existing sample signal frame and a standardized feature vector of the to-be-tested sample signal frame;

[0012] initializing an antibody population to generate an initialized antibody population;

[0013] iteratively evolving the initialized antibody population to obtain an iteratively evolved antibody population;

[0014] group-optimizing and screening the iteratively evolved antibody population to obtain a screening result;

[0015] repeating the iteratively evolving and group-optimizing and screening until the antibody population no longer changes, and generating an optimized discrimination zone for the antibody population that no longer changes;

[0016] calculating distances between the standardized feature vector of the to-be-tested sample signal frame and all antibodies in the antibody population that no longer changes, and judging whether the magnet is quenched according to the calculated distances.

[0017] Preferably, the preprocessing includes overlapping framing and windowing, and the existing sample signal frame is obtained by the following formula:

[0018] s i =w i ·S,

[0019] wherein S is the existing superconducting magnet voltage time series data, w i is the i-th Hanning framing window, · is a vector dot product operator, and s i is the i-th signal frame of the existing sample after framing.

[0020] Preferably, the extracted features include root mean square features, kurtosis features and maximum derivative features, and the root mean square features, kurtosis features and maximum derivative features of the existing sample signal frame are represented by the following formulas:

[0021]

[0022]

[0023]

[0024] Ftvec i = [RMS i , Kurt i , Jerkpeak i ],

[0025] where T denotes the length of the signal frame s i , RMS i denotes the root mean square feature of the signal frame s i , μ k and σ k denote the mean and standard deviation of the signal frame s i , E(…) is the expectation operator, Kurt i denotes the kurtosis feature of the signal frame s i , max is the maximum operator, Jerkpeak i is the maximum jerk feature of the signal frame s i , and Ftvec i is the feature vector of the signal frame s i .

[0026] Preferably, the normalized feature vector of the existing sample signal frame is obtained by:

[0027]

[0028] where SFtvec i is the normalized feature vector of the signal frame s i , and μ and σ are the mean vector and standard deviation vector of the existing sample.

[0029] Preferably, the initial antibody population is initialized and generated by:

[0030] IAB i (n) = μ n + r x σ n ,

[0031] where IAB i (n) denotes the value of the nth feature dimension of the ith initial antibody, μ n and σ n are the values of the nth dimension of the mean vector μ of the existing sample and the standard deviation vector σ of the existing sample, and r is a random number between -1 and 1.

[0032] Preferably, the initialized antibody population is iteratively evolved by:

[0033] NAB i (n) = (1 + r n) x IAB i (n),

[0034] wherein r n is a random number between -0.5 and 0.5, NAB i (n) represents the value of the i-th antibody in the n-th feature dimension after iteration.

[0035] Preferably, the population of antibodies after iteration is screened by population optimization, and the screening result comprises:

[0036] The distance between each antibody in the population of antibodies after iteration and the existing sample is calculated.

[0037] The population of antibodies after iteration is screened by population optimization according to the calculated distance, and the screening result is obtained.

[0038] Preferably, the distance between each antibody in the population of antibodies after iteration and the existing sample is calculated by the following formula:

[0039] BGD i,j = ||NAB i -SFtvec j ||,

[0040] wherein ||…|| is a two-norm operator, BGD i,j is the distance between the i-th antibody after iteration and the j-th sample, NAB i is the value of the i-th antibody after iteration, SFtvec j is the normalized feature vector of the signal frame S j .

[0041] Preferably, the optimization discrimination zone is generated by the following formula:

[0042]

[0043]

[0044] Rcr i = max{bgd i , gbd i,j},

[0045] wherein bgd i represents the distance between NAB i and the nearest SFtvec j , SFtvec i,j represents the sample with NAB i as the nearest antibody, gbd i,j represents the maximum value of the distance between all samples with NAB i as the nearest antibody and NAB i , Rcri represents the recognition radius finally assigned to the antibody NAB i

[0046] Preferably, judging whether the magnet quenches according to the calculated distance comprises:

[0047] If the distance between the calculated normalized feature vector of the sample to be tested signal frame and all antibodies in the antibody group that is no longer changing is less than or equal to the radius Rcr i , then the sample to be tested belongs to the i antibody category, otherwise the sample to be tested does not belong to the i antibody category, wherein if the i antibody is generated by the superconducting magnet quench signal frame iteration, the i antibody belongs to the quench category, and if the i antibody is generated by the non-quench signal frame iteration, the i antibody belongs to the non-quench category.

[0048] Through the above technical solution, various features of the superconducting magnet voltage can be extracted, and the antibody group iteration optimization of antibody group initialization, iteration evolution and optimization screening is performed in the feature space based on the sample group, a specific determination area based on the sample individual is formed, and finally the specific determination area is used to realize accurate detection of the sample quench. Using the detection method described in the present application, accurate quench detection can be realized in the case of large difference in sample distribution, and compared with the traditional threshold method, the false judgment and missed judgment problems caused by noise and signal interference are effectively limited. Compared with the method of single parameter data analysis, the adaptability and generalization ability are stronger, and the high precision requirement of the superconducting magnet quench detection is met. BRIEF DESCRIPTION OF DRAWINGS

[0049] The accompanying drawings, which are included to provide a further understanding of the embodiments of the application and constitute a part of the specification, illustrate the embodiments of the application and together with the text description serve to explain the principles of the application. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0050] Figure 1 A flow chart of a magnet quench detection method based on individual group independent optimization immune algorithm according to an embodiment of the present application is shown;

[0051] Figure 2 An antibody group and sample (antigen) and discrimination zoning diagram according to an embodiment of the present application is shown;

[0052] Figure 3 A quench detection result diagram according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0053] ​It should be noted that the embodiments and features of the embodiments in the present application can be combined with each other in the case of no conflict. The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The description of the at least one example embodiment is actually only illustrative, but not as any limitation on the present application and its application or use. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0054] It should be noted that the terms used herein are only intended to describe specific embodiments, and are not intended to limit the exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form, unless the context clearly indicates otherwise, and it should also be understood that when the terms "comprise" and / or "include" are used in the specification, there is a reference to the presence of a feature, step, operation, device, component and / or combinations thereof.

[0055] Unless specifically stated otherwise, the relative arrangement of components and steps, numerical expressions, and numerical values set forth in the various embodiments described herein are not limiting. Also, it should be understood that the various embodiments can be implemented in hardware, software, or a combination thereof without departing from the scope of the present application. Moreover, unless specifically stated otherwise, any use of a term used in a singular sense can also be considered in a plural sense, and vice versa. Also, it should be understood that any numerical range recited is intended to include all sub-ranges subsumed therein. For example, a range of "1 to 10" is intended to include all sub-ranges between and including the minimum value of 1 and the maximum value of 10, that is, all sub-ranges beginning with a minimum value equal to or greater than 1 and ending with a maximum value equal to or less than 10, as well as all sub-ranges beginning and ending with values that are respectively equal to or greater than 1 and equal to or less than 10. The same applies to ranges recited using only one of the limits, for example, the range "less than or equal to 10," is intended to include all sub-ranges beginning and ending with values that are respectively equal to or greater than 0 and equal to or less than 10, as well as all sub-ranges beginning and ending with values that are respectively equal to or greater than 0 and equal to or less than 10. In other words, whenever a numerical range is recited, it should be understood to include all sub-ranges encompassed therein. Also, any reference to a number of steps in a method is intended to include variations in the number of steps, for example, additional or fewer steps being included, unless the number of steps is clearly critical to the operation of the present application.

[0056] Figure 1 A flow chart of a magnet quench detection method based on individual population independent optimization immune algorithm according to an embodiment of the present application is shown.

[0057] As Figure 1 shown, the embodiment of the present application provides a magnet quench detection method based on individual population independent optimization immune algorithm, wherein the method comprises:

[0058] The existing sample and the to-be-tested sample are respectively preprocessed to obtain an existing sample signal frame and a to-be-tested sample signal frame, the existing sample includes existing superconducting magnet voltage time series data, and the to-be-tested sample includes to-be-tested superconducting magnet voltage time series data.

[0059] The feature vectors of the existing sample signal frames and the feature vectors of the to-be-tested sample signal frames are obtained by performing feature extraction on the existing sample signal frames and the to-be-tested sample signal frames respectively.

[0060] The feature vectors of the existing sample signal frames and the feature vectors of the to-be-tested sample signal frames are respectively normalized to obtain the normalized feature vectors of the existing sample signal frames and the normalized feature vectors of the to-be-tested sample signal frames.

[0061] The antibody population is initialized to generate an initialized antibody population.

[0062] The initialized antibody population is iteratively evolved to obtain an iteratively evolved antibody population.

[0063] The iteratively evolved antibody population is screened by population optimization to obtain a screening result.

[0064] The steps of iterative evolution and population optimization screening are repeated until the antibody population no longer changes, and an optimized discrimination region is generated for the antibody population that no longer changes (i.e., the final antibody population).

[0065] The distances between the normalized feature vectors of the to-be-tested sample signal frames and all antibodies in the antibody population that no longer changes are calculated, and whether the magnet is quenched is judged according to the calculated distances.

[0066] Thus, the category of the to-be-tested sample can be determined to obtain a detection result.

[0067] Through the above technical solution, various features of the superconducting magnet voltage can be extracted, and antibody population iterative optimization of antibody population initialization, iterative evolution and optimization screening is performed in the feature space based on a sample population to form a specialized discrimination region based on sample individuals. Finally, accurate detection of sample quenching is realized according to the specialized discrimination region. Using the detection method described in the present application, accurate quenching detection can be realized in the case of large differences in sample distribution, effectively limiting the misjudgment and missed judgment problems caused by noise and signal interference compared with the traditional threshold method. At the same time, compared with the method of single parameter data analysis, it has stronger adaptability and generalization ability, and meets the high-precision requirements of actual superconducting magnet quenching detection.

[0068] According to an embodiment of the present application, the preprocessing includes overlapping framing and windowing, and the existing sample signal frame is obtained by the following formula:

[0069] s i =w i ·S,

[0070] wherein S is the existing superconducting magnet voltage time series data, w i is the i-th Hanning framing window, · is the vector dot product operator, and s iThe i-th signal frame of the existing sample obtained after frame division.

[0071] For example, the time length of the frame division window is 0.1s, and the adjacent front and rear two frame division windows are overlapped by 0.05s respectively, and the window function is 0 outside 0.1s. That is, the frame division length can be 0.1s, the frame division overlap is 0.05s, and the window function is Hanning window, which can ensure the time accuracy of the detection result.

[0072] Through frame division, the original time sequence data can be divided into frame segments suitable for processing, to ensure the time accuracy of the detection result, and the frame division overlap ensures the continuity between the signal frames and the frames, and the windowing can prevent the frequency leakage problem of the signal in the frame division process.

[0073] According to an embodiment of the present application, the extracted features include the extracted root mean square feature, kurtosis feature and maximum derivative feature, and the root mean square feature, kurtosis feature and maximum derivative feature of the existing sample signal frame are represented by the following formula:

[0074]

[0075]

[0076]

[0077] Ftvec i =[RMS i ,Kurt i ,Jerkpeak i ],

[0078] Wherein, T represents the length of the signal frame s i , RMS i represents the root mean square feature (feature value) of the signal frame s i , μ k and σ k represent the mean and standard deviation corresponding to the signal frame s i , E(…) is the expectation operator, Kurt i represents the kurtosis feature (feature value) of the signal frame s i , max is the maximum value operator, Jerkpeak i is the maximum derivative feature (feature value) of the signal frame s i , and Ftvec i is the feature vector of the signal frame s i .

[0079] According to an embodiment of the present application, the normalized feature vector of the existing sample signal frame is obtained by the following formula:

[0080]

[0081] wherein SFtvec i is a normalized feature vector of the signal frame s i , μ and σ are a mean vector and a standard deviation vector of the existing samples (i.e., a vector composed of the mean and the standard deviation of each column of the feature vectors Ftvec i of all signal frames).

[0082] By normalizing the feature vectors of the signal frames, the weight of each dimension in the determination direction can be ensured to be consistent.

[0083] According to an embodiment of the present application, the initialized antibody population is generated by the following formula:

[0084] IAB i (n) = μ n + r x σ n ,

[0085] wherein IAB i (n) represents the value of the n-th feature dimension of the i-th initial antibody, μ n and σ n are the values of the n-th dimension of the mean vector μ and the standard deviation vector σ of the existing samples, and r is a random number between -1 and 1.

[0086] Thus, the initialized antibody population is generated.

[0087] According to an embodiment of the present application, the initialized antibody population is iteratively evolved by the following formula to obtain an iterated antibody population:

[0088] NAB i (n) = (1 + r n ) x IAB i (n),

[0089] wherein r n is a random number between -0.5 and 0.5, and NAB i (n) represents the value of the n-th feature dimension of the i-th antibody after iteration.

[0090] According to an embodiment of the present application, the iterated antibody population is subjected to population optimization screening to obtain a screening result, which includes:

[0091] calculating the distance between each antibody in the iterated antibody population and the existing samples;

[0092] performing population optimization screening on the iterated antibody population according to the calculated distance to obtain a screening result.

[0093] According to one embodiment of the present application, the distance between each antibody in the iterated antibody group and the existing sample is calculated by the following formula:

[0094] BGD i,j = ||NAB i -SFtvec j ||,

[0095] wherein ||...|| is the two-norm operator, BGD i,j is the distance between the ith antibody and the jth sample after iteration, NAB i is the value of the ith antibody after iteration, SFtvec j is the normalized eigenvector of the signal frame S j .

[0096] Thus, the antibody group with the shortest distance to all individuals in the sample group can be selected as the new initial antibody group, i.e., IAB, according to the BGD i,j result.

[0097] According to one embodiment of the present application, the optimized discrimination zone is generated by the following formula:

[0098]

[0099]

[0100] Rcr i = max{bgd i ,gbd i,j},

[0101] wherein bgd i represents the distance between NAB i and the nearest SFtvec j , SFtvec i,j represents the sample with NAB i as the nearest antibody, gbd i,j represents the maximum value of the distance between all samples with NAB i as the nearest antibody and NAB i , and Rcr i represents the recognition radius finally assigned to the antibody NAB i .

[0102] In the present application, the final antibody group (the antibody group that no longer changes), the sample (antigen), and the discrimination zone are shown as illustrated in Figure 2 .

[0103] According to one embodiment of the present application, whether the magnet is out of superconductivity is determined according to the calculated distance, comprising:

[0104] If the distance between the normalized feature vector of the calculated sample signal frame and all antibodies in the no longer changing antibody population is less than or equal to the radius Rcr i If the i-antibody is generated iteratively from a superconducting magnet quench signal frame, then the i-antibody belongs to the quench category; otherwise, the i-antibody does not belong to the i-antibody category. If the i-antibody is generated iteratively from a non-quench signal frame, then the i-antibody belongs to the non-quench category.

[0105] For example, if we obtain the distance BGD between the j-th test sample and the i-th antibody... i,j Compared to BGD i,j With Rcr i The size, if BGD i,j Less than or equal to Rcr i If the result is positive, the sample to be tested belongs to the i-antibody category; otherwise, it does not belong to the i-antibody category.

[0106] Therefore, the category of the sample to be tested can be determined, and the test results can be obtained.

[0107] In this invention, in the feature extraction step, the existing samples and the samples to be tested share the same feature extraction method; in the sample standardization step, the existing samples and the samples to be tested share the standardization parameters (i.e., μ and σ mentioned above); the method for calculating the distance between the standardized feature vector of the signal frame of the sample to be tested and all antibodies in the antibody population that no longer changes can be the same as the method for calculating the distance between each antibody in the iterated antibody population and the existing samples.

[0108] The magnet quench detection method based on individual group independent optimization immune algorithm described in this invention is described below with examples.

[0109] Taking voltage data from a superconducting magnet test platform as an example, normal antibody clusters are generated from normal voltage data according to the corresponding steps in this invention, and quench antibody clusters are generated from voltage data during quenching according to the corresponding steps. These form antibody clusters capable of detecting both normal and quenching faults, and discrimination regions are generated according to the corresponding steps for each. New voltage data (data to be tested) is then selected and detected according to the corresponding steps. The obtained quenching detection results are as follows: Figure 3 As shown in the figure. The results show that the magnet quench detection method based on individual group independent optimization immune algorithm proposed in this invention can effectively distinguish between the normal and quench conditions of superconducting magnets. In the data obtained from the actual test platform, the quench detection accuracy is higher than 98.5%, which can effectively detect the working state of superconducting magnets and meet the needs of high-speed detection of superconducting magnets in actual situations.

[0110] From the above embodiment, it can be seen that the magnet quench detection method based on the individual group independent optimization immune algorithm can effectively reduce the high false detection rate and the missed detection rate caused by parameter fluctuation when relying on threshold judgment by extracting the superconducting magnet parameter characteristic expression, and the detection method based on the individual group independent optimization immune algorithm constructs a specific recognition area based on sample individual distribution, avoids the problems of false detection and missed detection caused by uneven sample distribution or non-convex sample distribution, and is beneficial to improving the detection accuracy of the superconducting magnet quench detection.

[0111] In the description of the present application, it should be understood that the orientation words such as "front, rear, upper, lower, left, right", "transverse, vertical, perpendicular, horizontal" and "top, bottom" and the like indicate the orientation or positional relationship shown in the drawings, which are only for the convenience of describing the present application and simplifying the description, and do not indicate and imply that the devices or elements referred to must have a specific orientation or be constructed and operated in a specific orientation, therefore cannot be understood as a limitation on the protection scope of the present application; the orientation words "inner, outer" refer to the inner and outer of the contour of each component itself.

[0112] For the convenience of description, spatial relative terms such as "over", "above", "upper surface", "upper" and the like can be used herein to describe the spatial positional relationship of one device or feature with other devices or features as shown in the drawings. It should be understood that the spatial relative terms are intended to include different orientations in use or operation in addition to the orientation of the devices described in the drawings. For example, if the devices in the drawings are inverted, the device described as "above" or "over" other devices or structures will be positioned "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below" orientations. The device can also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatial relative descriptions used herein are interpreted accordingly.

[0113] In addition, it should be noted that the use of the words "first", "second" and the like to define parts is only for the convenience of distinguishing the corresponding parts, and the above words have no special meaning unless otherwise stated, therefore cannot be understood as a limitation on the protection scope of the present application.

[0114] The above description is only preferred embodiments of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A magnet quench detection method based on an individual-group independent optimization immune algorithm, characterized in that, The method includes: The existing samples and the samples to be tested are preprocessed separately to obtain the existing sample signal frame and the sample to be tested signal frame. The existing samples include the existing superconducting magnet voltage time series data, and the samples to be tested include the superconducting magnet voltage time series data to be tested. Feature extraction is performed on the existing sample signal frame and the sample signal frame to be tested to obtain the feature vector of the existing sample signal frame and the feature vector of the sample signal frame to be tested. The feature vectors of the existing sample signal frame and the feature vector of the sample signal frame to be tested are standardized respectively to obtain the standardized feature vectors of the existing sample signal frame and the standardized feature vectors of the sample signal frame to be tested. The antibody population is initialized to generate an initialized antibody population; The initial antibody population is iteratively evolved to obtain the iterative antibody population; The iterative antibody population was subjected to population optimization screening to obtain the screening results; Repeat the iterative evolution and population optimization screening steps until the antibody population no longer changes, and generate an optimized discriminant region for the antibody population that no longer changes; Calculate the distance between the standardized feature vector of the signal frame of the sample to be tested and all antibodies in the antibody population that no longer changes, and determine whether the magnet has lost quench based on the calculated distance; The antibody population is initialized and generated using the following formula: ,in, This represents the value of the i-th initial antibody in the n-th feature dimension. and The mean vector of existing samples Existing sample standard deviation vector The value in the nth dimension, where r is a random number between -1 and 1.

2. The method according to claim 1, characterized in that, Preprocessing includes overlapping framing and windowing, and the existing sample signal frames are obtained using the following formula: ,in, This is based on existing superconducting magnet voltage timing data. For the first One Hanning frame window, This is the vector dot product operator. The existing sample obtained after frame segmentation One signal frame.

3. The method according to claim 2, characterized in that, The extracted features include root mean square (RMS) features, kurtosis features, and maximum derivative features. The following formulas represent the RMS, kurtosis, and maximum derivative features of the existing sample signal frames: , , , Where T represents a signal frame Length, Indicates signal frame The root mean square characteristic, and Indicates signal frame The corresponding mean and standard deviation, E(...) is the expectation operator. This represents the kurtosis characteristic of the signal frame, with max being the maximum value operator. For signal frames Maximum derivative characteristic, Ftvec i For signal frames eigenvectors.

4. The method according to claim 3, characterized in that, The standardized feature vector of the existing sample signal frame is obtained by the following formula: Among them, SFtvec i For signal frames The standardized feature vector, and These are the mean vector and standard deviation vector of the existing samples.

5. The method according to claim 4, characterized in that, The initialized antibody population is iteratively evolved using the following formula to obtain the iteratively evolved antibody population: ,in, A random number between -0.5 and 0.

5. This represents the value of the i-th antibody in the n-th feature dimension after iteration.

6. The method according to claim 5, characterized in that, The iterative antibody population was subjected to population optimization screening, and the screening results included: Calculate the distance between each antibody in the iterative antibody population and the existing samples; The antibody population was optimized and screened based on the calculated distance to obtain the screening results.

7. The method according to claim 6, characterized in that, The distance between each antibody in the iterated antibody population and the existing samples is calculated using the following formula: Where, ||…|| are the 2-norm operators, BGD i,j NAB represents the distance between the i-th antibody and the j-th sample after iteration. i SFtvec represents the value of the i-th antibody after iteration. j For signal frame S j The standardized feature vector.

8. The method according to claim 7, characterized in that, The optimized discrimination region is generated using the following formula: , , ,in, Indicates separation With the most recent The distance between them Indicated by For the most recent antibody sample, GBD i,j Indicates finding For all samples of recent antibodies and The maximum distance between them, Rcr i This indicates the final allocation to the antibody. The recognition radius.

9. The method according to claim 8, characterized in that, Determining whether a magnet has lost quench based on calculated distance includes: If the distance between the normalized feature vector of the calculated sample signal frame and all antibodies in the no longer changing antibody population is less than or equal to the radius Rcr i If the i-antibody is generated iteratively from a superconducting magnet quench signal frame, then the i-antibody belongs to the quench category; otherwise, the i-antibody does not belong to the i-antibody category. If the i-antibody is generated iteratively from a non-quench signal frame, then the i-antibody belongs to the non-quench category.

Images