CN116184088A - A Fault Detection Method for Electromagnetic Radiation Emitting System Based on Electromagnetic Spectrum Features - Google Patents

Abstract

The invention discloses a fault detection method of an electromagnetic radiation emission system based on electromagnetic spectrum characteristics, comprising the following steps: S1. determining the test frequency band and M common fault states of the electromagnetic radiation emission system, and testing to obtain N groups of background environmental noises; S2 .In the mth kind of fault state, test the radiation emission amplitude of N groups of electromagnetic radiation emission systems; S3. Extract the electromagnetic characteristics of N groups of emission spectrum in the mth kind of fault state; S4. Get M kinds of fault states, each Emission electromagnetic spectrum of a group of fault states; S5. Update the electromagnetic characteristics of N groups of emission spectra under each fault state; S6. Build and train a fault identification model; S7. Use a mature fault identification model for electromagnetic radiation emission system Fault detection. The invention utilizes the corresponding relationship between the electromagnetic radiation emission characteristics of the equipment and the equipment to train the fault identification model for the fault detection of the electromagnetic radiation emission system, which reduces the difficulty of fault detection and shortens the troubleshooting cycle.

Description

A fault detection method for electromagnetic radiation emission system based on electromagnetic spectrum characteristics

Technical Field

The invention relates to electromagnetic radiation, and in particular to a fault detection method for an electromagnetic radiation emission system based on electromagnetic spectrum characteristics.

Background Art

For an electromagnetic radiation emission system that includes multiple electronic devices (or subsystems), each electronic device is often distributed in different locations of the system. For example, for an aircraft, these electronic devices (or subsystems) are concentrated in equipment compartments in the middle, bottom or tail of the aircraft, or inside the cockpit.

However, the internal space resources of the electromagnetic radiation emission system are very precious and scarce, and the equipment compartments where electronic equipment is placed are often very small, and the electronic equipment is closely arranged in them. In such a small space, the electromagnetic radiation emission signals generated by different devices affect each other and overlap, making it difficult to match the electromagnetic radiation emission characteristics of a specific location with the corresponding devices near that location when all devices are turned on.

Electronic devices are composed of various core circuit structures and other peripheral functional circuits. The excitation source signals and other functional signals generated by the internal circuits can be emitted into the environment in the form of electromagnetic radiation, especially high-frequency signals, such as clocks, I/O lines, and signals generated by internal switches. In addition, processing defects of printed circuit boards or design defects in the circuits themselves and attached peripherals will also lead to a large amount of electromagnetic emissions.

Since the location and category of electronic equipment are not clear, when a fault occurs in the electromagnetic radiation emission system, it is generally judged manually and checked one by one. This also takes a long time, has high requirements for testers, and has a low degree of test automation. As a result, the existing electromagnetic emission source detection methods cannot meet the fault detection needs of the electromagnetic radiation emission system.

Summary of the invention

The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method for fault detection of an electromagnetic radiation emission system based on electromagnetic spectrum characteristics. The method uses the relationship between the electromagnetic radiation emission characteristics of a device and the corresponding device to train a fault identification model. After obtaining a mature model, it is used for fault detection in the electromagnetic radiation emission system, which reduces the difficulty of fault detection and shortens the troubleshooting cycle.

The object of the present invention is achieved through the following technical solution: a method for detecting faults in an electromagnetic radiation emission system based on electromagnetic spectrum characteristics, comprising the following steps:

S1. Determine the test frequency band and M common fault conditions of the electromagnetic radiation emission system, and test to obtain N groups of background environmental noise

S2. Under the mth fault state, the radiation emission amplitude of N groups of electromagnetic radiation emission systems is obtained by testing.

S3. extracting the electromagnetic characteristics of the emission spectrum of the N groups according to the background environmental noise and the radiation emission amplitude of the electromagnetic radiation emission system under the mth fault state;

S4. When m=1, 2, ..., M, repeat steps S2 to S3 to obtain the emission electromagnetic spectrum of each group of fault states under M fault states;

S5. Extract a simplified set of electromagnetic characteristics of the emission spectrum under each fault state, update the N sets of electromagnetic characteristics of the emission spectrum under the fault state, and record all test frequency points contained in the simplified set of electromagnetic characteristics of the emission spectrum under M fault states;

S6. Build a fault recognition model based on artificial intelligence algorithm, and use N groups of updated emission spectrum electromagnetic features of each fault state as samples, and use the fault type as the sample label;

The fault recognition model is trained using the constructed samples and sample labels. When all the emission spectrum sample features of all groups under the fault state in M are trained, a mature fault recognition model is obtained.

S7. Use mature fault identification models to perform fault detection on electromagnetic radiation emission systems.

The beneficial effects of the present invention are: (1) the scheme of the present invention takes into account the influence of environmental interference and adopts corresponding processing means, so it has strong applicability;

(2) The present invention compresses the dimensions of the spectrum data that need to be processed, thereby reducing the burden on the artificial intelligence algorithm classifier.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a flow chart of the method of the present invention;

FIG2 is a schematic diagram of single radiation emission characteristics of a device set under different fault conditions in an embodiment;

FIG3 is a distribution diagram of radiation emission characteristics of a device set under different fault conditions in an embodiment;

FIG4 is a schematic diagram of the structure of a decision tree under full sample conditions in an embodiment.

DETAILED DESCRIPTION

The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings, but the protection scope of the present invention is not limited to the following.

For electronic devices, they will intentionally or unintentionally generate electromagnetic radiation emissions to the outside. And electromagnetic radiation emissions are an inherent property of electronic equipment, which is related to its internal structure, composition principle and working state. Therefore, the electromagnetic radiation emissions of electronic equipment are different, and there is a one-to-one mapping relationship between the radiation emissions and the test products in different states. Therefore, even if there are similar parts in the radiation emissions of electronic equipment and systems of the same type or model, the radiation emissions of each electronic device at the microscopic level are different. This also means that on the basis of obtaining the characteristics of electronic electromagnetic radiation emissions, the type and working state of the test product can be targetedly identified, and the purpose of identifying the status of the equipment based on electromagnetic radiation can be finally achieved. That is, the one-to-one correspondence between the electromagnetic emission characteristics of the equipment and the equipment is used to complete the fault location of the equipment, specifically:

As shown in FIG1 , a method for detecting electromagnetic radiation emission system faults based on electromagnetic spectrum characteristics includes the following steps:

S1. Determine the test frequency band and M common fault conditions of the electromagnetic radiation emission system, and test to obtain N groups of background environmental noise

The electromagnetic radiation emission system includes Q devices, namely device 1, device 2, ..., device Q;

In the common M fault states, each fault state corresponds to a different combination of faulty devices. That is to say, in each fault state, it is clear which devices are faulty and which are normal. Therefore, only the fault state needs to be known to complete the fault detection. For example:

Fault Status Faulty equipment State 1 Only device 1 is faulty State 2 Only device 1, device 2 failed … … Status m … … Status M …

The test frequency band is a vector of length K, denoted as:

Fre＝(fre ₁ , fre ₂ ,..., fre _k ,..., fre _K );

Wherein, fre _k represents the kth test frequency point, k = 1, 2, ..., K.

The process of testing N groups of background environmental noise includes:

S101. Perform background environmental noise test at each frequency point in the test frequency band Fre = (fre ₁ ,fre ₂ , ...,fre _k , ...,fre _K ) to obtain background environmental noise;

S102. Repeat the test in step S101 N times to obtain N groups of background environmental noise, recorded as:

表示第n次背景环境噪声测试得到的结果，n＝1,2,...,N；in,

represents the result of the nth background noise test, n = 1, 2, ..., N;

表示第n次测试时，在第k个频点处测得的背景环境噪声，k＝1,2,...,K。

It represents the background environmental noise measured at the kth frequency point during the nth test, where k = 1, 2, ..., K.

S2. Under the mth fault state, the radiation emission amplitude of N groups of electromagnetic radiation emission systems is obtained by testing.

S201. Testing the radiation emission amplitude of the magnetic radiation emission system at each frequency point in the test frequency band Fre = (fre ₁ ,fre ₂ , ...,fre _k , ...,fre _K ) to obtain the radiation emission amplitude test result;

S202. Repeat the test in step S201 N times to obtain N groups of radiation emission amplitude test results, recorded as:

表示第n次辐射发射幅度测试结果，n＝1,2,...,N；in,

Indicates the nth radiation emission amplitude test result, n = 1, 2, ..., N;

表示第n辐射发射幅度测试过程中在第k个频点处测得的结果。in

Represents the result measured at the kth frequency point during the nth radiated emission amplitude test.

S3. extracting the electromagnetic characteristics of the emission spectrum of the N groups according to the background environmental noise and the radiation emission amplitude of the electromagnetic radiation emission system under the mth fault state;

提取出明显高于背景噪声

的元素所对应的测试频点，作为第n组测试的发射频谱电磁特征：S301. Test results of the nth radiation emission amplitude under the mth state

Extracted noise significantly higher than the background

The test frequency points corresponding to the elements are used as the emission spectrum electromagnetic characteristics of the nth group of tests:

中的第k个元素

若与

中的第k个元素之差大于设定阈值，则认为

明显高于背景噪声，将对应的测试频点k加入集合

中；A1. For

The kth element in

If

If the difference of the kth element in is greater than the set threshold, it is considered

Obviously higher than the background noise, the corresponding test frequency k is added to the set

middle;

A2. When k = 1, 2, ..., K, repeat step A1 to obtain the electromagnetic characteristics of the emission spectrum:

中包含的频点数目；Where K′ represents the electromagnetic characteristics of the emission spectrum

The number of frequencies included in ;

S302. When n=1, 2, ..., N, repeat step S301 to obtain N groups of emission spectrum electromagnetic characteristics.

S4. When m=1, 2, ..., M, repeat steps S2 to S3 to obtain the emission electromagnetic spectrum of each group of fault states under M fault states;

S5. Extract a simplified set of electromagnetic characteristics of the emission spectrum under each fault state, update the N sets of electromagnetic characteristics of the emission spectrum under the fault state, and record all test frequency points contained in the simplified set of electromagnetic characteristics of the emission spectrum under M fault states;

其中j表示状态m的发射频谱电磁特征集中包含的测试频点数目；S501. When m=1,2,...,M, select all test frequency points contained in the N groups of emission spectrum electromagnetic characteristics in the mth state, record the test frequency points whose occurrence times are greater than 0.9*N, and add them to the same set to obtain the emission spectrum electromagnetic characteristic set of state m

Where j represents the number of test frequency points included in the emission spectrum electromagnetic characteristic set of state m;

S502. Compress the electromagnetic spectrum feature set under the mth fault state according to the electromagnetic feature set of the emission spectrum under all M states. The compression process follows the following three principles:

其中

为向上取整；(1) The compressed electromagnetic spectrum features should be able to effectively distinguish all fault states of the equipment, and the number of features should not be less than

in

To round up;

(2) For the emission spectrum electromagnetic features that exist in all M fault states, remove them from the emission spectrum feature set fre ^m in the mth state;

(3) When different electromagnetic spectrum features behave exactly the same in different states, only one needs to be retained;

For any two test frequencies fre _a and fre _b in fre ^m ;

Traverse the emission spectrum feature sets of M fault states, and add the fault state corresponding to the emission spectrum feature set containing the test frequency point fre _a to the set _Ma ;

Traverse the emission spectrum feature sets of M fault states, and add the fault state corresponding to the emission spectrum feature set containing the test frequency point fre _b to the set _Ma ;

If _Ma = _Mb , then only one of the spectral feature points fre _a and fre _b is retained in fre ^m , and the other is deleted from fre ^m ;

其中

通过压缩可以减少需要特征的维数，从而大大减少人工智能算法的训练复杂度。After (1) to (3), a simplified set of emission spectrum electromagnetic characteristics is obtained.

in

Compression can reduce the dimension of required features, thereby greatly reducing the training complexity of artificial intelligence algorithms.

S503. Update the emission spectrum electromagnetic characteristics of the nth group of tests under the mth fault state, and the updated result is:

S504. When n=1, 2, ..., N, repeat steps S502 to S503 to obtain N groups of updated emission spectrum electromagnetic characteristics under the mth fault state;

S505. When m=1, 2, ..., M, repeat step S504 to obtain N groups of updated emission spectrum electromagnetic characteristics for each fault state in the M fault states;

S506. Record the set of all test frequency points contained in the simplified emission spectrum electromagnetic feature set under M fault states:

S6. Build a fault recognition model based on artificial intelligence algorithm, and use N groups of updated emission spectrum electromagnetic features of each fault state as samples, and use the fault type as the sample label;

The fault recognition model is trained using the constructed samples and sample labels. When all the emission spectrum sample features of all groups under the fault state in M are trained, a mature fault recognition model is obtained.

The artificial intelligence algorithm includes but is not limited to a machine learning algorithm or a neural network algorithm. By using these algorithms to construct a classifier model, the model training can be realized; for example, the machine learning algorithm can adopt one of logistic regression (LogisticRegression), naive Bayes (Naive Bayes), nearest neighbors (K-Nearest Neighbors), decision tree (Decision Tree), and support vector machines (Support Vector Machines); the neural network algorithm can adopt convolutional neural networks (Convolutional Neural Networks, CNN), deep neural networks (Deep Neural Networks, DNN), etc.

S7. Use mature fault identification models to perform fault detection on electromagnetic radiation emission systems.

S701. When the electromagnetic radiation transmission system fails, the radiation emission amplitude of the magnetic radiation transmission system is tested at each frequency point in the test frequency band Fre = (fre ₁ ,fre ₂ , ...,fre _k , ...,fre _K ) to obtain the radiation emission amplitude test result under the current fault;

S702. Select any group of background environmental noise, compare it with the radiation emission amplitude test result under the current fault at each frequency point, and select the test frequency point where the difference between the radiation emission amplitude and the background environmental noise is greater than the set threshold, which constitutes the emission spectrum electromagnetic characteristics of the current fault:

fre _0-chara : fre _0-chara = [fre′ ₁ , fre′ ₂ ,...fre′ _k′ ];

Wherein, k′ represents the number of test frequency points included in fre _0-chara ;

S703. Find the intersection of the emission spectrum electromagnetic feature fre _0-chara of the current fault and the set fre _Chara consisting of all test frequency points contained in the simplified emission spectrum electromagnetic feature set under M fault states, and obtain:

送入成熟的故障识别模型，由成熟的故障识别模型输出的故障状态作为系统故障检测结果。S704.

The mature fault identification model is input, and the fault status output by the mature fault identification model is used as the system fault detection result.

In the embodiment of the present application, a specific fault detection process is introduced by taking the case where only one of the six devices cannot work normally as an example.

(1) Testing phase

For the six devices, there are six different test states, namely: device 1 does not work, and the other five devices work; device 2 does not work, and the other five devices work... device 6 does not work, and the other five devices work. 30 groups of tests were performed for the above six working states. The single test results in each state are shown in the solid line in Figure 2.

According to the method of this application, the characteristic frequency points of six devices under different fault conditions can be obtained as shown in the “*” in Figure 2 below;

(2) Feature screening stage

In order to further reduce the impact of random noise on the selected emission features, we count the number of occurrences of a certain emission feature under the same fault state. If the probability of the emission feature occurring is greater than 90%, we take it as an emission feature of the fault state. Finally, we can get the distribution of radiation emission features under various fault states as shown in Figure 3.

As can be seen, when the third type of equipment fails, the radiation emission features picked up from the entire aircraft equipment compartment are relatively few, only 8. For other types of equipment failures, the radiation emission features may be as many as 20. Since we are going to further adopt artificial intelligence algorithms, artificial intelligence algorithms are essentially a kind of probabilistic statistics. Therefore, if there are too many features of the sample, the algorithm will require more samples. Therefore, in order to further reduce the demand for data volume, we further trimmed the radiation features under different fault states. The basis for trimming is as follows: a. The radiation emission features that exist in all six fault states are not sufficient to distinguish the six fault states, so they are screened out from the final emission features (such as 1.201GHz in Figure 3 exists in all six states); b. For two different emission features that are consistent in different equipment fault states, we only need to keep one group (such as 300MHz and 1.15GHz in Figure 3 appear in the 1st, 2nd, 3rd, 5th, and 6th states); c. In theory, only three carefully selected radiation emission features are needed to identify eight different fault states. In summary, in order to compress the number of radiation emission features on the one hand and to ensure redundancy on the other hand, we finally selected five isolated peak signal radiation characteristics as radiation emission features, which are f1, f2, f3, f4, and f5.

The one-to-one correspondence between these five electromagnetic radiation characteristics and fault conditions is shown in the following table.

Feature 1 Feature 2 Feature 3 Feature 4 Feature 5 f1 f2 f3 f4 f5 Device 1 failure 0 0 0 1 0 Device 2 Failure 1 1 1 1 1 Device 3 failure 1 1 1 1 0 Device 4 failure 1 1 1 0 0 Device 5 failure 1 1 0 0 0 Device 6 failure 1 0 0 0 0

Among them, 1 means that the frequency point is a feature corresponding to the fault state, and 0 means that the frequency point is not a feature corresponding to the fault state;

It is worth noting that the characteristics of a certain type of fault are not necessarily related to the characteristics of a single test. Taking the failure of device 1 as an example, 1351MHz is not a feature of the failure of device 1, but a test when device 1 fails contains the feature of 1351MHz. In order to establish a more accurate prediction model, it is necessary to further introduce artificial intelligence algorithms.

(3) Feature Identification Stage

Based on the above principles, the corresponding logic of the decision tree algorithm is implemented through programming. In order to ensure the accuracy of fault prediction, all test data are used as samples to train the decision tree algorithm. The trained decision tree structure is shown in Figure 4, where x1, x2, x3, x4, and x5 correspond to radiation features 1 to 5 in each group of test data in Table 6. Taking the first group of test data in the case of equipment fault state 1 as an example, its feature vector is [0,0,0,1,0]. When making a decision, starting from the root node, x1＝0<0.5, so from the root node to the left to the next layer of nodes, x3＝0<0.5 at the child node layer, and finally reaching the decision node. The accuracy of the equipment fault state prediction result under this decision tree can reach 95%.

The above description shows and describes a preferred embodiment of the present invention, but as mentioned above, it should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the invention concept described herein through the above teachings or the technology or knowledge of the relevant field. Changes and variations made by those skilled in the art do not depart from the spirit and scope of the present invention, and should be within the scope of protection of the claims attached to the present invention.

Claims (9)

1. A method for detecting faults in an electromagnetic radiation emission system based on electromagnetic spectrum characteristics, characterized in that it comprises the following steps: S1. Determine the test frequency band and M common fault conditions of the electromagnetic radiation emission system, and test to obtain N groups of background environmental noise

S2. Under the mth fault state, the radiation emission amplitude of N groups of electromagnetic radiation emission systems is obtained by testing.

S3. extracting the electromagnetic characteristics of the emission spectrum of the N groups according to the background environmental noise and the radiation emission amplitude of the electromagnetic radiation emission system under the mth fault state; S4. When m=1, 2, ..., M, repeat steps S2 to S3 to obtain the emission electromagnetic spectrum of each group of fault states under M fault states; S5. Extract a simplified set of electromagnetic characteristics of the emission spectrum under each fault state, update the N sets of electromagnetic characteristics of the emission spectrum under the fault state, and record all test frequency points contained in the simplified set of electromagnetic characteristics of the emission spectrum under M fault states; S6. Build a fault recognition model based on artificial intelligence algorithm, and use N groups of updated emission spectrum electromagnetic features of each fault state as samples, and use the fault type as the sample label; The fault recognition model is trained using the constructed samples and sample labels. When all the emission spectrum sample features of all groups under the fault state in M are trained, a mature fault recognition model is obtained. S7. Use mature fault identification models to perform fault detection on electromagnetic radiation emission systems. 2. The electromagnetic radiation emission system fault detection method based on electromagnetic spectrum characteristics according to claim 1 is characterized in that: the electromagnetic radiation emission system in step S1 includes Q devices, namely device 1, device 2, ..., device Q; In the M common fault states described in step S1, each fault state corresponds to a different combination of faulty devices. 3. The electromagnetic radiation emission system fault detection method based on electromagnetic spectrum characteristics according to claim 1 is characterized in that: the test frequency band in step S1 is a vector of length K, which is expressed as: Fre＝(fre ₁ , fre ₂ ,..., fre _k ,..., fre _K ); Wherein, fre _k represents the kth test frequency point, k = 1, 2, ..., K. 4. The electromagnetic radiation emission system fault detection method based on electromagnetic spectrum characteristics according to claim 1 is characterized in that: the process of testing N groups of background environmental noise in step S1 comprises: S101. Perform background environmental noise test at each frequency point in the test frequency band Fre = (fre ₁ ,fre ₂ , ...,fre _k , ...,fre _K ) to obtain background environmental noise; S102. Repeat the test in step S101 N times to obtain N groups of background environmental noise, recorded as:

in,

represents the result of the nth background noise test, n = 1, 2, ..., N;

It represents the background environmental noise measured at the kth frequency point during the nth test, where k = 1, 2, ..., K. 5. The electromagnetic radiation emission system fault detection method based on electromagnetic spectrum characteristics according to claim 1, characterized in that: step S2 comprises the following steps: S201. Testing the radiation emission amplitude of the magnetic radiation emission system at each frequency point in the test frequency band Fre = (fre ₁ ,fre ₂ , ...,fre _k , ...,fre _K ) to obtain the radiation emission amplitude test result; S202. Repeat the test in step S201 N times to obtain N groups of radiation emission amplitude test results, recorded as:

in,

Indicates the nth radiation emission amplitude test result, n = 1, 2, ..., N;

in

Represents the result measured at the kth frequency point during the nth radiated emission amplitude test. 6. The electromagnetic radiation emission system fault detection method based on electromagnetic spectrum characteristics according to claim 1, characterized in that: step S3 comprises the following sub-steps: S301. Test results of the nth radiation emission amplitude in the mth state

Extracted noise significantly higher than the background

The test frequency points corresponding to the elements are used as the emission spectrum electromagnetic characteristics of the nth group of tests: A1. For

The kth element in

If

If the difference of the kth element in is greater than the set threshold, it is considered

Obviously higher than the background noise, the corresponding test frequency k is added to the set

middle; A2. When k = 1, 2, ..., K, repeat step A1 to obtain the electromagnetic characteristics of the emission spectrum:

Where K′ represents the electromagnetic characteristics of the emission spectrum

The number of frequencies included in ; S302. When n=1, 2, ..., N, repeat step S301 to obtain N groups of emission spectrum electromagnetic characteristics. 7. The electromagnetic radiation emission system fault detection method based on electromagnetic spectrum characteristics according to claim 6, characterized in that: step S5 comprises: S501. When m=1,2,...,M, select all test frequency points contained in the N groups of emission spectrum electromagnetic characteristics in the mth state, record the test frequency points whose occurrence times are greater than 0.9*N, and add them to the same set to obtain the emission spectrum electromagnetic characteristic set of state m

Where j represents the number of test frequency points included in the emission spectrum electromagnetic characteristic set of state m; S502. Compress the electromagnetic spectrum feature set under the mth fault state according to the electromagnetic feature set of the emission spectrum under all M states. The compression process follows the following three principles: (1) The compressed electromagnetic spectrum features should be able to effectively distinguish all fault states of the equipment, and the number of features should not be less than

in

To round up; (2) For the emission spectrum electromagnetic features that exist in all M fault states, remove them from the emission spectrum feature set fre ^m in the mth state; (3) When different electromagnetic spectrum features behave exactly the same in different states, only one needs to be retained; For any two test frequencies fre _a and fre _b in fre ^m ; Traverse the emission spectrum feature sets of M fault states, and add the fault state corresponding to the emission spectrum feature set containing the test frequency point fre _a to the set _Ma ; Traverse the emission spectrum feature sets of M fault states, and add the fault state corresponding to the emission spectrum feature set containing the test frequency point fre _b to the set _Ma ; If _Ma = _Mb , then only one of the spectral feature points fre _a and fre _b is retained in fre ^m , and the other is deleted from fre ^m ; After (1) to (3), a simplified set of emission spectrum electromagnetic characteristics is obtained.

in

S503. Update the emission spectrum electromagnetic characteristics of the nth group of tests under the mth fault state, and the updated result is:

S504. When n=1, 2, ..., N, repeat steps S502 to S503 to obtain N groups of updated emission spectrum electromagnetic characteristics under the mth fault state; S505. When m=1, 2, ..., M, repeat step S504 to obtain N groups of updated emission spectrum electromagnetic characteristics for each fault state in the M fault states; S506. Record the set of all test frequency points contained in the simplified emission spectrum electromagnetic feature set under M fault states:

8. The electromagnetic radiation emission system fault detection method based on electromagnetic spectrum characteristics according to claim 6, characterized in that: step S7 comprises: S701. When the electromagnetic radiation transmission system fails, the radiation emission amplitude of the magnetic radiation transmission system is tested at each frequency point in the test frequency band Fre = (fre ₁ ,fre ₂ , ...,fre _k , ...,fre _K ) to obtain the radiation emission amplitude test result under the current fault; S702. Select any group of background environmental noise, compare it with the radiation emission amplitude test result under the current fault at each frequency point, and select the test frequency point where the difference between the radiation emission amplitude and the background environmental noise is greater than the set threshold, which constitutes the emission spectrum electromagnetic characteristics of the current fault: fre _0-chara : fre _0-chara = [fre ₁ ′, fre′ ₂ ,...fre′ _k′ ]; Wherein, k′ represents the number of test frequency points included in fre _0-chara ; S703. Find the intersection of the emission spectrum electromagnetic feature fre _0-chara of the current fault and the set fre _Chara consisting of all test frequency points contained in the simplified emission spectrum electromagnetic feature set under M fault states, and obtain:

S704.

The mature fault identification model is input, and the fault status output by the mature fault identification model is used as the system fault detection result. 9. The electromagnetic radiation emission system fault detection method based on electromagnetic spectrum characteristics according to claim 1 is characterized in that: the artificial intelligence algorithm includes but is not limited to a machine learning algorithm or a neural network algorithm.

Images