Ray source state detection method, detection system, device and electronic equipment

By installing a sound sensor on the outer surface of the X-ray source enclosure, collecting and analyzing acoustic signatures, and identifying and locating the working status of the X-ray source, the problem of not being able to detect the status of the X-ray source in existing technologies is solved, thus achieving the effect of reducing the defect rate.

CN118915188APending Publication Date: 2024-11-08HANGZHOU RAYIN TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411073650.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

Existing technologies cannot effectively detect the working status of X-ray sources, resulting in high defect rates during production and use, especially since electronic components cannot be placed for testing in ultra-high voltage environments.

Method used

By installing a sound sensor on the outer surface of the X-ray source box, internal sound signals are collected, acoustic features are extracted, non-arc state and arc state are identified, and the target location of the arc phenomenon is determined by using time and propagation speed.

Benefits of technology

It enables rapid and accurate detection of the working status of the X-ray source without being affected by ultra-high voltage, reducing the defect rate during production and use.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118915188A_ABST
    Figure CN118915188A_ABST
Patent Text Reader

Abstract

The embodiment of the invention provides a ray source state detection method, system and device and electronic equipment. A processor can obtain a sound signal collected by a sound sensor and extract voiceprint features of the sound signal; the working state of the interior of the radiation source box body is determined by recognizing voiceprint features, and under the condition that the working state is determined to be an ignition state, the target position is determined according to the time when the sound sensor collects the sound signals, the installation position of the sound sensor and the propagation speed of the sound signals. As the ignition phenomenon is accompanied by specific sound, and the sound cannot be emitted during normal work, the voiceprint characteristics of sound signals in the non-ignition state and the ignition state are different, and the working state can be determined through the sound signals in the radiation source box body. As the sound sensor can collect the sound signals on the outer surface of the radiation source box body and does not need to be arranged in the box body, the working state of the radiation source can be detected, the reject ratio of the radiation source is reduced, and the target position is determined.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of electrical equipment detection, and in particular to a method, a detection system, a device, and an electronic device for detecting the state of a radiation source. Background Art

[0002] The radiation source is the core component of the X-ray security inspection system and is also the most common solution for generating controllable X-rays. Since the ionizing effect of the radiation source can cause radiation and damage to the human body, the radiation source must be shielded. Usually, the radiation source is made into a sealed container wrapped in a lead box. The principle of the radiation source to generate X-rays is to use ultra-high voltage to accelerate electrons to bombard the tungsten target to produce radiation. Under the ultra-high voltage environment, the electrons in the air will be accelerated by the electric field to form an arc, that is, a spark occurs. In the spark phenomenon, the high temperature generated by the arc discharge may melt the internal structure of the radiation source and may also cause a fire.

[0003] Due to the ultra-high voltage inside the radiation source, electronic components will be broken down by the ultra-high voltage, so it is impossible to place electronic components to detect the status of the radiation source, resulting in a high defect rate in the production and subsequent use of the radiation source. Summary of the invention

[0004] The purpose of the embodiments of the present application is to provide a method, a detection system, a device, and an electronic device for detecting the state of a radiation source, so as to detect the working state of the radiation source and reduce the defect rate of the radiation source during production and subsequent use. The specific technical solution is as follows:

[0005] In a first aspect, an embodiment of the present application provides a method for detecting a state of a radiation source, which is applied to a processor in a detection system, wherein the detection system further includes a sound sensor, which is installed on at least three outer surfaces of a radiation source box, and the sound sensor is used to collect a sound signal emitted from inside the radiation source box, wherein the method includes:

[0006] Acquire the sound signal collected by the sound sensor, and extract the voiceprint feature of the sound signal;

[0007] By identifying the voiceprint feature, determining the working state inside the radiation source box, wherein the working state includes a non-ignition state and an ignition state;

[0008] When it is determined that the working state is the ignition state, the target position where the ignition occurs inside the ray source box is determined based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor and the propagation speed of the sound signal inside the ray source box.

[0009] Optionally, the number of the sound sensors is four;

[0010] The step of determining the target position where the spark occurs inside the ray source box according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box includes:

[0011] The target position where the ignition occurs inside the ray source box is determined based on the time difference between the sound signals collected by the four sound sensors, the positional relationship between the installation positions of the four sound sensors, and the propagation speed of the sound signal inside the ray source box.

[0012] Optionally, the step of determining a target position where the ignition phenomenon occurs inside the ray source box according to the time difference between the sound signals collected by four sound sensors, the positional relationship between the installation positions of the four sound sensors, and the propagation speed of the sound signal inside the ray source box includes:

[0013] Among the four sound sensors, one is used as a first sound sensor, and for each of the remaining sound sensors, based on the time difference between the sound sensor and the first sound sensor in collecting the sound signal and the propagation speed of the sound signal inside the ray source box, the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor is calculated;

[0014] constructing a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the installation positions of the sound sensor and the first sound sensor;

[0015] The target position is solved based on the nonlinear equation so that the positional relationship between the installation positions of the four sound sensors satisfies the positional relationship required by the time difference of the sound signal emitted from the target position being transmitted to the four sound sensors.

[0016] Optionally, the step of calculating, for each of the remaining sound sensors, the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor based on the time difference between the sound sensor and the first sound sensor in collecting the sound signal and the propagation speed of the sound signal inside the ray source box, comprises:

[0017] For each of the remaining sound sensors, according to the time when the sound sensor and the first sound sensor collect the sound signal and the propagation speed of the sound signal inside the ray source box, the difference d between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor is calculated according to the following formula: 1j :

[0018] d 1j =v×(t 1 -t j )

[0019] Where j = 2, 3, 4, d 1j is the difference between the distance between the target position and the first sound sensor and the distance between the target position and the jth sound sensor, v is the propagation speed of sound inside the ray source box, t 1 is the time when the first sound sensor collects the sound signal, t j is the time when the j-th sound sensor collects the sound signal;

[0020] The step of constructing a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the installation positions of the sound sensor and the first sound sensor comprises:

[0021] According to the installation positions of the sound sensor and the first sound sensor, a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor is constructed according to the following formula:

[0022]

[0023] Where (x, y, z) is the target position, (x 1 ,y 1 ,z 1 ) is the installation position of the first sound sensor, (x j ,y j ,z j ) is the installation position of the jth sound sensor.

[0024] Optionally, the step of determining the working state inside the radiation source box by identifying the voiceprint feature includes:

[0025] The working state inside the radiation source box is determined based on the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to the sparking state.

[0026] Optionally, the step of determining the working state inside the radiation source box according to the similarity between the voiceprint feature and a pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or a pre-recorded sparking voiceprint feature corresponding to the sparking state comprises at least one of the following:

[0027] Calculate a first characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state; if the first characteristic distance is greater than a first preset threshold, determine that the working state inside the ray source box is the non-sparking state; if the first characteristic distance is not greater than the first preset threshold, determine that the working state inside the ray source box is the sparking state;

[0028] Calculate a second characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state; if the second characteristic distance is greater than a second preset threshold, determine that the working state inside the ray source box is the ignition state; if the second characteristic distance is not greater than the second preset threshold, determine that the working state inside the ray source box is the non-ignition state;

[0029] Calculate the third characteristic distances between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state and the ignition voiceprint feature corresponding to the ignition state respectively; if the third characteristic distance corresponding to the non-sparking voiceprint feature is greater than the third characteristic distance corresponding to the ignition voiceprint feature, determine that the working state inside the ray source box is the non-sparking state; if the third characteristic distance corresponding to the ignition voiceprint feature is greater than the third characteristic distance corresponding to the non-sparking voiceprint feature, determine that the working state inside the ray source box is the ignition state.

[0030] Optionally, the step of determining the working state inside the radiation source box by identifying the voiceprint feature includes:

[0031] The voiceprint feature is input into a pre-trained ignition state detection model, so that the ignition state detection model determines the matching probability between the voiceprint feature and the pre-learned sample voiceprint feature corresponding to the ignition state based on the model parameters obtained in pre-training, and according to the correspondence between the sample voiceprint feature corresponding to the ignition state and the ignition state, the ignition state corresponding to the sample voiceprint feature with the highest matching probability is output as the working state inside the radiation source box.

[0032] Optionally, the training method of the ignition state detection model includes:

[0033] Acquire multiple sample voiceprint features corresponding to the ignition state and an ignition state label corresponding to each sample voiceprint feature, wherein the ignition state label is used to identify that the working state corresponding to the sample voiceprint feature is the ignition state;

[0034] Input each sample voiceprint feature into the initial ignition state detection model, so that the initial ignition state detection model determines the matching probability between the sample voiceprint feature and the currently learned sample voiceprint feature corresponding to the ignition state based on the current model parameters, and outputs the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the estimated ignition state corresponding to each sample voiceprint feature according to the corresponding relationship between the currently learned sample voiceprint feature corresponding to the ignition state and the ignition state;

[0035] Calculating a loss function value based on the difference between the ignition state identified by the ignition state label corresponding to each sample voiceprint feature and the estimated ignition state;

[0036] According to the loss function value, back propagation is used to adjust the model parameters of the initial ignition state detection model until the initial ignition state detection model meets the convergence condition, thereby obtaining the ignition state detection model.

[0037] Optionally, after the step of determining a target position where the ignition phenomenon occurs inside the ray source box according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box, the method further includes:

[0038] Based on the spatial mapping relationship between the pre-built three-dimensional model and the radiation source box, determining the position corresponding to the target position in the three-dimensional model;

[0039] At the determined position, a preset ignition status indicator is displayed.

[0040] In a second aspect, an embodiment of the present application provides a device for detecting the state of a radiation source, which is applied to a processor in a detection system, wherein the detection system further includes a sound sensor, which is installed on at least three outer surfaces of a radiation source box, and the sound sensor is used to collect a sound signal emitted from inside the radiation source box, wherein the device includes:

[0041] A voiceprint feature extraction module, used to obtain the sound signal collected by the sound sensor and extract the voiceprint feature of the sound signal;

[0042] A working state determination module, used to determine the working state inside the radiation source box by identifying the voiceprint feature, wherein the working state includes a non-ignition state and an ignition state;

[0043] A target position determination module is used to determine the target position where the ignition phenomenon occurs inside the ray source box based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor and the propagation speed of the sound signal inside the ray source box when the working state is determined to be the ignition state.

[0044] Optionally, the number of the sound sensors is four;

[0045] The target position determination module comprises:

[0046] The target position determination submodule is used to determine the target position where the ignition phenomenon occurs inside the ray source box based on the time difference between the four sound sensors collecting the sound signals, the positional relationship between the installation positions of the four sound sensors, and the propagation speed of the sound signals inside the ray source box.

[0047] Optionally, the target position determination submodule includes:

[0048] a distance difference calculation unit, configured to use one of the four sound sensors as a first sound sensor, and for each of the remaining sound sensors, calculate the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor based on the time difference between the sound sensor and the first sound sensor in collecting the sound signal and the propagation speed of the sound signal inside the ray source box;

[0049] an equation constructing unit, for constructing a nonlinear equation for calculating a difference between a distance between the target position and the first sound sensor and a distance between the target position and the sound sensor according to installation positions of the sound sensor and the first sound sensor;

[0050] The target position solving unit is used to solve the target position based on the nonlinear equation so that the positional relationship between the installation positions of the four sound sensors satisfies the positional relationship required by the time difference of the sound signal emitted from the target position to be transmitted to the four sound sensors.

[0051] Optionally, the distance difference calculation unit includes:

[0052] The distance difference calculation subunit is used to calculate the difference d between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the following formula for each of the remaining sound sensors, based on the time when the sound sensor and the first sound sensor collect the sound signal and the propagation speed of the sound signal inside the ray source box: 1j :

[0053] d 1j =v×(t 1 -t j )

[0054] Where j = 2, 3, 4, d 1j is the difference between the distance between the target position and the first sound sensor and the distance between the target position and the jth sound sensor, v is the propagation speed of sound inside the ray source box, t 1 is the time when the first sound sensor collects the sound signal, t j is the time when the j-th sound sensor collects the sound signal;

[0055] The equation building unit comprises:

[0056] The equation construction subunit is used to construct a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the following formula according to the installation positions of the sound sensor and the first sound sensor:

[0057]

[0058] Where (x, y, z) is the target position, (x 1 ,y 1 ,z 1 ) is the installation position of the first sound sensor, (x j ,y j ,z j ) is the installation position of the jth sound sensor.

[0059] Optionally, the working status determination module includes:

[0060] A first working state determination submodule is used to determine the working state inside the radiation source box according to the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to the sparking state;

[0061] Optionally, the first working state determination submodule includes:

[0062] A first working state determining unit is used to calculate a first characteristic distance between the voiceprint feature and a non-sparking voiceprint feature corresponding to the non-sparking state; if the first characteristic distance is greater than a first preset threshold, it is determined that the working state inside the ray source box is the non-sparking state; if the first characteristic distance is not greater than the first preset threshold, it is determined that the working state inside the ray source box is the sparking state;

[0063] A second working state determining unit is used to calculate a second characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state; if the second characteristic distance is greater than a second preset threshold, it is determined that the working state inside the ray source box is the ignition state; if the second characteristic distance is not greater than the second preset threshold, it is determined that the working state inside the ray source box is the non-ignition state;

[0064] The third working state determination unit is used to respectively calculate the third characteristic distances between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state and the ignition voiceprint feature corresponding to the ignition state; if the third characteristic distance corresponding to the non-sparking voiceprint feature is greater than the third characteristic distance corresponding to the ignition voiceprint feature, it is determined that the working state inside the ray source box is the non-sparking state; if the third characteristic distance corresponding to the ignition voiceprint feature is greater than the third characteristic distance corresponding to the non-sparking voiceprint feature, it is determined that the working state inside the ray source box is the ignition state.

[0065] Optionally, the working status determination module includes:

[0066] The second working state determination submodule is used to input the voiceprint feature into a pre-trained ignition state detection model, so that the ignition state detection model determines the matching probability between the voiceprint feature and the pre-learned sample voiceprint feature corresponding to the ignition state based on the model parameters obtained in pre-training, and according to the correspondence between the sample voiceprint feature corresponding to the ignition state and the ignition state, outputs the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the working state inside the radiation source box.

[0067] Optionally, the device further comprises:

[0068] A sample acquisition module, used to acquire a plurality of sample voiceprint features corresponding to the ignition state and an ignition state label corresponding to each sample voiceprint feature, wherein the ignition state label is used to identify that the working state corresponding to the sample voiceprint feature is the ignition state;

[0069] An estimated state acquisition module is used to input each sample voiceprint feature into an initial ignition state detection model, so that the initial ignition state detection determines the matching probability between the sample voiceprint feature and the currently learned sample voiceprint feature corresponding to the ignition state based on the current model parameters, and outputs the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the estimated ignition state corresponding to each sample voiceprint feature according to the corresponding relationship between the currently learned sample voiceprint feature corresponding to the ignition state and the ignition state;

[0070] A loss function value calculation module, used to calculate the loss function value based on the difference between the ignition state identified by the ignition state label corresponding to each sample voiceprint feature and the estimated ignition state;

[0071] A parameter adjustment module is used to adjust the model parameters of the initial ignition state detection model by back propagation according to the loss function value until the initial ignition state detection model meets the convergence condition to obtain the ignition state detection model.

[0072] Optionally, the device further comprises:

[0073] A position determination module, used to determine the position corresponding to the target position in the three-dimensional model based on the spatial mapping relationship between the pre-built three-dimensional model and the ray source box;

[0074] The identification display module is used to display a preset ignition status identification at a determined position.

[0075] In a third aspect, an embodiment of the present application provides a system for detecting the state of a radiation source, the detection system comprising a processor and a sound sensor, the number of the sound sensors being at least four, the sound sensors being installed on at least three outer surfaces of a radiation source box,

[0076] The sound sensor is used to collect sound signals emitted from inside the radiation source box;

[0077] The processor is used to obtain the sound signal collected by the sound sensor and extract the sound print features of the sound signal; determine the working state inside the ray source box by identifying the sound print features; when it is determined that the working state is the ignition state, determine the target position where the ignition phenomenon occurs inside the ray source box according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor and the propagation speed of the sound signal inside the ray source box, wherein the working state includes a non-ignition state and an ignition state.

[0078] In a fourth aspect, an embodiment of the present application provides an electronic device, including:

[0079] Memory, used to store computer programs;

[0080] The processor is used to implement any method described in the first aspect when executing a program stored in the memory.

[0081] In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, any method described in the first aspect is implemented.

[0082] Beneficial effects of the embodiments of the present application:

[0083] In the solution provided by the embodiment of the present application, the processor can obtain the sound signal collected by the sound sensor and extract the sound print features of the sound signal; by identifying the sound print features, the working state inside the ray source box is determined, wherein the working state includes a non-sparking state and a sparking state. Since the ignition phenomenon is usually accompanied by specific sounds, such as crackling, popping, hissing, etc., and these sounds will not be emitted when the ray source is not working, the sound print features of the sound signal emitted in the non-sparking state and the sparking state inside the ray source box are different, and the sound signal inside the ray source box can be collected by the sound sensor to determine whether the working state inside the ray source is a non-sparking state or a sparking state. In the case of determining that the working state is the sparking state, the target position where the sparking phenomenon occurs inside the ray source box is determined according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. Since the sound sensor can collect sound signals on the outer surface of the ray source box without placing the sound sensor inside the ray source box, it is possible to avoid the sound sensor being broken down by the ultra-high voltage inside the ray source box, detect the working state of the ray source, and reduce the defective rate of the ray source during production and later use. Since the installation position of the sound sensor on the outer surface of the ray source box is different, the time when the sound signal accompanying the ignition phenomenon reaches the sound sensor is often different. Therefore, the processor can determine the target position based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. In this way, when it is determined that the working state is the ignition state, the target position where the ignition phenomenon occurs can be quickly and accurately located. Of course, implementing any product or method of the present application does not necessarily require all of the advantages described above to be achieved at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application, and for ordinary technicians in this field, other embodiments can also be obtained based on these drawings.

[0085] Figure 1 It is a schematic diagram of a radiation source box in the current related technology;

[0086] Figure 2 A flow chart of a method for detecting a radiation source state provided in an embodiment of the present application;

[0087] Figure 3(a) is based on Figure 2 A schematic diagram of a radiation source box of the illustrated embodiment;

[0088] Figure 3(b) is based on Figure 2 Another schematic diagram of the radiation source box of the illustrated embodiment;

[0089] Figure 4 Based on Figure 2 A flow chart of a target position solving method of the illustrated embodiment;

[0090] Figure 5 Based on Figure 2 A flow chart of a first implementation method of determining the working state of the illustrated embodiment;

[0091] Figure 6 Based on Figure 2 A flow chart of a second implementation method of determining the working state of the illustrated embodiment;

[0092] Figure 7 Based on Figure 2 A flow chart of a third implementation method of determining the working state of the illustrated embodiment;

[0093] Figure 8 Based on Figure 2 A flow chart of a training method of an ignition state detection model of the illustrated embodiment;

[0094] Fig. 9 Based on Figure 2 A flow chart showing a preset ignition status indicator of the illustrated embodiment;

[0095] Fig.10 Based on Figure 2 A schematic diagram of a three-dimensional model corresponding to the ray source box of the illustrated embodiment;

[0096] Fig.11 A schematic diagram of the structure of a device for detecting the state of a radiation source provided in an embodiment of the present application;

[0097] Fig.12 A schematic diagram of the structure of a radiation source state detection system provided in an embodiment of the present application;

[0098] Fig.13 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0099] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field based on the present application belong to the scope of protection of the present application.

[0100] In the current related technology, the schematic diagram of the radiation source box is as follows: Figure 1 As shown, Figure 1 The view in the upper left corner is a cross-section of the front view of the ray source housing. The ray tube 101 can be used to generate X-rays. The fan-shaped range above the ray tube 101 is the range that can generate X-rays. The ray source housing also includes a voltage multiplier plate 102 and a transformer 103. Figure 1 The view in the lower left corner is a top view of the ray source box, and X-rays can be emitted from the X-ray beam outlet 104 on the outer surface of the ray source box. Figure 1 The view in the lower right corner is a cross-section of the side view of the radiation source box. The radiation source box 105 can be filled with insulating oil. The radiation source box 105 includes an oil tank cover 106, a radiation tube 101, a voltage multiplier plate 102, a transformer 103 and an X-ray beam outlet 104.

[0101] In order to detect the working state of a radiation source and reduce the defect rate of the radiation source during production and later use, the embodiments of the present application provide a radiation source state detection method, detection system, device, electronic device, computer readable storage medium and computer program product. The following first introduces a radiation source state detection method provided by the embodiments of the present application.

[0102] The method for detecting the state of a ray source provided in the embodiment of the present application can be applied to any processor that needs to detect the state of a ray source, for example, a digital signal processor, a central processing unit, a field programmable gate array, etc., which is not specifically limited here. For the sake of clarity, it is subsequently referred to as a processor in this article.

[0103] like Figure 2 As shown, a method for detecting the state of a radiation source is applied to a processor in a detection system, wherein the detection system further comprises a sound sensor, wherein the sound sensor is mounted on at least three outer surfaces of a radiation source housing, and the sound sensor is used to collect sound signals emitted from inside the radiation source housing, wherein the method comprises:

[0104] S201, obtaining a sound signal collected by the sound sensor, and extracting a voiceprint feature of the sound signal;

[0105] S202, determining the working state inside the radiation source box by identifying the voiceprint feature;

[0106] Wherein, the working state includes a non-ignition state and an ignition state.

[0107] S203, when it is determined that the working state is the ignition state, determine the target position where the ignition phenomenon occurs inside the ray source box according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor and the propagation speed of the sound signal inside the ray source box.

[0108] It can be seen that in the embodiment of the present application, the processor can obtain the sound signal collected by the sound sensor and extract the sound print features of the sound signal; by identifying the sound print features, the working state inside the ray source box is determined, wherein the working state includes a non-sparking state and a sparking state. Since the ignition phenomenon is usually accompanied by specific sounds, such as crackling, popping, hissing, etc., and these sounds will not be emitted when the ray source is not working, the sound print features of the sound signal emitted from the ray source box in the non-sparking state and the sparking state are different, and the sound signal inside the ray source box can be collected by the sound sensor to determine whether the working state inside the ray source is a non-sparking state or a sparking state. In the case of determining that the working state is the sparking state, the target position where the sparking phenomenon occurs inside the ray source box is determined according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. Since the sound sensor can collect sound signals on the outer surface of the ray source box without placing the sound sensor inside the ray source box, it is possible to avoid the sound sensor being broken down by the ultra-high voltage inside the ray source box, detect the working state of the ray source, and reduce the defective rate of the ray source during production and later use. Since the installation position of the sound sensor on the outer surface of the ray source box is different, the time when the sound signal accompanying the ignition phenomenon reaches the sound sensor is often different. Therefore, the processor can determine the target position based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. In this way, when it is determined that the working state is the ignition state, the target position where the ignition phenomenon occurs can be quickly and accurately located.

[0109] When the working state of the radiation source box is a non-sparking state and a sparking state, different sounds are usually made, so the working state of the radiation source box can be determined to be a non-sparking state or a sparking state by the sound made by the radiation source box when working. The non-sparking state is a non-sparking working state in which no sparking phenomenon occurs inside the radiation source box, and the sparking state is an abnormal working state in which a sparking phenomenon occurs inside the radiation source box.

[0110] In order to collect the sound signals emitted from the inside of the ray source box, a sound sensor can be installed on the outer surface of the ray source box. The sound sensor can be an electret microphone, a MEMS (Micro-Electro-Mechanical System) microphone, or other sensor for collecting sound signals, which is not specifically limited here. For example, a schematic diagram of a ray source box can be shown in FIG3(a), where a first sound sensor 302, a second sound sensor 303, a third sound sensor 304, and a fourth sound sensor 305 are installed on the outer surface of the ray source box 301. Another schematic diagram of the ray source box can be shown in FIG3(b), and FIG3(b) is similar to FIG3(a). Figure 1 The difference is that in the view in the lower left corner of Figure 3(b), the first sound sensor 302 and the second sound sensor 303 are installed on the surface of the ray source box, and the third sound sensor and the fourth sound sensor are also installed on the outer surface of the ray source box. However, since the view in the upper left corner and the view in the lower right corner of Figure 3(b) are both cross-sections of the ray source box, the third sound sensor and the fourth sound sensor are not shown in Figure 3(b).

[0111] Since the voiceprint feature can characterize the acoustic features of the sound signal, the voiceprint features of the sound signal emitted from the ray source box are different when the working state inside the ray source box is a non-ignition state and an ignition state, so the sound signal collected by the sound sensor can be obtained, and the voiceprint feature of the sound signal can be extracted, that is, step S201 is executed. The way to extract the voiceprint feature can be to perform cepstrum analysis, linear prediction coding, and formant extraction on the sound signal, which is not specifically limited here.

[0112] For example, assuming that sound sensor 1-sound sensor 4 is installed outside the ray source box, then sound signals 1-sound signal 4 collected by sound sensor 1-sound sensor 4 can be obtained respectively, and then the voiceprint features of sound signal 1-sound signal 4 can be extracted respectively to obtain voiceprint feature 1-voiceprint feature 4.

[0113] In one embodiment, after obtaining the sound signal collected by the sound sensor, the sound signal may be preprocessed, and then the voiceprint feature may be extracted from the preprocessed sound signal. The preprocessing method may include noise reduction processing, normalization processing, etc., which are not specifically limited here.

[0114] Since the sound print features of the sound signal collected by the sound sensor are different when the working state inside the ray source box is a non-ignition state and an ignition state, the working state inside the ray source box can be determined by identifying the sound print features, that is, executing step S202. For example, the sound sensor collects sound signal 1 and sound signal 2 at 8:00:00 and 8:05:00, respectively, then the sound print features of sound signal 1 and sound signal 2 can be extracted respectively to obtain sound print feature 1 and sound print feature 2. By identifying sound print feature 1 and sound print feature 2, it can be determined that the working state inside the ray source at 8:00:00 and 8:05:00 is a non-ignition state and an ignition state, respectively.

[0115] Since four sound sensors can accurately locate the three-dimensional coordinates of the sound source, in order to accurately locate the target position where the ignition occurs, at least four sound sensors can be installed on the outer surface of the ray source box, and the sound sensors are installed on at least three outer surfaces of the ray source box. The number of sound sensors can be 4, 5, 6, etc., which is not specifically limited here.

[0116] Since the installation positions of the sound sensor on the outer surface of the ray source box are different, the time when the sound signal accompanying the ignition phenomenon reaches the sound sensor is often different. Therefore, when the working state is determined to be the ignition state, the target position where the ignition phenomenon occurs inside the ray source box can be determined based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. The specific calculation method for determining the target position can be geometric triangulation, Gauss-Newton algorithm, least squares method, etc., which is not specifically limited here.

[0117] In one embodiment, the target position can be determined based on the time when the four sound sensors collect the sound signal, the installation positions of the four sound sensors, and the propagation speed of the sound signal inside the ray source box. If the number of the sound sensors is greater than four, the four sound sensors can be any four sound sensors installed on at least three outer surfaces of the ray source box.

[0118] Of course, if the purpose is only to determine whether the working state inside the ray source box is a non-ignition state or an ignition state, without having to locate the target position where the ignition occurs, then the number of sound sensors can be at least one, and the sound sensor can be installed on at least one outer surface of the ray source box.

[0119] It can be seen that in the embodiment of the present application, since the occurrence of the ignition phenomenon is usually accompanied by specific sounds, such as crackling, popping, hissing, etc., and these sounds will not be emitted when the ray source is not working, the sound signal emitted in the non-ignition state and the ignition state inside the ray source box has different soundprint characteristics, and the sound signal inside the ray source box can be collected by the sound sensor to determine whether the working state inside the ray source is a non-ignition state or an ignition state. Since the sound sensor can collect the sound signal on the outer surface of the ray source box without placing the sound sensor inside the ray source box, it can be avoided that the sound sensor is broken down by the ultra-high voltage inside the ray source box, and the working state of the ray source can be detected, and the defective rate of the ray source during production and later use can be reduced. Since the installation position of the sound sensor on the outer surface of the ray source box is different, the time when the sound signal accompanied by the ignition phenomenon reaches the sound sensor is often different, so the processor can determine the target position according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. In this way, when it is determined that the working state is the ignition state, the target position where the ignition phenomenon occurs can be quickly and accurately located.

[0120] As an implementation of an embodiment of the present application, the number of the above-mentioned sound sensors may be four. In this case, the step of determining the target position where the ignition phenomenon occurs inside the ray source box according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box may include:

[0121] The target position where the ignition occurs inside the ray source box is determined based on the time difference between the sound signals collected by the four sound sensors, the positional relationship between the installation positions of the four sound sensors, and the propagation speed of the sound signal inside the ray source box.

[0122] Since determining the target position where the ignition phenomenon occurs in three-dimensional space requires at least the time and installation position when the three groups of sound sensors collect the sound signals, and the three groups of sound sensors need to be composed of four sound sensors, in order to use the least number of sound sensors to determine the target position, the number of the above-mentioned sound sensors can be four.

[0123] Since the time difference in collecting sound signals by the four sound sensors can reflect the distance difference between the target position and the installation positions of the four sound sensors, and the time difference in collecting sound signals by the four sound sensors can be obtained by calculation, the positional relationship between the installation positions of the four sound sensors is known. Therefore, the processor can determine the target position based on the above time difference, positional relationship and the propagation speed of the sound signal inside the radiation source box.

[0124] It can be seen that in the embodiment of the present application, the processor can determine the target position where the spark phenomenon occurs inside the ray source box according to the time difference between the four sound sensors collecting the sound signal, the positional relationship between the installation positions of the four sound sensors, and the propagation speed of the sound signal inside the ray source box. Since the time difference between the four sound sensors collecting the sound signal can reflect the distance difference between the target position and the installation positions of the four sound sensors, the target position can be determined based on the time difference, the positional relationship between the installation positions, and the propagation speed. In this way, the target position where the spark phenomenon occurs can be quickly and accurately determined using the least number of sound sensors.

[0125] As an implementation method of the present application, Figure 4 As shown, the step of determining the target position where the ignition phenomenon occurs inside the ray source box according to the time difference between the sound signals collected by the four sound sensors, the positional relationship between the installation positions of the four sound sensors, and the propagation speed of the sound signal inside the ray source box may include:

[0126] S401, using one of the four sound sensors as a first sound sensor, and for each of the remaining sound sensors, calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor based on the time difference between the sound sensor and the first sound sensor in collecting the sound signal and the propagation speed of the sound signal inside the ray source box;

[0127] In order to determine the location of the sound source in three-dimensional space, that is, the target location where the ignition occurs, it is necessary to determine the time difference of the sound signals collected by the three groups of sound sensors and the positional relationship of the installation positions. The three groups of sound sensors can be composed of: among the four sound sensors, one is used as the first sound sensor, and the first sound sensor and each of the remaining sound sensors are respectively composed of three groups of sound sensors.

[0128] The distance between the target position and a certain sound sensor is equal to the product of the time when the sound sensor collects the sound signal and the propagation speed of the sound signal inside the ray source box, and the difference in distance between the target position and two sound sensors is equal to the product of the time difference when the two sound sensors collect the sound signals and the propagation speed of the sound signal inside the ray source box.

[0129] In order to solve the difference in distance between each group of sound sensors and the target position, the processor can calculate the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor for each of the remaining sound sensors based on the time difference between the sound signal collected by the sound sensor and the first sound sensor and the propagation speed of the sound signal inside the radiation source box.

[0130] For example, assuming that the four sound sensors are sound sensor 1 to sound sensor 4, and the first sound sensor is sound sensor 1, then the three groups of sound sensors are sound sensor 1 and sound sensor 2, sound sensor 1 and sound sensor 3, and sound sensor 1 and sound sensor 4. The time when sound sensor 1 to sound sensor 4 collect sound signals is 0.0002 seconds, 0.0003 seconds, 0.0004 seconds, and 0.0001 seconds respectively, and the propagation speed of the sound signal in the ray source box is 1500 meters per second.

[0131] Next, taking sound sensor 1 and sound sensor 2 as an example, the difference in distance between the target position and the installation position of sound sensor 1 and sound sensor 2 is solved. The time difference between sound sensor 1 and sound sensor 2 collecting sound signals is 0.0001 seconds, so it can be calculated that the difference in distance between the target position and the installation position of sound sensor 1 and sound sensor 2 is 0.15 meters. Similarly, it can be calculated that the difference in distance between the target position and the installation position of sound sensor 1 and sound sensor 3 is 0.3 meters, and the difference in distance between the target position and the installation position of sound sensor 1 and sound sensor 4 is 0.15 meters.

[0132] S402, constructing a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the installation positions of the sound sensor and the first sound sensor;

[0133] Since the installation positions of the sound sensor and the first sound sensor are known, after calculating the values ​​of the distance between the target position and the first sound sensor and the difference between the distance between the target position and the sound sensor, the processor can construct a nonlinear equation for calculating the distance between the target position and the first sound sensor and the difference between the distance between the target position and the sound sensor according to the installation positions of the sound sensor and the first sound sensor. In this way, three sets of nonlinear equations for solving the target position can be obtained.

[0134] S403, solving the target position based on the nonlinear equation, so that the positional relationship between the installation positions of the four sound sensors satisfies the positional relationship required by the time difference of the sound signal emitted from the target position to be transmitted to the four sound sensors.

[0135] If the time difference between the two sound sensors collecting the sound signals is large, it means that the distance difference between the target position and the two sound sensors is large. Correspondingly, if the time difference between the two sound sensors collecting the sound signals is small, it means that the distance difference between the target position and the two sound sensors is small. Since the positional relationship between the installation positions of the four sound sensors is fixed, the processor can solve the target position based on the constructed nonlinear equation group so that the positional relationship between the installation positions of the four sound sensors meets the positional relationship required by the time difference of the sound signal transmitted from the target position to the four sound sensors.

[0136] It can be seen that in the embodiment of the present application, the processor can construct three nonlinear equations for solving the target position, and then solve the nonlinear equations together to obtain the target position. Since the target position solved can make the positional relationship between the installation positions of the four sound sensors meet the positional relationship required by the time difference of the sound signal transmitted from the target position to the four sound sensors, the target position where the ignition phenomenon occurs can be accurately determined.

[0137] As an implementation manner of an embodiment of the present application, the above step of calculating, for each of the remaining sound sensors, the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor based on the time difference between the sound sensor and the first sound sensor in collecting the sound signal and the propagation speed of the sound signal inside the ray source box, may include:

[0138] For each of the remaining sound sensors, according to the time when the sound sensor and the first sound sensor collect the sound signal and the propagation speed of the sound signal inside the ray source box, the difference d between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor is calculated according to the following formula:1j :

[0139] d 1j =v×(t 1 -t j )

[0140] t 1 is the time when the first sound sensor collects the sound signal, t j is the time when the jth sound sensor collects the sound signal, and j in the formula = 2, 3, 4, so the time when the second sensor to the fourth sensor collects the sound signal can be expressed as t 2 ,t 3 ,t 4 The first sound sensor may be any one of the above four sound sensors. After the first sound sensor is determined, the remaining three sound sensors are the second sound sensor, the third sound sensor, and the fourth sound sensor.

[0141] v is the propagation speed of sound inside the ray source box, d 1j represents the difference between the distance between the target position and the first sound sensor and the distance between the target position and the jth sound sensor, so the difference d between the distance between the target position and the first sound sensor and the distance between the target position and the second sound sensor can be determined 12 =v×(t 1 -t 2 ), the difference d between the distance between the target position and the first sound sensor and the distance between the target position and the third sound sensor 13 =v×(t 1 -t 3 ), the difference d between the distance between the target position and the first sound sensor and the distance between the target position and the fourth sound sensor 14 =v×(t 1 -t 4 ).

[0142] Accordingly, the step of constructing a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the installation positions of the sound sensor and the first sound sensor may include:

[0143] According to the installation positions of the sound sensor and the first sound sensor, a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor is constructed according to the following formula:

[0144]

[0145] (x1 ,y 1 ,z 1 ) is the installation position of the first sound sensor, (x j ,y j ,z j ) is the position of the jth sound sensor, so the installation positions of the second to fourth sensors can be expressed as (x 1 ,y 1 ,z 1 )、(x 2 ,y 2 ,z 2 )、(x 3 ,y 3 ,z 3 )、(x 4 ,y 4 ,z 4 ). Therefore, the following formula can be obtained:

[0146]

[0147] Next, a nonlinear equation group can be constructed according to the above formula to calculate the target position (x, y, z) where the spark occurs inside the ray source box.

[0148] It can be seen that in the embodiment of the present application, the processor can construct a nonlinear equation group and calculate the target position according to the time when the four sound sensors collect the sound signal, the installation position of the four sound sensors, and the propagation speed of the sound signal inside the ray source box. In this way, when troubleshooting, the problem point, that is, the target position, can be directly located, and the problem convergence is faster.

[0149] As an implementation of the embodiment of the present application, the step of determining the working state inside the radiation source box by identifying the voiceprint feature may include:

[0150] The working state inside the radiation source box is determined based on the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to the sparking state.

[0151] If the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state is high, it means that the working state inside the ray source box is likely to be the non-sparking state. Correspondingly, if the similarity between the voiceprint feature and the pre-recorded sparking voiceprint feature corresponding to the sparking state is high, it means that the working state inside the ray source box is likely to be the sparking state, so the working state inside the ray source box can be determined based on the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to the sparking state.

[0152] In one embodiment, the processor can determine the working state inside the radiation source box based only on the similarity between the voiceprint feature and a pre-recorded non-sparking voiceprint feature corresponding to a non-sparking state, or can determine the working state inside the radiation source box based only on the similarity between the voiceprint feature and a pre-recorded ignition voiceprint feature corresponding to an ignition state, or can determine the working state inside the radiation source box based on the similarity between the voiceprint feature and a pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and an ignition voiceprint feature corresponding to the ignition state.

[0153] In another embodiment, the ignition phenomenon may include multiple situations, and the sounds accompanying each situation are different. For example, arc discharge is usually accompanied by a "crackling" sound, surface discharge is usually accompanied by a "buzzing" sound, spark discharge is usually accompanied by a "click" sound, local discharge is usually accompanied by a short "hissing" sound, and corona discharge is usually accompanied by a continuous "hissing" sound. Therefore, the above-mentioned ignition state may include at least one ignition state among the arc discharge state, surface discharge state, spark discharge state, local discharge state and corona discharge state, and each ignition state corresponds to a different ignition sound pattern feature.

[0154] In this case, the processor can determine the working state inside the ray source box according to the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to each sparking state. Specifically, the processor can calculate the similarity between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state and the sparking voiceprint feature corresponding to each sparking state; and determine the state to which the corresponding voiceprint feature with higher similarity belongs as the working state inside the ray source box.

[0155] For example, assuming that voiceprint feature A and voiceprint feature B are the voiceprint features of the sound signals collected by the sound sensor at 12:00:05 and 12:00:30 respectively, the similarity between voiceprint feature A and the ignition voiceprint feature corresponding to the pre-recorded corona discharge state is the highest, and the similarity between voiceprint feature B and the ignition voiceprint feature corresponding to the pre-recorded partial discharge state is the highest, then it can be determined that the ignition states inside the radiation source box at 12:00:05 and 12:00:30 are the corona discharge state and the partial discharge state respectively.

[0156] It can be seen that in the embodiment of the present application, the processor can determine the working state inside the ray source box according to the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to the sparking state. If the similarity between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state is high, it means that the working state inside the ray source box is likely to be the non-sparking state. If the similarity between the voiceprint feature and the sparking voiceprint feature corresponding to the sparking state is high, it means that the working state inside the ray source box is likely to be the sparking state. Therefore, the working state inside the ray source box can be determined according to the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to the sparking state. In this way, the working state inside the ray source can be determined quickly and accurately.

[0157] As an implementation of an embodiment of the present application, the step of determining the working state inside the radiation source box according to the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to the sparking state may include at least one of the following:

[0158] As a first implementation manner, a first characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state is calculated; if the first characteristic distance is greater than a first preset threshold, it is determined that the working state inside the ray source box is the non-sparking state; if the first characteristic distance is not greater than the first preset threshold, it is determined that the working state inside the ray source box is the sparking state;

[0159] As a second implementation manner, a second characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state is calculated; if the second characteristic distance is greater than a second preset threshold, it is determined that the working state inside the ray source box is the ignition state; if the second characteristic distance is not greater than the second preset threshold, it is determined that the working state inside the ray source box is the non-ignition state;

[0160] As a third implementation method, the third characteristic distances between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state and the ignition voiceprint feature corresponding to the ignition state are calculated respectively; if the third characteristic distance corresponding to the non-sparking voiceprint feature is greater than the third characteristic distance corresponding to the ignition voiceprint feature, it is determined that the working state inside the radiation source box is the non-sparking state; if the third characteristic distance corresponding to the ignition voiceprint feature is greater than the third characteristic distance corresponding to the non-sparking voiceprint feature, it is determined that the working state inside the radiation source box is the ignition state.

[0161] Next, the above three implementation modes will be described respectively. A flow chart of the first implementation mode can be shown as follows: Figure 5 As shown, the following steps may be specifically included:

[0162] S501, calculating a first characteristic distance between the voiceprint feature and the non-ignition voiceprint feature corresponding to the non-ignition state; if the first characteristic distance is greater than a first preset threshold, executing step S502; if the first characteristic distance is not greater than the first preset threshold, executing step S503;

[0163] If the characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state is large, it means that the working state of the ray source box is likely to be the non-sparking state. On the contrary, if the characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state is small, it means that the working state of the ray source box is likely to be the sparking state. Then, the first characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state can be calculated, that is, step S501 is executed. Among them, the method of calculating the characteristic distance between voiceprint features can be to calculate the Euclidean distance, cosine characteristic distance, Manhattan distance, etc. between voiceprint features, which is not specifically limited here.

[0164] S502, determining that the working state inside the radiation source box is the non-ignition state;

[0165] If the first characteristic distance is greater than the first preset threshold, it means that the first characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state is large, then it can be determined that the working state inside the ray source box is the non-sparking state, that is, step S502 is executed. The first preset threshold can be set by the user according to actual detection needs, and can be 80%, 90%, 95%, etc., which are all reasonable.

[0166] S503: Determine that the working state inside the radiation source box is the ignition state.

[0167] If the first characteristic distance is not greater than the first preset threshold, it means that the first characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state is small, then it can be determined that the working state inside the ray source box is the ignition state, that is, execute step S503.

[0168] For example, assuming that the first preset threshold is 80%, voiceprint feature 1 and voiceprint feature 2 are the voiceprint features of the sound signals collected by the sound sensor at 17:00:05 and 17:00:30 respectively, and the first feature distances between voiceprint feature 1 and voiceprint feature 2 and the non-ignition voiceprint features corresponding to the pre-recorded non-ignition state are 85% and 10% respectively, then it can be determined that the working states inside the radiation source box at 17:00:05 and 17:00:30 are the non-ignition state and the ignition state respectively.

[0169] A flow chart of the second embodiment described above can be as follows: Figure 6 As shown, the following steps may be specifically included:

[0170] S601, calculating a second characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state; if the second characteristic distance is greater than a second preset threshold, executing step S602; if the second characteristic distance is not greater than the second preset threshold, executing step S603;

[0171] If the characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state is large, it means that the working state of the ray source box is likely to be the ignition state. On the contrary, if the characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state is small, it means that the working state of the ray source box is likely to be the non-ignition state. Then, the second characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state can be calculated, that is, step S601 is executed.

[0172] S602, determining that the working state inside the radiation source box is the ignition state;

[0173] If the second characteristic distance is greater than the second preset threshold, it means that the second characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state is greater, then it can be determined that the working state inside the ray source box is the ignition state, that is, step S602 is executed. The second preset threshold can be set by the user according to actual detection needs, and can be 85%, 93%, 95%, etc., which are all reasonable.

[0174] S603, determining that the working state inside the radiation source box is the non-ignition state.

[0175] If the second characteristic distance is not greater than the second preset threshold, it means that the second characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state is small, then it can be determined that the working state inside the ray source box is a non-ignition state, that is, execute step S603.

[0176] For example, assuming that the second preset threshold is 93%, voiceprint feature 3 and voiceprint feature 4 are the voiceprint features of the sound signals collected by the sound sensor at 20:30:00 and 20:30:05 respectively, and the second characteristic distances between voiceprint feature 3 and voiceprint feature 4 and the ignition voiceprint feature corresponding to the pre-recorded ignition state are 5% and 97% respectively, then it can be determined that the working states inside the radiation source box at 20:30:00 and 20:30:05 are the non-ignition state and the ignition state respectively.

[0177] A flow chart of the third embodiment can be as follows: Figure 7 As shown, the following steps may be specifically included:

[0178] S701, respectively calculating the third characteristic distances between the voiceprint feature and the non-ignition voiceprint feature corresponding to the non-ignition state and the ignition voiceprint feature corresponding to the ignition state; if the third characteristic distance corresponding to the non-ignition voiceprint feature is greater than the third characteristic distance corresponding to the ignition voiceprint feature, executing step S702; if the third characteristic distance corresponding to the ignition voiceprint feature is greater than the third characteristic distance corresponding to the non-ignition voiceprint feature, executing step S703;

[0179] If the characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state is greater, it means that the working state inside the ray source box is the non-sparking state. Correspondingly, if the characteristic distance between the voiceprint feature and the sparking voiceprint feature corresponding to the sparking state is greater, it means that the working state inside the ray source box is the sparking state. Therefore, the processor can calculate the third characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state and the sparking voiceprint feature corresponding to the sparking state, that is, execute step S701.

[0180] S702, determining that the working state inside the radiation source box is the non-sparking state;

[0181] After obtaining the third characteristic distance between the voiceprint feature and the non-ignition voiceprint feature corresponding to the non-ignition state and the ignition voiceprint feature corresponding to the ignition state, the processor can determine the state to which the voiceprint feature with the higher third characteristic distance belongs as the working state inside the ray source box. Specifically, when the third characteristic distance corresponding to the non-ignition voiceprint feature is greater than the third characteristic distance corresponding to the ignition voiceprint feature, the processor can determine that the working state inside the ray source box is the non-ignition state.

[0182] S703, determining that the working state inside the radiation source box is the ignition state.

[0183] When the third characteristic distance corresponding to the ignition voiceprint feature is greater than the third characteristic distance corresponding to the non-ignition voiceprint feature, the processor may determine that the working state inside the ray source box is the ignition state.

[0184] For example, assuming that voiceprint feature 5 and voiceprint feature 6 are the voiceprint features of the sound signals collected by the sound sensor at 9:20:30 and 9:20:35 respectively, the third characteristic distances between voiceprint feature 5 and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and the ignition voiceprint feature corresponding to the ignition state are 23% and 81% respectively, and the third characteristic distances between voiceprint feature 6 and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and the ignition voiceprint feature corresponding to the ignition state are 94% and 35% respectively, then it can be determined that the working states inside the radiation source box at 9:20:30 and 9:20:35 are the ignition state and the non-sparking state respectively.

[0185] It can be seen that in the embodiment of the present application, the processor can use any of the above three methods to determine the working state inside the ray source box. Since the non-sparking voiceprint feature corresponding to the non-sparking state and the sparking voiceprint feature corresponding to the sparking state correspond to the sound signals collected when the working state inside the ray source is the non-sparking state and the sparking state, respectively, the characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state and the sparking voiceprint feature corresponding to the sparking state can be calculated, so as to quickly and accurately determine the working state inside the ray source.

[0186] As an implementation of the embodiment of the present application, the step of determining the working state inside the radiation source box by identifying the voiceprint feature may include:

[0187] The voiceprint feature is input into a pre-trained ignition state detection model, so that the ignition state detection model determines the matching probability between the voiceprint feature and the pre-learned sample voiceprint feature corresponding to the ignition state based on the model parameters obtained in pre-training, and according to the correspondence between the sample voiceprint feature corresponding to the ignition state and the ignition state, the ignition state corresponding to the sample voiceprint feature with the highest matching probability is output as the working state inside the radiation source box.

[0188] In order to determine the working state inside the ray source box, an ignition state detection model can be pre-trained, and the ignition state detection model can learn the corresponding relationship between the sample voiceprint features corresponding to the ignition state and the ignition state. When the voiceprint features need to be identified, the voiceprint features can be input into the ignition state detection model, and the ignition state detection model can determine the matching probability between the voiceprint features and the pre-learned sample voiceprint features corresponding to the ignition state based on the model parameters.

[0189] The matching probability can characterize the matching degree between the voiceprint feature and the sample voiceprint feature. If the matching probability is high, it means that the matching degree between the voiceprint feature and the sample voiceprint feature is high. Correspondingly, if the matching probability is low, it means that the matching degree between the voiceprint feature and the sample voiceprint feature is low. According to the above correspondence, the processor can output the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the working state inside the output ray source box. If the matching probability between the voiceprint feature and each sample voiceprint feature is low, then the working state inside the ray source box corresponding to the sample feature can be output as a non-sparking state. Among them, the above ignition state detection model can be a convolutional neural network, a recurrent neural network, a time-delay neural network, etc., which is not specifically limited here.

[0190] For example, assuming that voiceprint feature A and voiceprint feature B are the voiceprint features of sound signals collected by the sound sensor at 14:10:25 and 14:10:35 respectively, the processor can input voiceprint feature A and voiceprint feature B into a pre-trained ignition state detection model, and the ignition state detection model can determine the matching probability between voiceprint feature A and voiceprint feature B and the sample voiceprint features corresponding to the pre-learned ignition state based on the model parameters obtained in pre-training, and output the ignition state corresponding to the sample voiceprint feature with the highest matching probability according to the correspondence between the sample voiceprint features corresponding to the ignition state and the ignition state, so as to obtain the working states inside the radiation source box at 14:10:25 and 14:10:35 as the non-ignition state and the ignition state respectively.

[0191] In one embodiment, when the ignition state includes at least one of an arc discharge state, a surface discharge state, a spark discharge state, a partial discharge state, and a corona discharge state, and the ignition voiceprint features corresponding to each ignition state are different, the ignition state detection model can pre-learn the correspondence between the sample voiceprint features corresponding to each ignition state and the ignition state.

[0192] When it is necessary to identify the voiceprint features, the processor can input the voiceprint features into a pre-trained ignition state detection model, so that the ignition state detection model determines the matching probability between the voiceprint features and the sample voiceprint features corresponding to each ignition state learned in advance based on the pre-trained model parameters, and outputs the ignition state corresponding to the sample voiceprint features with the highest matching probability as the working state inside the radiation source box according to the correspondence between the sample voiceprint features corresponding to each ignition state and the ignition state.

[0193] It can be seen that in the embodiment of the present application, the processor can input the voiceprint feature into the pre-trained ignition state detection model, so that the ignition state detection model can determine the matching probability between the voiceprint feature and the sample voiceprint feature corresponding to the pre-learned ignition state based on the model parameters obtained by pre-training, and output the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the working state inside the ray source box according to the corresponding relationship between the sample voiceprint feature corresponding to the ignition state and the ignition state. Since the ignition state detection model has pre-learned the corresponding relationship between the sample voiceprint feature corresponding to the ignition state and the ignition state, when it is necessary to identify the voiceprint feature, the processor can input the voiceprint feature into the ignition state detection model, and the ignition state detection model can output the working state inside the ray source box based on the voiceprint feature and the above-mentioned corresponding relationship. In this way, the working state inside the ray source box can be quickly and accurately determined through the ignition state detection model.

[0194] As an implementation method of the present application, Figure 8 As shown, the training method of the above ignition state detection model may include:

[0195] S801, obtaining a plurality of sample voiceprint features corresponding to the ignition state and an ignition state label corresponding to each sample voiceprint feature;

[0196] In the training phase of the ignition state detection model, in order to enable the initial ignition state detection model to learn the correspondence between the sample voiceprint features corresponding to the ignition state and the ignition state, the processor can obtain multiple sample voiceprint features corresponding to the ignition state and the ignition state label corresponding to each sample voiceprint feature. The ignition state label is used to identify the working state corresponding to the sample voiceprint feature as the ignition state.

[0197] S802, inputting each of the sample voiceprint features into an initial ignition state detection model, so that the initial ignition state detection model determines the matching probability between the sample voiceprint features and the currently learned sample voiceprint features corresponding to the ignition state based on the current model parameters, and outputs the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the estimated ignition state corresponding to each sample voiceprint feature according to the corresponding relationship between the currently learned sample voiceprint features corresponding to the ignition state and the ignition state;

[0198] Next, the processor can input each sample voiceprint feature into the initial ignition state detection model, and the initial ignition state detection model can determine the matching probability between the sample voiceprint feature and the sample voiceprint feature corresponding to the currently learned ignition state based on the current model parameters of the initial ignition state detection model, and output the ignition state corresponding to the sample voiceprint feature with the highest matching probability according to the corresponding relationship between the sample voiceprint feature corresponding to the currently learned ignition state and the ignition state as the estimated ignition state corresponding to each sample voiceprint feature. Then, the processor can obtain the estimated ignition state corresponding to each sample voiceprint feature.

[0199] S803, calculating a loss function value based on the difference between the ignition state identified by the ignition state label corresponding to each sample voiceprint feature and the estimated ignition state;

[0200] Since the estimated ignition state corresponding to each sample voiceprint feature is predicted by the initial ignition state detection model, and the ignition state label corresponding to each sample voiceprint feature is the actual working state inside the radiation source box, there is usually a certain difference between the two. Therefore, the loss function value can be calculated based on the difference between the ignition state label corresponding to each sample voiceprint feature and the estimated ignition state.

[0201] S804, adjusting the model parameters of the initial ignition state detection model by back propagation according to the loss function value until the initial ignition state detection model meets the convergence condition, thereby obtaining the ignition state detection model.

[0202] After calculating the loss function value, the processor can adjust the model parameters of the initial ignition state detection model by back propagation according to the loss function value until the initial ignition state detection model meets the convergence conditions to obtain the ignition state detection model.

[0203] Specifically, when adjusting the model parameters of the initial ignition state detection model, the processor can use the direction of decreasing the loss function value as the optimization direction, and adjust the model parameters of the initial ignition state detection model by back propagation. Among them, the model parameters of the initial ignition state detection model can be adjusted by using a gradient descent algorithm, a stochastic gradient descent algorithm, etc., which are not specifically limited here.

[0204] For example, the processor may obtain sample voiceprint features 1-sample voiceprint features 3, and corresponding state labels 1-state labels 3, where state labels 1-state labels 3 are all ignition states. The processor may input sample voiceprint features 1-sample voiceprint features 3 into the initial ignition state detection model, and the initial ignition state detection model may output estimated states 1-estimated states 3 corresponding to sample voiceprint features 1-sample voiceprint features 3.

[0205] Next, the processor can calculate the loss function value 1-loss function value 3 based on the difference between state label 1 and estimated state 1, the difference between state label 2 and estimated state 2, and the difference between state label 3 and estimated state 3, and adjust the model parameters of the initial ignition state detection model through back propagation according to the loss function value 1-loss function value 3 until the initial ignition state detection model meets the convergence conditions, thereby obtaining the ignition state detection model.

[0206] It can be seen that in the embodiment of the present application, the processor can obtain multiple sample voiceprint features corresponding to the ignition state and the ignition state label corresponding to each sample voiceprint feature; input each sample voiceprint feature into the initial ignition state detection model to obtain the estimated ignition state corresponding to each sample voiceprint feature determined by the initial ignition state detection model; calculate the loss function value based on the difference between the ignition state label corresponding to each sample voiceprint feature and the estimated ignition state; according to the loss function value, adjust the model parameters of the initial ignition state detection model by back propagation until the initial ignition state detection model meets the convergence condition, and obtain the ignition state detection model. In this way, the initial ignition state detection model can be efficiently trained, and then the ignition state detection model that can accurately determine the internal working state of the ray source box according to the voiceprint features can be obtained.

[0207] As an implementation method of the present application, Fig. 9 As shown, after the step of determining the target position where the ignition phenomenon occurs inside the ray source box according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor and the propagation speed of the sound signal inside the ray source box, the method may further include:

[0208] S901, determining a position corresponding to the target position in the three-dimensional model based on a spatial mapping relationship between the pre-built three-dimensional model and the ray source box;

[0209] In order to realize the visualization of the target position, a three-dimensional model corresponding to the ray source box can be pre-built. When the target position is determined, the processor can determine the position corresponding to the target position in the three-dimensional model based on the spatial mapping relationship between the pre-built three-dimensional model and the ray source box.

[0210] S902, displaying a preset ignition status indicator at the determined position.

[0211] After determining the position corresponding to the target position in the three-dimensional model, the processor can display a preset ignition status mark at the determined position. For example, the preset ignition status mark can be a circle pattern, a triangle pattern, a five-pointed star pattern, etc., which is not specifically limited here. In this way, the user can accurately know the target position where the ignition phenomenon occurs inside the ray source box through the preset ignition status mark position displayed in the three-dimensional model, and then formulate a solution for solving the ignition state in a targeted manner.

[0212] For example, a schematic diagram of the three-dimensional model corresponding to the ray source box can be as follows: Fig.10 As shown, Fig.10 The left figure is a schematic diagram of the three-dimensional model corresponding to the ray source box. When the target position is determined, the schematic diagram of the three-dimensional model can be as follows Fig.10 As shown in the right figure, a preset ignition status mark of a triangular pattern can be displayed at the position corresponding to the target position in the three-dimensional model.

[0213] It can be seen that in the embodiment of the present application, the processor can determine the position corresponding to the target position in the three-dimensional model based on the spatial mapping relationship between the pre-built three-dimensional model and the ray source box; and display the preset ignition state mark at the determined position. Since the preset ignition state mark can be displayed at the target position determined in the three-dimensional model, the user can intuitively see the target position where the ignition phenomenon occurs inside the ray source box, and then formulate a solution for solving the ignition state in a targeted manner.

[0214] In the technical solution of this application, the operations involved in obtaining, storing, using, processing, transmitting, providing and disclosing user personal information are all carried out with the user's authorization.

[0215] Corresponding to the above-mentioned method for detecting the state of a radiation source, an embodiment of the present application further provides a device for detecting the state of a radiation source. The following is an introduction to a device for detecting the state of a radiation source provided in an embodiment of the present application.

[0216] like Fig.11 As shown, a device for detecting the state of a radiation source is applied to a processor in a detection system, wherein the detection system further comprises a sound sensor, wherein the sound sensor is mounted on the outer surface of a radiation source box, and the sound sensor is used to collect a sound signal emitted from the inside of the radiation source box. The device comprises:

[0217] The voiceprint feature extraction module 1101 is used to obtain the sound signal collected by the sound sensor and extract the voiceprint feature of the sound signal;

[0218] A working state determination module 1102 is used to determine the working state inside the radiation source box by identifying the voiceprint feature, wherein the working state includes a non-ignition state and an ignition state;

[0219] The target position determination module 1103 is used to determine the target position where the ignition phenomenon occurs inside the ray source box based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor and the propagation speed of the sound signal inside the ray source box when the working state is determined to be the ignition state.

[0220] It can be seen that in the embodiment of the present application, the processor can obtain the sound signal collected by the sound sensor and extract the sound print features of the sound signal; by identifying the sound print features, the working state inside the ray source box is determined, wherein the working state includes a non-sparking state and a sparking state. Since the ignition phenomenon is usually accompanied by specific sounds, such as crackling, popping, hissing, etc., and these sounds will not be emitted when the ray source is not working, the sound print features of the sound signal emitted from the ray source box in the non-sparking state and the sparking state are different, and the sound signal inside the ray source box can be collected by the sound sensor to determine whether the working state inside the ray source is a non-sparking state or a sparking state. In the case of determining that the working state is the sparking state, the target position where the sparking phenomenon occurs inside the ray source box is determined according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. Since the sound sensor can collect sound signals on the outer surface of the ray source box without placing the sound sensor inside the ray source box, it is possible to avoid the sound sensor being broken down by the ultra-high voltage inside the ray source box, detect the working state of the ray source, and reduce the defective rate of the ray source during production and later use. Since the installation position of the sound sensor on the outer surface of the ray source box is different, the time when the sound signal accompanying the ignition phenomenon reaches the sound sensor is often different. Therefore, the processor can determine the target position based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. In this way, when it is determined that the working state is the ignition state, the target position where the ignition phenomenon occurs can be quickly and accurately located.

[0221] As an implementation of the embodiment of the present application, the number of the sound sensors may be four. In this case, the target position determination module 1103 may include:

[0222] The target position determination submodule is used to determine the target position where the ignition phenomenon occurs inside the ray source box based on the time difference between the four sound sensors collecting the sound signals, the positional relationship between the installation positions of the four sound sensors, and the propagation speed of the sound signals inside the ray source box.

[0223] As an implementation of an embodiment of the present application, the above-mentioned target position determination submodule may include:

[0224] a distance difference calculation unit, configured to use one of the four sound sensors as a first sound sensor, and for each of the remaining sound sensors, calculate the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor based on the time difference between the sound sensor and the first sound sensor in collecting the sound signal and the propagation speed of the sound signal inside the ray source box;

[0225] an equation constructing unit, for constructing a nonlinear equation for calculating a difference between a distance between the target position and the first sound sensor and a distance between the target position and the sound sensor according to installation positions of the sound sensor and the first sound sensor;

[0226] The target position solving unit is used to solve the target position based on the nonlinear equation so that the positional relationship between the installation positions of the four sound sensors satisfies the positional relationship required by the time difference of the sound signal emitted from the target position to be transmitted to the four sound sensors.

[0227] As an implementation of an embodiment of the present application, the distance difference calculation unit may include:

[0228] The distance difference calculation subunit is used to calculate the difference d between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the following formula for each of the remaining sound sensors, based on the time when the sound sensor and the first sound sensor collect the sound signal and the propagation speed of the sound signal inside the ray source box: 1j :

[0229] d 1j =v×(t 1 -t j )

[0230] Where j = 2, 3, 4, d 1j is the difference between the distance between the target position and the first sound sensor and the distance between the target position and the jth sound sensor, v is the propagation speed of sound inside the ray source box, t 1is the time when the first sound sensor collects the sound signal, t j is the time when the j-th sound sensor collects the sound signal;

[0231] The above equation building unit may include:

[0232] The equation construction subunit is used to construct a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the following formula according to the installation positions of the sound sensor and the first sound sensor:

[0233]

[0234] Where (x, y, z) is the target position, (x 1 ,y 1 ,z 1 ) is the installation position of the first sound sensor, (x j ,y j ,z j ) is the installation position of the jth sound sensor.

[0235] As an implementation of the embodiment of the present application, the working status determination module 1102 may include:

[0236] The first working state determination submodule is used to determine the working state inside the radiation source box according to the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the ignition voiceprint feature corresponding to the ignition state.

[0237] As an implementation of the embodiment of the present application, the first working state determination submodule may include:

[0238] A first working state determining unit is used to calculate a first characteristic distance between the voiceprint feature and a non-sparking voiceprint feature corresponding to the non-sparking state; if the first characteristic distance is greater than a first preset threshold, it is determined that the working state inside the ray source box is the non-sparking state; if the first characteristic distance is not greater than the first preset threshold, it is determined that the working state inside the ray source box is the sparking state;

[0239] A second working state determining unit is used to calculate a second characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state; if the second characteristic distance is greater than a second preset threshold, it is determined that the working state inside the ray source box is the ignition state; if the second characteristic distance is not greater than the second preset threshold, it is determined that the working state inside the ray source box is the non-ignition state;

[0240] The third working state determination unit is used to respectively calculate the third characteristic distances between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state and the ignition voiceprint feature corresponding to the ignition state; if the third characteristic distance corresponding to the non-sparking voiceprint feature is greater than the third characteristic distance corresponding to the ignition voiceprint feature, it is determined that the working state inside the ray source box is the non-sparking state; if the third characteristic distance corresponding to the ignition voiceprint feature is greater than the third characteristic distance corresponding to the non-sparking voiceprint feature, it is determined that the working state inside the ray source box is the ignition state.

[0241] As an implementation of the embodiment of the present application, the working status determination module 1102 may include:

[0242] The second working state determination submodule is used to input the voiceprint feature into a pre-trained ignition state detection model, so that the ignition state detection model determines the matching probability between the voiceprint feature and the pre-learned sample voiceprint feature corresponding to the ignition state based on the model parameters obtained in pre-training, and according to the correspondence between the sample voiceprint feature corresponding to the ignition state and the ignition state, outputs the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the working state inside the radiation source box.

[0243] As an implementation of the embodiment of the present application, the above device may further include:

[0244] A sample acquisition module, used to acquire a plurality of sample voiceprint features corresponding to the ignition state and an ignition state label corresponding to each sample voiceprint feature, wherein the ignition state label is used to identify that the working state corresponding to the sample voiceprint feature is the ignition state;

[0245] An estimated state acquisition module is used to input each sample voiceprint feature into an initial ignition state detection model, so that the initial ignition state detection determines the matching probability between the sample voiceprint feature and the currently learned sample voiceprint feature corresponding to the ignition state based on the current model parameters, and outputs the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the estimated ignition state corresponding to each sample voiceprint feature according to the corresponding relationship between the currently learned sample voiceprint feature corresponding to the ignition state and the ignition state;

[0246] A loss function value calculation module, used to calculate the loss function value based on the difference between the ignition state identified by the ignition state label corresponding to each sample voiceprint feature and the estimated ignition state;

[0247] A parameter adjustment module is used to adjust the model parameters of the initial ignition state detection model by back propagation according to the loss function value until the initial ignition state detection model meets the convergence condition to obtain the ignition state detection model.

[0248] As an implementation of the embodiment of the present application, the above device may further include:

[0249] A position determination module, used to determine the position corresponding to the target position in the three-dimensional model based on the spatial mapping relationship between the pre-built three-dimensional model and the ray source box;

[0250] The identification display module is used to display a preset ignition status identification at a determined position.

[0251] Corresponding to the above-mentioned method for detecting the state of a radiation source, an embodiment of the present application further provides a system for detecting the state of a radiation source. The following is an introduction to a system for detecting the state of a radiation source provided in an embodiment of the present application.

[0252] like Fig.12 As shown, a detection system for a radiation source state includes a processor 1201 and a sound sensor 1202. The sound sensor 1202 is installed on at least three outer surfaces of a radiation source box.

[0253] The sound sensor 1202 is used to collect sound signals emitted from inside the ray source box;

[0254] The processor 1201 is used to obtain the sound signal collected by the sound sensor 1202 and extract the voiceprint features of the sound signal; determine the working state inside the ray source box by identifying the voiceprint features; when it is determined that the working state is the ignition state, determine the target position where the ignition phenomenon occurs inside the ray source box according to the time when the sound sensor 1202 collects the sound signal, the installation position of the sound sensor 1202 and the propagation speed of the sound signal inside the ray source box, wherein the working state includes a non-ignition state and an ignition state.

[0255] It can be seen that in the embodiment of the present application, the processor can obtain the sound signal collected by the sound sensor and extract the sound print features of the sound signal; by identifying the sound print features, the working state inside the ray source box is determined, wherein the working state includes a non-sparking state and a sparking state. Since the ignition phenomenon is usually accompanied by specific sounds, such as crackling, popping, hissing, etc., and these sounds will not be emitted when the ray source is not working, the sound print features of the sound signal emitted from the ray source box in the non-sparking state and the sparking state are different, and the sound signal inside the ray source box can be collected by the sound sensor to determine whether the working state inside the ray source is a non-sparking state or a sparking state. In the case of determining that the working state is the sparking state, the target position where the sparking phenomenon occurs inside the ray source box is determined according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. Since the sound sensor can collect sound signals on the outer surface of the ray source box without placing the sound sensor inside the ray source box, it is possible to avoid the sound sensor being broken down by the ultra-high voltage inside the ray source box, detect the working state of the ray source, and reduce the defective rate of the ray source during production and later use. Since the installation position of the sound sensor on the outer surface of the ray source box is different, the time when the sound signal accompanying the ignition phenomenon reaches the sound sensor is often different. Therefore, the processor can determine the target position based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. In this way, when it is determined that the working state is the ignition state, the target position where the ignition phenomenon occurs can be quickly and accurately located.

[0256] The present application also provides an electronic device, such as Fig.13 As shown, including:

[0257] Memory 1301, used for storing computer programs;

[0258] The processor 1302 is used to implement the method for detecting the state of the radiation source described in any of the above embodiments when executing the program stored in the memory 1301.

[0259] Furthermore, the electronic device may further include a communication bus and / or a communication interface, and the processor 1302, the communication interface, and the memory 1301 communicate with each other via the communication bus.

[0260] It can be seen that in the embodiment of the present application, the electronic device can obtain the sound signal collected by the sound sensor and extract the sound print features of the sound signal; by identifying the sound print features, the working state inside the ray source box is determined, wherein the working state includes a non-sparking state and a sparking state. Since the sparking phenomenon is usually accompanied by specific sounds, such as crackling, popping, hissing, etc., and these sounds will not be emitted when the ray source is not working, the sound print features of the sound signal emitted in the non-sparking state and the sparking state inside the ray source box are different, and the sound signal inside the ray source box can be collected by the sound sensor to determine whether the working state inside the ray source is a non-sparking state or a sparking state. In the case of determining that the working state is the sparking state, the target position where the sparking phenomenon occurs inside the ray source box is determined according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. Since the sound sensor can collect sound signals on the outer surface of the ray source box without placing the sound sensor inside the ray source box, it can avoid the sound sensor being broken down by the ultra-high voltage inside the ray source box, and can detect the working state of the ray source, thereby reducing the defective rate of the ray source during production and later use. Since the installation position of the sound sensor on the outer surface of the ray source box is different, the time when the sound signal accompanying the ignition phenomenon reaches the sound sensor is often different. Therefore, the electronic device can determine the target position based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box. In this way, when it is determined that the working state is the ignition state, the target position where the ignition phenomenon occurs can be quickly and accurately located.

[0261] The communication bus mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0262] The communication interface is used for communication between the above electronic device and other devices.

[0263] The memory may include a random access memory (RAM) or a non-volatile memory (NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0264] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0265] In another embodiment provided in the present application, a computer-readable storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method for detecting the state of any radiation source are implemented.

[0266] In another embodiment provided in the present application, a computer program product including instructions is also provided, which, when executed on a computer, enables the computer to execute any method for detecting the state of a ray source in the above embodiments.

[0267] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a solid-state hard disk (SSD), etc.

[0268] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0269] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, detection system, electronic device, computer-readable storage medium, and computer program product, since they are basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0270] The above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application are included in the protection scope of the present application.

Claims

1. A method for detecting a radiation source state, characterized in that: A processor applied to a detection system, wherein the detection system further comprises a sound sensor, wherein the sound sensor is mounted on at least three outer surfaces of a ray source box, and the sound sensor is used to collect a sound signal emitted from inside the ray source box. The method comprises: Acquire the sound signal collected by the sound sensor, and extract the voiceprint feature of the sound signal; By identifying the voiceprint feature, determining the working state inside the radiation source box, wherein the working state includes a non-ignition state and an ignition state; When it is determined that the working state is the ignition state, the target position where the ignition occurs inside the ray source box is determined based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor and the propagation speed of the sound signal inside the ray source box.

2. The method according to claim 1, characterized in that The number of the sound sensors is four; The step of determining the target position where the spark occurs inside the ray source box according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box includes: The target position where the ignition occurs inside the ray source box is determined based on the time difference between the sound signals collected by the four sound sensors, the positional relationship between the installation positions of the four sound sensors, and the propagation speed of the sound signal inside the ray source box.

3. The method according to claim 2, characterized in that The step of determining the target position where the spark phenomenon occurs inside the ray source box according to the time difference between the sound signals collected by the four sound sensors, the positional relationship between the installation positions of the four sound sensors, and the propagation speed of the sound signal inside the ray source box includes: Among the four sound sensors, one is used as a first sound sensor, and for each of the remaining sound sensors, based on the time difference between the sound sensor and the first sound sensor in collecting the sound signal and the propagation speed of the sound signal inside the ray source box, the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor is calculated; constructing a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the installation positions of the sound sensor and the first sound sensor; The target position is solved based on the nonlinear equation so that the positional relationship between the installation positions of the four sound sensors satisfies the positional relationship required by the time difference of the sound signal emitted from the target position being transmitted to the four sound sensors.

4. The method according to claim 3, characterized in that The step of calculating, for each of the remaining sound sensors, the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor based on the time difference between the sound sensor and the first sound sensor in collecting the sound signal and the propagation speed of the sound signal inside the ray source box, comprises: For each of the remaining sound sensors, according to the time when the sound sensor and the first sound sensor collect the sound signal and the propagation speed of the sound signal inside the ray source box, the difference d between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor is calculated according to the following formula: 1j : d 1j =v×(t1-t j ) Where j = 2, 3, 4, d 1j is the difference between the distance between the target position and the first sound sensor and the distance between the target position and the jth sound sensor, v is the propagation speed of sound inside the ray source box, t1 is the time when the first sound sensor collects the sound signal, t j is the time when the j-th sound sensor collects the sound signal; The step of constructing a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the installation positions of the sound sensor and the first sound sensor comprises: According to the installation positions of the sound sensor and the first sound sensor, a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor is constructed according to the following formula: Wherein, (x, y, z) is the target position, (x1, y1, z1) is the installation position of the first sound sensor, (x j ,y j ,z j ) is the installation position of the jth sound sensor.

5. The method according to any one of claims 1 to 4, characterized in that: The step of determining the working state inside the radiation source box by identifying the voiceprint feature includes: The working state inside the radiation source box is determined based on the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to the sparking state.

6. The method according to claim 5, characterized in that The step of determining the working state inside the radiation source box according to the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to the sparking state comprises at least one of the following: Calculate a first characteristic distance between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state; if the first characteristic distance is greater than a first preset threshold, determine that the working state inside the ray source box is the non-sparking state; if the first characteristic distance is not greater than the first preset threshold, determine that the working state inside the ray source box is the sparking state; Calculate a second characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state; if the second characteristic distance is greater than a second preset threshold, determine that the working state inside the ray source box is the ignition state; if the second characteristic distance is not greater than the second preset threshold, determine that the working state inside the ray source box is the non-ignition state; Calculate the third characteristic distances between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state and the ignition voiceprint feature corresponding to the ignition state respectively; if the third characteristic distance corresponding to the non-sparking voiceprint feature is greater than the third characteristic distance corresponding to the ignition voiceprint feature, determine that the working state inside the ray source box is the non-sparking state; if the third characteristic distance corresponding to the ignition voiceprint feature is greater than the third characteristic distance corresponding to the non-sparking voiceprint feature, determine that the working state inside the ray source box is the ignition state.

7. The method according to any one of claims 1 to 4, characterized in that: The step of determining the working state inside the radiation source box by identifying the voiceprint feature includes: The voiceprint feature is input into a pre-trained ignition state detection model, so that the ignition state detection model determines the matching probability between the voiceprint feature and the pre-learned sample voiceprint feature corresponding to the ignition state based on the model parameters obtained in pre-training, and according to the correspondence between the sample voiceprint feature corresponding to the ignition state and the ignition state, the ignition state corresponding to the sample voiceprint feature with the highest matching probability is output as the working state inside the radiation source box.

8. The method according to claim 7, characterized in that The training method of the ignition state detection model includes: Acquire multiple sample voiceprint features corresponding to the ignition state and an ignition state label corresponding to each sample voiceprint feature, wherein the ignition state label is used to identify that the working state corresponding to the sample voiceprint feature is the ignition state; Input each sample voiceprint feature into the initial ignition state detection model, so that the initial ignition state detection model determines the matching probability between the sample voiceprint feature and the currently learned sample voiceprint feature corresponding to the ignition state based on the current model parameters, and outputs the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the estimated ignition state corresponding to each sample voiceprint feature according to the corresponding relationship between the currently learned sample voiceprint feature corresponding to the ignition state and the ignition state; Calculating a loss function value based on the difference between the ignition state identified by the ignition state label corresponding to each sample voiceprint feature and the estimated ignition state; According to the loss function value, back propagation is used to adjust the model parameters of the initial ignition state detection model until the initial ignition state detection model meets the convergence condition, thereby obtaining the ignition state detection model.

9. The method according to any one of claims 1 to 4, characterized in that: After the step of determining the target position where the spark occurs inside the ray source box according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor, and the propagation speed of the sound signal inside the ray source box, the method further includes: Based on the spatial mapping relationship between the pre-built three-dimensional model and the radiation source box, determining the position corresponding to the target position in the three-dimensional model; At the determined position, a preset ignition status indicator is displayed.

10. A device for detecting the state of a radiation source, characterized in that: A processor used in a detection system, wherein the detection system further comprises a sound sensor, wherein the sound sensor is mounted on at least three outer surfaces of a ray source box, and the sound sensor is used to collect sound signals emitted from inside the ray source box. The device comprises: A voiceprint feature extraction module, used to obtain the sound signal collected by the sound sensor and extract the voiceprint feature of the sound signal; A working state determination module, used to determine the working state inside the radiation source box by identifying the voiceprint feature, wherein the working state includes a non-ignition state and an ignition state; A target position determination module is used to determine the target position where the ignition phenomenon occurs inside the ray source box based on the time when the sound sensor collects the sound signal, the installation position of the sound sensor and the propagation speed of the sound signal inside the ray source box when the working state is determined to be the ignition state.

11. The device according to claim 10, characterized in that The number of the sound sensors is four; The target position determination module comprises: A target position determination submodule, for determining a target position where the spark phenomenon occurs inside the ray source box according to the time difference between the sound signals collected by the four sound sensors, the positional relationship between the installation positions of the four sound sensors, and the propagation speed of the sound signal inside the ray source box; The target position determination submodule includes: a distance difference calculation unit, configured to use one of the four sound sensors as a first sound sensor, and for each of the remaining sound sensors, calculate the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor based on the time difference between the sound sensor and the first sound sensor in collecting the sound signal and the propagation speed of the sound signal inside the ray source box; an equation constructing unit, for constructing a nonlinear equation for calculating a difference between a distance between the target position and the first sound sensor and a distance between the target position and the sound sensor according to installation positions of the sound sensor and the first sound sensor; a target position solving unit, configured to solve the target position based on the nonlinear equation so that the positional relationship between the installation positions of the four sound sensors satisfies the positional relationship required by the time difference of the sound signal emitted from the target position being transmitted to the four sound sensors; The distance difference calculation unit comprises: The distance difference calculation subunit is used to calculate the difference d between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the following formula for each of the remaining sound sensors, based on the time when the sound sensor and the first sound sensor collect the sound signal and the propagation speed of the sound signal inside the ray source box: 1j : d 1j =v×(t1-t j ) Where j = 2, 3, 4, d 1j is the difference between the distance between the target position and the first sound sensor and the distance between the target position and the jth sound sensor, v is the propagation speed of sound inside the ray source box, t1 is the time when the first sound sensor collects the sound signal, t j is the time when the j-th sound sensor collects the sound signal; The equation building unit comprises: The equation construction subunit is used to construct a nonlinear equation for calculating the difference between the distance between the target position and the first sound sensor and the distance between the target position and the sound sensor according to the following formula according to the installation positions of the sound sensor and the first sound sensor: Wherein, (x, y, z) is the target position, (x1, y1, z1) is the installation position of the first sound sensor, (x j ,y j ,z j ) is the installation position of the j-th sound sensor; The working status determination module comprises: A first working state determination submodule is used to determine the working state inside the radiation source box according to the similarity between the voiceprint feature and the pre-recorded non-sparking voiceprint feature corresponding to the non-sparking state and / or the sparking voiceprint feature corresponding to the sparking state; The first working state determination submodule includes: A first working state determining unit is used to calculate a first characteristic distance between the voiceprint feature and a non-sparking voiceprint feature corresponding to the non-sparking state; if the first characteristic distance is greater than a first preset threshold, it is determined that the working state inside the ray source box is the non-sparking state; if the first characteristic distance is not greater than the first preset threshold, it is determined that the working state inside the ray source box is the sparking state; A second working state determining unit is used to calculate a second characteristic distance between the voiceprint feature and the ignition voiceprint feature corresponding to the ignition state; if the second characteristic distance is greater than a second preset threshold, it is determined that the working state inside the ray source box is the ignition state; if the second characteristic distance is not greater than the second preset threshold, it is determined that the working state inside the ray source box is the non-ignition state; The third working state determination unit is used to respectively calculate the third characteristic distances between the voiceprint feature and the non-sparking voiceprint feature corresponding to the non-sparking state and the ignition voiceprint feature corresponding to the ignition state; if the third characteristic distance corresponding to the non-sparking voiceprint feature is greater than the third characteristic distance corresponding to the ignition voiceprint feature, it is determined that the working state inside the ray source box is the non-sparking state; if the third characteristic distance corresponding to the ignition voiceprint feature is greater than the third characteristic distance corresponding to the non-sparking voiceprint feature, it is determined that the working state inside the ray source box is the ignition state; The working status determination module comprises: The second working state determination submodule is used to input the voiceprint feature into a pre-trained ignition state detection model, so that the ignition state detection model determines the matching probability between the voiceprint feature and the pre-learned sample voiceprint feature corresponding to the ignition state based on the model parameters obtained by pre-training, and outputs the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the working state inside the ray source box according to the corresponding relationship between the sample voiceprint feature corresponding to the ignition state and the ignition state; The device also includes: A sample acquisition module, used to acquire a plurality of sample voiceprint features corresponding to the ignition state and an ignition state label corresponding to each sample voiceprint feature, wherein the ignition state label is used to identify that the working state corresponding to the sample voiceprint feature is the ignition state; An estimated state acquisition module is used to input each sample voiceprint feature into an initial ignition state detection model, so that the initial ignition state detection determines the matching probability between the sample voiceprint feature and the currently learned sample voiceprint feature corresponding to the ignition state based on the current model parameters, and outputs the ignition state corresponding to the sample voiceprint feature with the highest matching probability as the estimated ignition state corresponding to each sample voiceprint feature according to the corresponding relationship between the currently learned sample voiceprint feature corresponding to the ignition state and the ignition state; A loss function value calculation module, used to calculate the loss function value based on the difference between the ignition state identified by the ignition state label corresponding to each sample voiceprint feature and the estimated ignition state; A parameter adjustment module, used for adjusting the model parameters of the initial ignition state detection model by back propagation according to the loss function value, until the initial ignition state detection model meets the convergence condition, thereby obtaining the ignition state detection model; The device also includes: A position determination module, used to determine the position corresponding to the target position in the three-dimensional model based on the spatial mapping relationship between the pre-built three-dimensional model and the ray source box; The identification display module is used to display a preset ignition status identification at a determined position.

12. A system for detecting the state of a radiation source, characterized in that: The detection system includes a processor and a sound sensor, wherein the sound sensor is installed on at least three outer surfaces of the radiation source box. The sound sensor is used to collect sound signals emitted from inside the radiation source box; The processor is used to obtain the sound signal collected by the sound sensor and extract the sound print features of the sound signal; determine the working state inside the ray source box by identifying the sound print features; when it is determined that the working state is the ignition state, determine the target position where the ignition phenomenon occurs inside the ray source box according to the time when the sound sensor collects the sound signal, the installation position of the sound sensor and the propagation speed of the sound signal inside the ray source box, wherein the working state includes a non-ignition state and an ignition state.