CN115575491A - Self powered gas sensing device, system and method - Google Patents

Abstract

A self-powered gas sensing system, apparatus and method for identifying properties of gases in a gaseous environment to be measured are disclosed. The self-powered gas sensing system includes: a self-powered gas sensing device, including: an energy harvester configured to obtain mechanical energy and convert the mechanical energy into electrical energy; a breakdown arrestor arranged in the gas environment to be measured In the above, the breakdown arrester is configured to obtain the electric energy, wherein the electric energy causes the breakdown arrester to perform breakdown discharge in the gas environment to be tested, and the current generated by the breakdown discharge leads to the emission of electromagnetic wave signals , and wherein the electromagnetic wave signal carries information associated with a property of the gas in the gas environment to be measured; and a gas property identification device configured to receive the electromagnetic wave signal from the self-powered gas sensing device, And identifying the property of the gas in the gas environment to be measured based on the electromagnetic wave signal.

Description

Self powered gas sensing device, system and method

technical field

The present disclosure relates to the field of sensing technology, and more particularly, to a self-powered gas sensing device, system and method.

Background technique

Gas sensors play an important role in industrial exhaust, indoor environment detection and other fields, and have a broad application background. Traditional gas sensors usually rely on external power sources to provide energy, such as catalytic combustion gas sensors, semiconductor gas sensors, thermal conductivity gas sensors, infrared gas sensors, and so on. The need to rely on an external power supply greatly limits the application range of the gas sensor. Even though wireless energy transmission has appeared now, this power supply method is limited by a limited transmission distance, and there is still a demand for an external power supply to power the energy transmitting part.

Self-powered technology is a technology that collects other forms of energy (such as solar energy, wind energy, mechanical energy, heat energy, etc.) However, due to the different impedance and output characteristics of energy harvesters, this power supply method based on self-energy technology usually requires the introduction of energy management circuits, which introduces additional energy requirements and increases the size of the entire system.

At the same time, existing gas sensors often have limitations in detection range and other defects. For example, catalytic combustion gas sensors are limited to testing combustible gases and have the risk of ignition and explosion. Semiconductor gas sensors are greatly interfered by background gas, Susceptible to temperature, thermal conductivity gas sensors are significantly affected by temperature, poor detection accuracy, low sensitivity, and infrared gas sensors have high cost, complex systems, and are limited to testing gases that absorb infrared radiation, etc. In addition, many gas sensors cannot be separated from the subsequent sensor data analysis device too far (for example, not more than 1m), because for example, if a transmission medium such as a wire is required to transmit the sensing signal, a too long wire will have too much impedance and thus affect the transmission. The effectiveness of the sensing signal and increase the volume of the system, but in some application scenarios, the gas sensor and the sensor data analysis device cannot be placed close enough due to space constraints or the data analysis device is not suitable for movement.

For this reason, there is an urgent need for a sensor that can integrate energy harvesting and conversion functions and gas detection functions, is small in size, can be used to safely sense various types of gases, has high sensing accuracy, and has a long transmission distance for sensing signals. self-powered gas sensor.

Contents of the invention

According to an aspect of the present disclosure, there is provided a self-powered gas sensing system for identifying properties of a gas in a gas environment to be measured, comprising: a self-powered gas sensing device including: an energy harvester configured to acquire mechanical energy, and convert the mechanical energy into electrical energy; a breakdown arrester is arranged in the gas environment to be measured, and the breakdown arrester is configured to obtain the electrical energy, wherein the electrical energy causes the breakdown The discharger performs a breakdown discharge in the gas environment to be tested, and the current generated by the breakdown discharge causes the emission of an electromagnetic wave signal, and wherein the electromagnetic wave signal carries information associated with the properties of the gas in the gas environment to be tested and a gas attribute identification device configured to receive the electromagnetic wave signal from the self-powered gas sensing device, and identify an attribute of the gas in the gas environment to be measured based on the electromagnetic wave signal.

According to an embodiment of the present disclosure, wherein the breakdown arrestor includes a first electrode and a second electrode separated by gas insulation in the gas environment to be measured, and wherein the first electrode and the second electrode have Opposite discharge tips, there is a discharge gap between the opposite discharge tips, the electric energy forms an electric field between the tips of the first electrode and the second electrode, and the electric field makes the electric field between the first electrode and the second electrode The gas in the gas environment to be measured is broken down between the discharge gaps of the discharge tips to perform a breakdown discharge.

According to an embodiment of the present disclosure, the properties of the gas include at least one of the following: gas composition, gas concentration, and gas pressure.

According to an embodiment of the present disclosure, wherein the gas property identification device includes: a signal receiving unit configured to receive the electromagnetic wave signal from the self-powered gas sensing device, and convert the electromagnetic wave signal into a predetermined electrical parameter and an identification unit configured to identify the property of the gas in the gas environment to be measured based on the time series signal of the predetermined electrical parameter.

According to an embodiment of the present disclosure, wherein the gas property identification device further includes: a signal processing unit configured to perform signal processing on the time series signal of the predetermined electrical parameter to obtain a processed signal, wherein the identification unit is based on The processed signal identifies the properties of the gas in the gas environment to be measured; wherein the signal processing includes at least one of the following operations: removing noise signals or intercepting effective signals; and extracting spectrum signals.

According to an embodiment of the present disclosure, wherein the identification unit includes a machine learning model, the machine learning model is trained to generate an identification result indicating the property of the gas in the gas environment to be measured for the processed signal, wherein, The machine learning model is trained on a training set to obtain a trained machine learning model, and the training set includes a plurality of processed signals and real labels of gas properties corresponding to each of the plurality of processed signals.

According to an embodiment of the present disclosure, wherein the training set is obtained by: selecting a first number of reference gas environments with different gas properties; obtaining a second number of processed signals for each reference gas environment; a reference gas environment, using the second number of processed signals and the true labels of their corresponding gas properties as a training subset; and using the training subsets for all reference gas environments together as a training set.

According to an embodiment of the present disclosure, wherein the second quantity of processed signals for each reference gas environment is obtained by: for the reference gas environment, an airtight gas placed in the gas filled in the reference gas environment The breakdown arrester in the chamber, based on the electrical energy obtained from the energy harvester, performs a breakdown discharge in the reference gas environment, wherein the current generated by the breakdown discharge causes the emission of an electromagnetic wave signal; obtaining The electromagnetic wave signal, and converting the electromagnetic wave signal into a second number of time-series signals of predetermined electrical parameters; and processing each time-series signal of predetermined electrical parameters to obtain the second number of processed signals.

According to an embodiment of the present disclosure, wherein each of the processed signals included in the training set is a time-domain signal, the machine learning model is a bidirectional long-short-term memory recurrent neural network (bi-LSTM) model, wherein, The machine learning model is trained in the following manner: each time domain signal included in the training set is input forward and reversely to the bi-LSTM model to obtain a prediction corresponding to each time domain signal label, and based on the difference between each predicted label and the real label of the corresponding time-domain signal, the model parameters of the bi-LSTM model are updated by the error backpropagation method and the gradient descent method until the model parameters are relative to the set Converge on the training set.

According to an embodiment of the present disclosure, wherein each of the processed signals included in the training set is a spectral image signal, the machine learning model is a convolutional neural network (CNN) model, wherein the machine learning model Training in the following manner: each spectral image signal included in the training set is input to the CNN model to obtain a prediction label corresponding to each spectral image signal, and based on each prediction label and the corresponding spectral image signal The difference of the true label, update the model parameters of the CNN model through the error backpropagation method and the gradient descent method, until the model parameters converge with respect to the training set.

According to another aspect of the present disclosure, a self-powered gas sensing device is provided, including: an energy harvester configured to harvest mechanical energy and convert the mechanical energy into electrical energy; In a gas environment, the breakdown arrester is configured to obtain the electrical energy, wherein the electrical energy causes the breakdown arrester to perform a breakdown discharge in the gas environment to be tested, and the current generated by the breakdown discharge causes an electromagnetic wave The emission of a signal, wherein the electromagnetic wave signal carries information associated with the property of the gas in the gas environment to be measured.

According to an embodiment of the present disclosure, wherein the breakdown arrestor includes a first electrode and a second electrode separated by gas insulation in the gas environment to be measured, and wherein the first electrode and the second electrode have Opposite discharge tips, there is a discharge gap between the discharge tips, the electric energy forms an electric field between the tips of the first electrode and the second electrode, and the discharge of the electric field at the discharge tips of the first electrode and the second electrode The gas in the gas environment to be measured is broken down between the gaps to perform breakdown discharge.

According to yet another aspect of the present disclosure, there is provided a self-powered gas sensing method for identifying properties of a gas in a gas environment to be tested, comprising: placing a breakdown arrestor in the gas environment to be tested; utilizing The energy harvester obtains mechanical energy and converts the mechanical energy into electrical energy; the electrical energy is obtained by using the breakdown arrester, wherein the electrical energy causes the breakdown arrester to perform breakdown discharge in the gas environment to be measured, and The current generated by the breakdown discharge causes the emission of an electromagnetic wave signal, and wherein the electromagnetic wave signal carries information associated with the properties of the gas in the gas environment to be measured; and using the gas property identification device based on the electromagnetic wave signal, Identifying properties of gases in the gas environment to be measured.

According to an embodiment of the present disclosure, wherein, using the gas attribute identification device to identify the attribute of the gas in the gas environment to be measured based on the electromagnetic wave signal includes: receiving the electromagnetic wave signal, and converting the electromagnetic wave signal into a predetermined A time-series signal of an electrical parameter; performing signal processing on the time-series signal of the predetermined electrical parameter to obtain a processed signal; and identifying the property of the gas in the gas environment to be measured based on the processed signal.

According to an embodiment of the present disclosure, wherein the gas property identification device includes a machine learning model, the self-powered gas sensing method further includes: acquiring a plurality of processed signals and gas properties corresponding to each of the plurality of processed signals as a training set; and using the training set to train the machine learning model, so that the trained machine learning model can generate a recognition result indicating the property of the gas in the gas environment under test for the processed signal.

According to the self-powered gas sensing device, system, and method of the embodiments of the present disclosure, the properties of the gas in the gas environment where the self-powered gas sensing device is located, such as gas composition, concentration, and air pressure, can be identified. The advanced gas sensing technology can realize the sensing of most gases without external power supply, and the generated sensing signal itself is a wireless signal so that it is not constrained by the signal transmission line, and does not require additional energy management modules and wireless signals The transmitter module has a small volume and high recognition accuracy of gas properties, so that the self-powered gas sensing system has a wider range of applications.

Description of drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings that need to be used in the description of the embodiments will be briefly introduced below. Apparently, the drawings in the following description are only some exemplary embodiments of the present disclosure, and those skilled in the art can obtain other drawings according to these drawings without creative efforts.

Fig. 1 shows a schematic structural block diagram of a self-powered gas sensing system according to an embodiment of the present disclosure.

Fig. 2 shows a schematic structural block diagram of a self-powered gas sensing device according to an embodiment of the present disclosure.

Figures 3A-3B illustrate example structures of triboelectric nanogenerators.

FIG. 4 shows an equivalent circuit model of the self-powered gas sensing device described in FIG. 2 .

Fig. 5A shows a schematic structural block diagram of a gas property identification device according to an embodiment of the present disclosure.

Fig. 5B shows a schematic diagram of a structure for obtaining a second quantity of processed signals based on a self-powered gas sensing device.

Figure 5C shows a schematic diagram of the voltage-time time series obtained for 10 reference gas environments.

Figure 5D shows a flow diagram for obtaining multiple processed signals for each reference gas environment.

Fig. 6 shows a schematic diagram illustrating the effectiveness of a self-powered gas sensing system based on a confusion matrix.

FIG. 7 shows a schematic flowchart of a self-powered gas sensing method according to an embodiment of the present disclosure.

Fig. 8 shows a schematic structural block diagram of a computing device that can be used to implement an identification unit according to an embodiment of the present disclosure.

detailed description

Embodiments of the present invention will be described in detail below. It should be emphasized that the following description is only exemplary and not intended to limit the scope of the invention and its application.

In order to make the objects, technical solutions, and advantages of the present disclosure more apparent, exemplary embodiments according to the present disclosure will be described in detail below with reference to the accompanying drawings. Apparently, the described embodiments are only some of the embodiments of the present disclosure, rather than all the embodiments of the present disclosure, and it should be understood that the present disclosure is not limited by the exemplary embodiments described here.

In the present specification and drawings, substantially the same or similar steps and elements are denoted by the same or similar reference numerals, and repeated descriptions of these steps and elements will be omitted. The terms "first" and "second" are used for descriptive purposes only, and should not be understood as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or in the description of the embodiment of the invention, "plurality" means two or more, unless otherwise specified Specific limits.

Fig. 1 shows a schematic block diagram of a self-powered gas sensing system 100 according to an embodiment of the present disclosure.

As shown in FIG. 1 , the self-powered gas sensing system 100 may include two parts: a self-powered gas sensing device 110 and a gas property identification device 120 .

The self-powered gas sensing device 110 can be used to harvest mechanical energy and be self-powered by converting the mechanical energy into electrical energy. In addition, the self-powered gas sensing device 110 can also sense the properties of the gas in the gas environment to be measured, and generate electromagnetic wave signals, which carry information related to the properties of the gas in the gas environment to be measured.

For example, the properties of the gas may include at least one of the following: gas composition, gas concentration, and gas pressure. Gas composition can be for pure gas environment or for mixed gas environment. In the case of a mixed gas environment, gas composition refers not only to the type of gas composition included, but also to their respective gas concentrations.

For example, mechanical energy can come from human-applied forces, wind, waves, vehicle motion, raindrops, liquid flow, and so on.

The gas attribute identification device 120 can be used to receive the electromagnetic wave signal from the self-powered gas sensing device 110, and identify the attribute of the gas in the gas environment to be measured based on the electromagnetic wave signal.

Due to the electromagnetic wave signal transmission between the self-powered gas sensing device 110 and the gas property identification device 120, the two can be separated by a relatively long distance and do not need to be connected by wires, for example, they can be separated by 30m, and the gas property identification device 120 can also receive electromagnetic waves. signal and identify gas properties. Therefore, it can be better applied in specific scenarios such as limited space or where it is not suitable to move the gas property identification device.

More details of the self-powered gas sensing device 110 and the gas property identification device 120 will be described in detail later.

According to the above self-powered gas sensing system, the properties of the gas in the gas environment where the self-powered gas sensing device is located can be realized, and the sensing of most gases can be realized without external power supply, and the generated sensing signal itself is The wireless signal is not constrained by the signal transmission line, and does not require additional energy management modules and wireless signal transmission modules, so the structure is simple and small, and the identification accuracy of gas properties is high, making the self-powered gas sensing system more powerful. scope of application.

The self-powered gas sensing device in the self-powered gas sensing system 100 described in FIG. 1 will be described in detail below with reference to FIGS. 2-4 .

FIG. 2 shows a block diagram of a self-powered gas sensing device according to an embodiment of the present disclosure.

As shown in Figure 2, the self-powered gas sensing device 110 includes an energy harvester 110-1 and a breakdown arrestor 110-2.

The energy harvester 110-1 is configured to harvest mechanical energy and convert the mechanical energy into electrical energy.

The energy harvester 110-1 can be placed in the gas environment to be tested, and can also be placed outside the gas environment to be tested. No matter where it is placed, the energy harvester 110-1 needs to capture mechanical energy and convert the mechanical energy into electrical energy to realize the self-powering of the self-powered gas sensing device 110.

By way of example and not limitation, energy harvester 110-1 may be a triboelectric nanogenerator as described in the context of this disclosure.

The working principle of the triboelectric nanogenerator is introduced as follows. It should be understood that the working principle described below is exemplary only to help better understand the embodiments of the present disclosure, and different triboelectric nanogenerators have different working principles.

The triboelectric nanogenerator 110-1 may include a triboelectric generation layer, a first output electrode, and a second output electrode (not shown), wherein the triboelectric generation layer is used to generate electricity between the first output electrode and the second output electrode based on the obtained mechanical energy. Charges of opposite polarity are formed on the second output electrode, thereby forming an electric field between the first electrode and the second electrode of the breakdown discharger connected to the first output electrode and the second output electrode as described later ( supply power to the breakdown arrester). Of course, if it is another type of energy harvester, there are also a first output electrode and a second output electrode to form an electric field between the first electrode and the second electrode of the breakdown discharger.

An exemplary structure of the triboelectric nanogenerator 110-1 can be shown in FIGS. 3A-3B , including a sliding type triboelectric nanogenerator (FS-TENG) ( FIG. 3A ), and a contact separation type triboelectric nanogenerator (CS-TENG) ( Fig. 3B), the specific working process will be described with reference to Fig. 3A-3B. Of course, the following exemplary structures are only for helping to better understand the present disclosure, not for limitation. Any other suitable triboelectric nanogenerator structure can be used.

The friction nanogenerator can be a sliding friction generator, as shown in Fig. 3A. The sliding friction generator is composed of two different thin-film polymers or thin-film polymers and electrodes coated with metal electrode materials on the back. Type material film layer, and bottom electrode. The sliding friction generator receives external force (mechanical energy), so that the first type material film layer and the second type material film layer move relative to each other, and charge transfer occurs due to the triboelectric effect, so that the top electrode and the bottom electrode A potential difference (voltage difference) is formed so that an electric field can be formed between the first electrode and the second electrode of the breakdown arrester connected to the top electrode and the bottom electrode (powering the breakdown arrestor).

Exemplarily, FIG. 3A shows a generator made of polytetrafluoroethylene (PTFE) and nylon (nylon) two materials with opposite frictional polarities and its working mechanism. When PTFE and nylon are in contact, because PTFE has stronger electrons than nylon, negative charges will be generated on the surface of PTFE while positive charges will be generated on the surface of nylon. When PTFE and nylon are completely heavy, there is no potential difference between the electrode on the nylon and the electrode under the PTFE. When the two have dislocation movement under the action of external force (mechanical energy), the frictional charge on the dislocation area cannot be completely offset. There is a potential difference between the two electrodes. When PTFE and nylon overlap again, the dislocation area will disappear, and the potential difference generated by the triboelectric charge will also disappear, and free electrons will flow from the electrode on the nylon to the electrode on the PTFE. The electrical energy converted from mechanical energy can be output from between the two electrodes.

Another exemplary triboelectric nanogenerator may be a contact-separation triboelectric nanogenerator (CS-TENG). As shown in Fig. 3B, the triboelectric nanogenerator includes a first electrode layer, a first friction material layer, a second friction material layer having a different electronegativity material from the first friction material layer, and a second friction nanogenerator from top to bottom. The material of at least one of the electrode layer, the first friction material layer and the second friction material layer is an insulating material (for maintaining triboelectric charge for a long time), for example, the material of the first friction material layer can be polyimide (PI) film (Kapton), the material of the second friction material layer may be polymethyl methacrylate (PMMA).

When pressure (mechanical energy) is applied to the first electrode layer, the first friction material layer and the second friction material layer are in contact with each other. Charges of different polarities are charged, but the charges are only formed on the contact surface and the positive and negative charges can be offset, so no potential difference is formed between the first electrode layer and the second electrode layer at this time.

Once the contact surfaces of the first friction material layer and the second friction material layer are released to separate, the electrodes on the upper and lower surfaces of the first friction material layer and the second friction material layer will have a potential difference. This potential difference can form an electric field during the release process between the first electrode and the second electrode of the breakdown arrester connected to the first electrode layer and the second electrode layer (supplying the breakdown arrester with electrical energy).

That is to say, when there is no pressure signal input, no induced charge will be generated on the two electrode layers of the triboelectric nanogenerator, so there is no potential difference between the two, so it cannot provide electric energy to the breakdown discharger. When there is a pressure signal input, the first friction material layer and the second friction material layer generate induced charges, so that when the pressure signal stops inputting (during the release process as described above), the first electrode layer and the second friction material layer will Inductive charges are generated on the second electrode layer to form a potential difference between the two electrode layers, and further, the generated electric energy is provided to the breakdown discharger.

The working principle of the breakdown arrestor 110-2 will be introduced below. It should be understood that the working principles to be described below are exemplary only to help better understand the embodiments of the present disclosure.

The breakdown arrester 110-2 is set in the gas environment to be tested, and the breakdown arrester is configured to obtain the electrical energy obtained by converting the acquired mechanical energy by the triboelectric nanogenerator 110-1, wherein the electrical energy makes the breakdown The discharger performs breakdown discharge in the gas environment to be tested, and the current generated by the breakdown discharge leads to the emission of electromagnetic wave signals.

Due to different gas environments to be measured (with different gas properties), the characteristics of the generated electromagnetic wave signals, such as waveforms, amplitudes, spectral features, etc. are different, therefore, the electromagnetic wave signals can carry The information associated with the properties of the gases in . Based on the analysis of the electromagnetic wave signal, the properties of the gas in the gas environment to be measured can be identified.

The breakdown arrester 110-2 may include a first electrode and a second electrode separated by gas insulation in the gas environment to be measured, and wherein the first electrode and the second electrode have opposite discharge tips, the There is a discharge gap between the opposite discharge tips, the electric energy forms an electric field between the discharge tips of the first electrode and the second electrode, and the electric field makes the discharge gap between the discharge tips of the first electrode and the second electrode The gas in the gas environment to be measured is broken down to perform a breakdown discharge.

For example, the discharge tip can be a metal structure and can be formed by electron beam evaporation process. The minimum distance between two metal structures is controlled at 5-500 microns, which can be observed by microscope. When the electrical energy is obtained from the triboelectric nanogenerator 110-1, an electric field will be formed between the first electrode and the second electrode of the breakdown discharger, and the electric field intensity will reach the maximum at the discharge tip, when the electric field intensity at the discharge tip When it is large enough, the gas in the gas environment to be measured is broken down between the discharge gaps of the discharge tips of the first electrode and the second electrode, so as to perform a breakdown discharge.

The breakdown discharge will form a current between the discharge tips, and this current will cause the emission of electromagnetic wave signals.

Specifically, a strong electric field is generated between the opposing discharge tips of the first and second electrodes of the breakdown discharger, causing electrons to move from the cathode to the anode. During the movement, electrons collide with air molecules at high speed, resulting in the generation of new electrons, positive ions and negative ions, causing electron avalanche and further generating plasma. The plasma can be viewed as a collective oscillation of clusters of charged particles, creating a region of zero impedance that acts as a conductor for electrons. Therefore, a pulse current is generated in the emission system, and its amplitude and pulse rise time are related to voltage, gap width and gas properties (gas composition, pressure, concentration, etc.). Eventually, as the electric field weakens, the weakened electron avalanche cannot continue to support the path, causing the circuit to continue returning to the open circuit state.

The process of generating the electromagnetic wave signal will be described in detail with reference to FIG. 4 .

Figure 4 shows an equivalent circuit model of the self-powered gas sensing device (including the energy harvester and the breakdown arrestor) described in Figures 2-3B.

As shown in Figure 4, the voltage source Vi and the equivalent capacitance Ci correspond to the equivalent circuit model of the energy harvester (such as a triboelectric nanogenerator); C, R and L are the capacitance and inductance of the energy harvester and the breakdown discharger and resistance, which can be derived from parasitic capacitance, parasitic self-inductance, parasitic resistance, etc. two output electrodes) and the first electrode and the second electrode of the breakdown discharger will form a conduction loop. Once the pulse current in the breakdown discharge occurs, it will generate an underdamped oscillation current signal in the emission system composed of the triboelectric nanogenerator and the breakdown discharge device. The oscillation can generate a changing magnetic field and a changing electric field in the surroundings to generate an omnidirectionally propagating electromagnetic wave signal, which is finally transmitted to the gas attribute identification device for reception. Optionally, the electromagnetic wave signal can be received by a receiving coil or a receiving capacitor in the gas property identification device.

Research can be conducted to evaluate the output performance of the electromagnetic wave signal caused by the breakdown and the relationship between the influencing factors. The main influencing factors can be determined empirically, including the breakdown arrester voltage (U), the gap distance between electrodes of the breakdown arrester (d), the movement direction of the triboelectric nanogenerator (-), the energy harvester (such as triboelectric nanogenerator) The length of the connecting wire with the breakdown arrester (l), the space conductor distribution (a), the distance between the breakdown arrester and the receiver (D), the gas composition (N), the gas pressure (P), and the gas concentration (C) , temperature (T), humidity (H).

By sequentially changing the parameter value of one influencing factor, while keeping the parameter values of other influencing factors unchanged, the following conclusions can be drawn: the change of most of the above influencing factors will change the amplitude of the electromagnetic wave signal, but will not affect the waveform and spectrum, but Changes in the length of the connecting wire (l) between the energy harvester (such as a triboelectric nanogenerator) and the breakdown discharger, as well as changes in gas composition (N), gas pressure (P), and gas concentration (C) will cause the waveform and spectrum of the electromagnetic wave signal Change. This also shows that under the condition that the structure of the same self-powered gas sensing system does not change, that is, when the length (l) of the connecting wire between the energy harvester (such as a triboelectric nanogenerator) and the breakdown discharger is fixed, the The property of the gas in the gas environment to be measured where the self-powered gas sensing system is located is identified through the electromagnetic wave signal.

By means of the self-powered gas sensing device as described with reference to FIGS. Powered by an external power supply, the generated electromagnetic wave signal itself is a wireless signal, so it is not restricted by the signal transmission line and can carry information about the properties of the gas, so that the properties of the gas can be determined by analyzing the electromagnetic wave signal.

The composition and working process of the gas property identification device in the self-powered gas sensing system described in Fig. 1 will be described in detail below in conjunction with Figs. 5A-5D.

Fig. 5A shows a schematic structural block diagram of a gas property identification device according to an embodiment of the present disclosure.

As shown in Fig. 5A, the gas attribute identification device 120 may include: a signal receiving unit 120-1 and an identification unit 120-2.

The signal receiving unit 120-1 is configured to receive the electromagnetic wave signal from the self-powered gas sensing device 110, and convert the electromagnetic wave signal into a time-series signal of an electrical parameter, that is, a time-series signal of an electrical parameter changing with time.

For example, the signal receiving unit 120-1 may include an electromagnetic wave receiving circuit and a sampling circuit. The electromagnetic wave receiving circuit can convert the electromagnetic wave signal into a predetermined electrical parameter signal (for example, the predetermined electrical parameter can be voltage or current), and the sampling circuit can sample the predetermined electrical parameter signal according to a predetermined sampling frequency, thereby obtaining the timing of the predetermined electrical parameter Signal. An example of the electromagnetic wave receiving circuit may be at least one of a receiving coil and a receiving antenna.

The identification unit 120-2 is configured to identify the property of the gas in the gas environment to be measured based on the time series signal of the predetermined electrical parameter.

As mentioned earlier, if the properties of the gas are different, the waveform and spectrum characteristics of the electromagnetic wave signal will be different accordingly. Therefore, the electromagnetic wave signal can be obtained by analyzing the time series signal of predetermined electrical parameters obtained after the electromagnetic wave signal is received and sampled. The properties of the gas in the gas environment to be measured can be obtained from the signal.

Because the timing signal of predetermined electrical parameters obtained by receiving and sampling the electromagnetic wave signal may include some redundant signals and noise signals. In addition, considering multiple influencing factors will affect the characteristics (waveform, amplitude or spectrum) of the electromagnetic wave signal. For example, even for the same gas environment, the difference in the magnitude of the mechanical energy will cause the amplitude of the electromagnetic wave signal to be different, but this will not affect Waveform and spectrum features, so the spectrum signal can also be extracted in the signal processing process, and can optionally be identified based on the spectrum signal as described later, so in some cases, the gas attribute identification device can also include signal processing The unit 120-3 is configured to preprocess the timing signal of predetermined electrical parameters to obtain a processed signal, and then the identification unit 120-2 will perform identification based on the processed signal. Of course, the signal processing unit 120-3 may also be included in the identification unit 120-2.

For example, the signal processing unit 120-3 may remove noise signals and/or intercept effective signals from the time series signals of the predetermined electrical parameters output by the electromagnetic wave receiving circuit to obtain processed signals. That is, in order to improve the accuracy of sensing and/or facilitate identification, before inputting the time-series signals of predetermined electrical parameters into the identification unit 120-2 for identification, the time-series signals of predetermined electrical parameters are signal-processed (for example, noise removal signal, intercept effective signal, and extract spectrum signal, etc.), and then obtain the properties of the gas in the gas environment to be measured based on the processed signal.

Optionally, regarding the manner in which the identification unit 120-2 identifies the properties of the gas, one manner may include: The identification unit 120-2 may be based on the time-varying trend of the value of the predetermined electrical parameter in the time series signal of the acquired predetermined electrical parameter (this trend of change can reflect the waveform) is analyzed, and according to this trend of change, it can be determined that the gas environment to be measured corresponds to or is the closest to which gas environment (one of the reference gas environments, according to application scenarios (for example, industry, automobile exhaust, Indoor environment, etc.) and predetermine possible gas environment as a reference gas environment).

In addition, another method may include: the identification unit 120-2 may also obtain a spectrum signal (such as a spectrum image) according to the time series signal of a predetermined electrical parameter, and by analyzing the spectrum signal, determine that the gas environment to be tested corresponds to or is the most Which gas environment to approach.

Optionally, the identification unit 120-2 can obtain the properties of the gas in the gas environment to be measured based on the time series signals of predetermined electrical parameters based on machine learning.

That is, the identification unit 120-2 may include a machine learning model, and the machine learning module can be trained to generate a property indicating gas for a time-series signal of a predetermined electrical parameter (in order to improve the identification accuracy, the processed signal after signal processing is generally used). recognition results.

As mentioned above, for the first approach, the processed signal is a time-domain signal, so the type of machine learning model suitable for training time-domain signals can be selected.

For the second way, the processed signal is a frequency domain signal (spectral image), so the type of machine learning model suitable for image classification can be selected.

Of course, in addition to using a machine learning model, the identification unit 120-2 may also identify the attributes of the gas in the gas environment to be measured in other ways, and the present disclosure is not limited thereto. For example, according to the application, the characteristics of the electromagnetic wave signal for each of the common reference gas environments (for example, the variation trend of the value of the electrical parameters mentioned above relative to time, the spectral characteristics) Store correspondingly, and compare the characteristics of the processed signal with the stored characteristics by sampling the electromagnetic wave signal in the gas environment to be tested and signal processing, and determine the corresponding or closest gas environment for the characteristic .

Next, the identification unit 120-2 is mainly described in detail for identifying the properties of the gas in the gas environment to be measured based on the machine learning model.

The machine learning model is trained on the training set to obtain the trained machine learning model, and the training set includes a plurality of processed signals and true labels of gas properties corresponding to each of the plurality of processed signals.

For example, for time-domain signals, the machine learning model can be a bidirectional long-short-term memory recurrent neural network (bi-LSTM) model.

This bidirectional long-short-term memory recurrent neural network (bi-LSTM) model can be trained as follows: Each time domain signal included in the training set is input into the bi-LSTM model forwardly and backwardly, to obtain each The predicted label corresponding to the time domain signal, and based on the difference between each predicted label and the real label of the corresponding time domain signal, the model parameters of the bi-LSTM model are updated by the error backpropagation method and the gradient descent method, until the model parameters converge with respect to the training set. The way to obtain the training set will be described later.

For example, regarding the bi-LSTM model, the bidirectional LSTM layer has 50 neurons in each direction, and there are 100 neurons in total; then, the output after the bidirectional LSTM layer is connected to the fully connected layer (including 32 neurons ), after batch normalization (Batch Normalization, momentum momentum is set to 0.8), it is activated by an activation function (for example, ReLu), and then connected to the fully connected layer in the model output structure, which is set according to the number of categories The number of neurons (for example, if the model is trained to identify gases in ten gas environments, then the number of neurons is set to 10), and finally activated by the softmax function, so that each training set in the training set The predicted label for the sample (processed signal). There may be differences between the predicted label and the real label of the training sample, so the difference can be used to update the model parameters, that is, in the process of updating the model parameters, the error back propagation method and the gradient descent method can be used to update the model parameters.

For example, for spectral signals, the machine learning model can be, for example, a convolutional neural network model, etc., which can be used for image classification.

The machine learning model that can be used for image classification, such as a convolutional neural network model, etc., can be trained as follows: each spectrum image signal included in the training set is input to the CNN model to obtain each spectrum. The predicted label corresponding to the image signal, and based on the difference between each predicted label and the corresponding real label of the spectral image signal, the model parameters of the CNN model are updated by the error backpropagation method and the gradient descent method until the model parameters Converged with respect to the training set. Likewise, the way to obtain the training set will be described later.

For example, a convolutional neural network can adopt a general architecture such as AlexNet, ResNet architecture, etc., and likewise output a predicted label for each training sample (processed signal) in the training set. There may also be differences between the predicted label and the real label of the training sample, so this difference can also be used to update the model parameters, that is, in the process of updating the model parameters, the error back propagation method and the gradient descent method can be used to update the model parameters. renew.

Usually, the machine learning model is trained on the training set, so before training the machine learning model, it is necessary to obtain the training set.

As mentioned above, each processed signal is obtained by signal processing the received electromagnetic wave signal, and the characteristics of the electromagnetic wave signal generated based on the breakdown discharge process in different gas environments are different, so based on different gas The electromagnetic wave signal in the environment can be used to obtain a plurality of processed signals. Therefore, the training set can be obtained in the following way.

First, a first number of reference gas environments with different gas properties are selected.

As an example and not a limitation, 10 kinds of reference gas environments (as shown in Table 1) are used in the embodiments of the present disclosure, according to specific application scenarios (for example, industry, automobile exhaust, indoor environment, etc.) need to actually identify the gas environment, but different and more reference gas environments can be selected to generate the training set. Wherein, in the following content of the present disclosure, the air pressure of 10 kinds of reference gas environments is a standard atmospheric pressure. Of course, if the final training machine learning model needs to identify the gas pressure of the gas environment to be tested, then identifying the gas pressure should also be used as a training goal. At this time, the gas pressure can also be used as a variable, for example, at least two reference gas environments can be The gas pressure is set to be different, but the gas composition and gas concentration are the same.

[Table 1] 10 reference gas environments

A second quantity of processed signals is then obtained for each reference gas environment.

Fig. 5B shows a schematic diagram of a structure for obtaining a second quantity of processed signals based on a self-powered gas sensing device. Select the structure of the self-powered gas sensing device to be used to identify the gas environment to be measured, i.e. select the structure and parameters of the energy harvester (e.g. triboelectric nanogenerator) and the breakdown arrestor. As in the structure of the self-powered gas sensing device described above with reference to FIGS. The discharger provides electrical energy. That is to say, after determining the structure of the self-powered gas sensing device to be used to identify the properties of the gas in the gas environment to be measured, as will be described later, the self-powered gas sensing device on which the training set is generated based on the reference gas environment is based. The structure of the gas sensing device also needs to be the same structure (different structures (for example, the connection wire length (l) of the energy harvester and the breakdown amplifier may be different) will cause the waveform and Spectral features are different), so that the machine learning model can be trained most accurately.

Optionally, for each reference gas environment in turn, perform the following operations i-iii to obtain the second number of processed signals.

Operation i, the breakdown arrester is placed in the airtight chamber filled with the gas in the reference gas environment, for example, the airtight chamber can be evacuated first, and then the gas corresponding to the reference gas environment can be injected. When the reference gas environment is subsequently replaced, the airtight chamber can be evacuated again, and then the gas corresponding to the next reference gas environment can be injected.

Operation ii, harvesting mechanical energy by an energy harvester (e.g., a triboelectric nanogenerator) to convert the mechanical energy into electrical energy, wherein the breakdown arrestor is activated in the reference gas based on the electrical energy harvested from the energy harvester The breakdown discharge is carried out in the environment, and the current generated by the breakdown discharge leads to the emission of electromagnetic wave signals.

Operation iii, after the electromagnetic wave signal is emitted, the electromagnetic wave receiving circuit can acquire the electromagnetic wave signal, and convert the electromagnetic wave signal into a second number of time series signals of electrical parameters (for example, the second number of time periods of the same time length are respectively A plurality of sampling points and corresponding electrical parameter values form a second number of time series signals of predetermined electrical parameters), for example, as shown in Figure 5C, as an example, the number of types (first number) of the reference gas environment is 10, for The number of time-series signals of the voltage obtained for each reference gas environment is 100, as described later, including the second number 80 for the training set plus the third number 20 for the test set. Thereafter, the time-series signals of each predetermined electrical parameter are processed to obtain the second number of processed signals.

In order to more clearly illustrate the above process of obtaining the second number of processed signals, Fig. 5D also shows a flowchart for obtaining the second number of processed signals for each reference gas environment.

As shown in Figure 5D, the initial state of the gas chamber is an air environment. First, the gas chamber is evacuated to 0.001 MPa (MPa), and the gas (target gas) corresponding to the reference gas environment is injected to 0.1 MPa (one atmosphere) , and repeat this process several times (here, 5 for example) to ensure that the target gas fills the air chamber and there are no other gases in the air chamber. Then, the electrical energy output by the energy harvester (for example, a triboelectric nanogenerator) causes the breakdown discharger to perform breakdown discharge to realize the emission of electromagnetic wave signals, and then convert the electromagnetic wave signals into predetermined The time-series signals of electrical parameters and further obtain processed signals, when the number of time-series signals of predetermined electrical parameters reaches the preset number 100 (as described later, the second number 80 for the training set plus the first number 80 for the test set When the third quantity is 20), stop the acquisition of the processed signal for the current reference gas environment, re-evacuate the gas chamber to 0.001MPa, and inject the gas corresponding to the next reference gas environment (shown as air environment) to 0.1 MPa (one atmospheric pressure), and repeat this process for a number of times (here, 5 for example) to obtain a second number of processed signals for the next reference gas environment.

Then, for each kind of reference gas environment, the processed signals of the second quantity and the true labels of the respective gas properties are used as the training subset of this kind of reference gas environment.

Finally, the respective training subsets of the first quantity of reference gas environments are collectively used as a training set.

For example, in the case where the second number is 80, the 80 processed signals obtained for each reference gas environment (of course, 80 is only an example here) and the real labels of the respective gas properties corresponding to the 80 processed signals (for example, air, nitrogen, or helium, etc., can be marked with the number (1, 2, 3...10) in the table) together as the training set, in this example, the training set includes 800 sets of training data .

Optionally, the 80 processed signals (which may be time domain signals or frequency spectrum signals) may be in the form of images. That is to say, after obtaining a time-domain signal or a frequency spectrum signal from each time-series signal of predetermined electrical parameters, a time-domain image or frequency spectrum image can be further generated to facilitate the training of the machine learning model.

In addition, generally after the machine learning model is trained, the test set is used to evaluate the performance of the trained machine learning model, that is, the performance of the trained machine learning model is tested by using training samples that the model has not seen.

In an embodiment of the present disclosure, the performance of the trained machine learning model, i.e., the effectiveness of the self-powered gas sensing system, can be verified in the following manner.

First, for each reference gas environment, when acquiring the second number of processed signals, a third number of processed signals is additionally acquired, and the third number of processed signals and the real labels of the respective corresponding gas properties as a test subset.

Then, the test subsets of the first quantity of reference gas environments are collectively used as a test set.

For example, 100 processed signals are acquired for each reference gas environment and divided into a first part and a second part, wherein the first part includes a second number of 80 processed signals (and their labels as 80 groups) as described above. training data), the second part includes a third number of 20 processed signals (and their labels as 20 sets of test data). For example, since there are 10 reference gas environments, 10 test subsets together constitute the test set, and there are 200 sets of test data in total.

Next, all the processed signals in the test set are input into the trained machine learning model, and the recognition results (predicted labels) are obtained from the trained machine learning model.

For example, the test data in the test set can be sequentially or randomly input into the trained machine learning model according to the true label type, and the recognition result (predicted label) is obtained from the trained machine learning model.

Finally, based on the recognition result (predicted label) of each processed signal obtained from the trained machine learning model and the real label of the gas property corresponding to the processed signal, the effective value of the self-powered gas sensing system is obtained. sex.

For example, taking the 200 sets of test data in the test set as an example, if the recognition results of the 197 sets of test data (processed signals) are consistent with the labels, and the recognition results of the 3 sets of test data (processed signals) are inconsistent with the labels, then you can The effectiveness of the self-powered gas sensing system was determined to be 98.5%.

In machine learning, this validity can also be obtained automatically through the confusion matrix based on the test set.

Fig. 6 shows a schematic diagram showing effectiveness based on a confusion matrix. Still take the training set and data set obtained for the above-mentioned 10 kinds of reference gas environments as an example.

From the confusion matrix shown in Figure 6, it can be seen that for all the test data in the 7 test subsets obtained for the reference gas environments numbered 1, 2, 4-8, the predicted labels (recognition results of the trained machine learning model) and The real labels are consistent, for example, all are 1, all are 2, all are one of 4-8, etc., which can also be understood as, for each of the reference gas environments numbered 1, 2, 4-8 The recognition results of 20 test data in the obtained test subset are all correct. On the other hand, for some of the test data in the 3 test subsets obtained for the reference gas environments numbered 3, 9-10, the predicted labels (recognition results of the trained machine learning model) were inconsistent with the real labels. For example, in the test subset obtained for the reference gas environment numbered 3, there is a test data whose predicted label (recognition result) is 2, but the real label should be 3, so the test subset obtained for the reference gas environment numbered 3 The correct rate of the recognition results of the set is 95%, and the error rate is 5%.

Based on the comparison results of the predicted labels and the real labels of all test data obtained for all reference gas environments numbered 1-10, it can be determined that the effectiveness of the self-powered gas sensing system is 98.5% based on the confusion matrix. Successful recognition of the input signal.

Through the gas attribute identification device described above with reference to FIGS. Analysis allows the properties of the gas to be determined, and sensing of most gases is possible. In addition, the specific identification process of gas properties can be identified through a machine learning model, and the training set of the machine learning model can be obtained through electromagnetic wave signals generated for different gas environments. In addition, a test set was obtained along with the training set, which can be used to test the effectiveness of the self-powered gas sensing system.

According to yet another aspect of the present disclosure, there is also provided a self-powered gas sensing method that can be used to identify properties of gases in a gaseous environment to be measured.

FIG. 7 shows a schematic flow diagram of a self-powered gas sensing method for identifying properties of a gas in a gas environment to be measured, according to an embodiment of the present disclosure.

As shown in FIG. 7, in step S710, the breakdown arrester is placed in the gas environment to be tested.

Alternatively, the breakdown arrestor may be a breakdown arrestor as in a self-powered gas sensing device as described with reference to Figures 2-4.

In step S720, the energy harvester is used to obtain mechanical energy and convert the mechanical energy into electrical energy.

Alternatively, the energy harvester can be a triboelectric nanogenerator. The triboelectric nanogenerator may be a triboelectric nanogenerator in a self-powered gas sensing device as described with reference to Figures 2-4. Moreover, more details of step S720 have also been described in detail above, so it will not be repeated here.

Optionally, the triboelectric nanogenerator can be placed in the gas environment to be tested together with the breakdown arrester, but sometimes, for example, in the case of space constraints, the triboelectric nanogenerator can also not be placed in the gas environment to be measured (the breakdown arrester The wear amplifier needs to be placed in the gas environment to be tested).

In step S730, the electrical energy is obtained by using the breakdown arrester, wherein the electrical energy causes the breakdown arrester to perform breakdown discharge in the gas environment to be tested, and the current generated by the breakdown discharge causes an electromagnetic wave signal and wherein the electromagnetic wave signal carries information associated with properties of the gas in the gas environment to be measured.

As mentioned above, the first output electrode and the second output electrode of the energy harvester (for example, a triboelectric nanogenerator) are connected to the first electrode and the second electrode of the breakdown arrester to provide electrical energy to the breakdown arrester, To form an electric field between the first electrode and the second electrode, and when the strength of the electric field is large enough, the breakdown amplifier can perform breakdown discharge, so a current loop will be formed. In this current loop, due to the existence of inductance and capacitance , so there will be oscillation, which can generate a changing magnetic field and a changing electric field around, so as to generate an omnidirectionally propagating electromagnetic wave signal, which is finally transmitted to the gas attribute identification device for reception. In addition, for different gas environments, the amplitude and frequency spectrum of the electromagnetic wave signal will be different, so the gas property identification device can know the property of the gas in the gas environment at this time based on the analysis of the emitted electromagnetic wave signal.

In step S740, the property of the gas in the gas environment to be measured is identified based on the electromagnetic wave signal by using the gas property identification device.

Optionally, step S740 may include using the gas property identification device to perform the following sub-steps.

In sub-step S740-1, the electromagnetic wave signal is received, and the electromagnetic wave signal is converted into a time series signal of predetermined electrical parameters.

For example, the electromagnetic wave signal can be received by the receiving coil and converted into a predetermined electrical parameter signal, and the time series signal of the predetermined electrical parameter can be obtained through the sampling circuit.

In sub-step S740-2, signal processing is performed on the time series signal of the predetermined electrical parameter to obtain a processed signal.

For example, signal processing may include at least one of the following operations: removing noise signals; intercepting effective signals; and extracting spectrum signals.

In sub-step S740-3, based on the processed signal, the properties of the gas in the gas environment to be measured are identified.

For example, a trained machine learning model may be utilized to identify properties of the gas in the gaseous environment under test based on the processed signal.

The way to obtain the training set for training the machine learning model and the test set for testing the performance of the machine learning model can refer to the descriptions in Figures 5B-5D.

More details of the self-powered gas sensing method are the same as those for the self-powered gas sensing device and system above, so the description will not be repeated here.

Similarly, through the above self-powered gas sensing method, the identification of the properties of the gas in the gas environment where the self-powered gas sensing device is located can be realized, and the sensing of most gases can be realized without external power supply , and the generated electromagnetic wave signal itself is a wireless signal, so it is not restricted by the signal transmission line and can carry the information of gas properties, so that the properties of the gas can be determined by analyzing the electromagnetic wave signal. In addition, the specific identification process of gas properties can be identified through a machine learning model, and the training set of the machine learning model can be obtained through electromagnetic wave signals generated for different gas environments. In addition, a test set was obtained along with the training set, which can be used to test the effectiveness of the self-powered gas sensing system.

According to yet another aspect of the present disclosure, a computing device is also provided. The computing device can be used to implement the recognition unit in the gas attribute recognition device as described above or perform various operations of the recognition unit. In addition, the computing device can also implement at least a part of the operations of the signal processing unit in the gas property identification device.

FIG. 8 shows a structural block diagram of a computing device 800 according to an embodiment of the present disclosure.

The computer device includes: a processor; and a memory, on which instructions are stored, and the instructions, when executed by the processor, cause the processor to perform the machine learning-based identification of the gas environment to be tested performed by the identification unit. The individual operations involved in the process of the properties of the gas.

The computing device may be a computer terminal, a mobile terminal or other devices.

The processor can be an integrated circuit chip with signal processing capability. The above-mentioned processor may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components. The steps and logical block diagrams of the operations performed by the identification unit and optionally the signal processing unit in the embodiments of the present disclosure may be realized or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc., and can be of X84 architecture or ARM architecture.

The memory can be a non-volatile memory such as read only memory (ROM), programmable read only memory (PROM), erasable programmable read only memory (EPROM), electrically erasable programmable read only memory (EEPROM) or flash. It should be noted that the memory of the methods described in this disclosure is intended to include, but not be limited to, these and any other suitable types of memory.

The display screen of the computing device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a button, a trackball or a touch pad provided on the terminal shell, or It is an external keyboard, trackpad or mouse.

It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functions and operations of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, program segment, or part of code that includes at least one executable program for implementing specified logical functions. instruction. It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or they may sometimes be executed in the reverse order, depending upon the functionality involved. It should also be noted that each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations, can be implemented by a dedicated hardware-based system that performs the specified functions or operations , or may be implemented by a combination of dedicated hardware and computer instructions.

In general, the various example embodiments of the present disclosure may be implemented in hardware or special purpose circuits, software, firmware, logic, or any combination thereof. Certain aspects may be implemented in hardware, while other aspects may be implemented in firmware or software, which may be executed by a controller, microprocessor or other computing device. When aspects of the embodiments of the present disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it is to be understood that the blocks, devices, systems, techniques, or methods described herein may serve as non-limiting Examples are implemented in hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controllers or other computing devices, or some combination thereof.

The exemplary embodiments of the present disclosure described in detail above are illustrative only and not restrictive. Those skilled in the art should understand that without departing from the principle and spirit of the present disclosure, various modifications and combinations can be made to the embodiments or their features, and such modifications should fall within the scope of the present disclosure.

Claims (15)

1. A self-powered gas sensing system for identifying properties of a gas in a gas environment under test, comprising:

a self-powered gas sensing device comprising:

an energy harvester configured to capture mechanical energy and convert the mechanical energy into electrical energy;

a breakdown discharger disposed in the gas environment to be tested, the breakdown discharger being configured to obtain the electrical energy, wherein the electrical energy causes the breakdown discharger to perform a breakdown discharge in the gas environment to be tested, a current generated by the breakdown discharge causes emission of an electromagnetic wave signal, and wherein the electromagnetic wave signal carries information associated with properties of the gas in the gas environment to be tested; and

a gas property identification device configured to receive the electromagnetic wave signal from the self-powered gas sensing device and identify a property of gas in the gas environment under test based on the electromagnetic wave signal.

2. The self-powered gas sensing system of claim 1, wherein the surge arrestor comprises first and second electrodes isolated via gas insulation in the gas environment under test, and

the first electrode and the second electrode are provided with opposite discharge tips, a discharge gap is arranged between the opposite discharge tips, the electric energy forms an electric field between the discharge tips of the first electrode and the second electrode, and the electric field enables gas in the gas environment to be detected to be broken down between the discharge gaps of the discharge tips of the first electrode and the second electrode, so that breakdown discharge is carried out.

3. The self-powered gas sensing system of claim 1, wherein the property of the gas comprises at least one of: gas composition, gas concentration, gas pressure.

4. A self-powered gas sensing system according to any of claims 1 to 3 wherein the gas property identification means comprises:

a signal receiving unit configured to receive the electromagnetic wave signal from the self-powered gas sensing device and convert the electromagnetic wave signal into a time series signal of a predetermined electrical parameter; and

an identification unit configured to identify a property of the gas in the gas environment to be measured based on the timing signal of the predetermined electrical parameter.

5. The self-powered gas sensing system of claim 4, wherein the gas property identification device further comprises:

a signal processing unit configured to perform signal processing on the time-series signal of the predetermined electrical parameter resulting in a processed signal,

wherein the identification unit identifies a property of the gas in the gas environment to be measured based on the processed signal;

wherein the signal processing comprises at least one of: removing noise signals; intercepting a valid signal; and extracting the spectral signal.

6. The self-powered gas sensing system of claim 5, wherein the recognition unit comprises a machine learning model trained to generate recognition results indicative of properties of gas in the gas environment under test for the processed signal,

wherein the machine learning model is trained on a training set to obtain a trained machine learning model, and the training set includes a plurality of processed signals and real labels of gas attributes corresponding to the plurality of processed signals.

7. The self-powered gas sensing system of claim 6, wherein the training set is derived by:

selecting a first number of reference gas environments having different gas properties;

obtaining a second number of processed signals for each reference gas environment;

for each reference gas environment, taking the second number of processed signals and the respective corresponding real label of the gas property as a training subset of the reference gas environment; and

and taking the training subsets of the first number of reference gas environments together as a training set.

8. A self-powered gas sensing system according to claim 7, wherein the second number of processed signals for each reference gas environment is derived by: with respect to the reference gas environment, the gas environment,

performing, by the breakdown discharger placed in a closed gas chamber filled with a gas in the reference gas environment, breakdown discharge in the reference gas environment based on the electric energy obtained from the energy collector, wherein a current generated by the breakdown discharge causes emission of an electromagnetic wave signal;

acquiring the electromagnetic wave signals, and converting the electromagnetic wave signals into a second number of time sequence signals of a preset electrical parameter; and

each time series signal of a predetermined electrical parameter is processed to obtain the second number of processed signals.

9. The self-powered gas sensing system of claim 7, wherein each of the processed signals comprised by the training set is a time domain signal, the machine learning model is a bi-directional long-short-term memory recurrent neural network (bi-LSTM) model,

wherein the machine learning model is trained by:

respectively inputting each time domain signal included in the training set to the bi-LSTM model in a forward direction and in a reverse direction to obtain a prediction label corresponding to each time domain signal, and

updating model parameters of the bi-LSTM model by an error backpropagation method and a gradient descent method based on a difference of each predicted label and a true label of a corresponding time domain signal until the model parameters converge with respect to the training set.

10. The self-powered gas sensing system of claim 7, wherein each of the processed signals included in the training set is a spectral image signal, the machine learning model is a Convolutional Neural Network (CNN) model,

wherein the machine learning model is trained by:

inputting each spectral image signal included in the training set into the CNN model to obtain a prediction label corresponding to each spectral image signal, and

updating model parameters of the CNN model by an error back propagation method and a gradient descent method based on a difference of each predicted label and a true label of a corresponding spectral image signal until the model parameters converge with respect to the training set.

11. A self-powered gas sensing device comprising:

an energy harvester configured to take mechanical energy and convert the mechanical energy into electrical energy;

a breakdown discharger disposed in the gas environment to be tested, the breakdown discharger configured to obtain the electrical energy, wherein the electrical energy causes the breakdown discharger to perform breakdown discharge in the gas environment to be tested, and a current generated by the breakdown discharge causes emission of an electromagnetic wave signal,

wherein the electromagnetic wave signal carries information associated with a property of a gas in the gas environment to be measured.

12. The self-powered gas sensing device of claim 11, wherein the breakdown discharger comprises a first electrode and a second electrode separated via gas insulation in the gas environment to be tested, and

the first electrode and the second electrode are provided with opposite discharge tips, a discharge gap is arranged between the discharge tips, the electric energy forms an electric field between the tips of the first electrode and the second electrode, and the electric field breaks down gas in the gas environment to be detected between the discharge gaps of the discharge tips of the first electrode and the second electrode so as to carry out breakdown discharge.

13. A self-powered gas sensing method for identifying properties of a gas in a gas environment to be measured, comprising:

placing a breakdown discharger in the gas environment to be tested;

acquiring mechanical energy by using an energy collector, and converting the mechanical energy into electric energy;

acquiring the electric energy by using the breakdown discharger, wherein the electric energy enables the breakdown discharger to perform breakdown discharge in the gas environment to be tested, and the current generated by the breakdown discharge causes the emission of an electromagnetic wave signal, and the electromagnetic wave signal carries information related to the property of the gas in the gas environment to be tested; and

and identifying the attribute of the gas in the gas environment to be detected based on the electromagnetic wave signal by using a gas attribute identification device.

14. The self-powered gas sensing method of claim 13, wherein identifying the property of the gas in the gas environment under test based on the electromagnetic wave signal with a gas property identification device comprises:

receiving the electromagnetic wave signal, and converting the electromagnetic wave signal into a time sequence signal of a preset electrical parameter;

performing signal processing on the time sequence signal of the preset electrical parameter to obtain a processed signal; and

based on the processed signals, attributes of gas in the gas environment to be measured are identified.

15. The self-powered gas sensing method of claim 13, wherein the gas property identification device comprises a machine learning model, the self-powered gas sensing method further comprising:

acquiring a plurality of processed signals and real labels of gas attributes corresponding to the plurality of processed signals respectively to serve as a training set; and

training the machine learning model with the training set such that the trained machine learning model is capable of generating, for the processed signals, recognition results indicative of attributes of the gas in the gas environment under test.

Images