An unmanned aerial vehicle battery state of health online prediction method

Abstract

The application belongs to the technical field of battery health state monitoring, and relates to a kind of unmanned aerial vehicle battery health state online prediction method, comprising the following steps: by exciting quantum dot tracer in battery monitoring area, obtain initial fluorescence spectrum, and real-time comparison to identify fluorescence quenching caused by gas production, generate preliminary aging signal;Further analyze the spectral drift characteristics to analyze lithium aging and simple gas production;Based on the diagnosis result, trigger ultrasonic detection, extract the time-frequency characteristics of acoustic echo to construct quantitative acoustic fingerprint;Finally, through the cross-validation of optical and acoustic information, a comprehensive health state report containing risk level and aging type is generated;The application realizes early and multi-dimensional online diagnosis and early warning of battery aging, and overcomes the problems of hysteresis, positioning difficulty and inaccurate risk assessment caused by the dependence of traditional methods on macroscopic electric parameters.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of battery health state monitoring, and relates to an unmanned aerial vehicle battery health state online prediction method. BACKGROUND

[0002] The safety and endurance of the lithium ion battery of the unmanned aerial vehicle directly affect the operation efficiency and flight safety thereof. With the cycle use and time elapse, aging occurs in the battery, which is specifically manifested as capacity attenuation, internal resistance rise, and possibly accompanied by physical and chemical state deterioration such as gas generation and lithium precipitation. These changes not only reduce the performance of the unmanned aerial vehicle, but also constitute potential safety hazards, and therefore, early, accurate and multi-dimensional online diagnosis of the aging state of the battery becomes a technical challenge to be solved in the field, and the key lies in how to capture and distinguish the subtle features of different aging modes in a non-invasive manner in real time.

[0003] At present, the health management method based on the battery management system is generally adopted in the industry, the macro electrical parameters such as voltage, current and temperature are monitored, and a battery model is established to estimate the health state and the remaining service life. Although this kind of method can reflect the overall electrical performance attenuation trend of the battery, its sensing dimension is limited to external electrical variables, and the identification sensitivity of early signs of local aging such as gas generation and lithium precipitation in the battery is insufficient, and it is difficult to obtain direct information of physical deformation and chemical changes in the battery in real time, and it is difficult to obtain direct information of physical deformation and chemical changes in the battery in real time.

[0004] However, the existing method only relies on macro electrical parameters and model estimation, resulting in single diagnosis dimension and certain result hysteresis, and it is often difficult to determine aging when the performance of the battery has been significantly reduced, it is difficult to realize early warning, accurately locate the aging area during the operation of the unmanned aerial vehicle, and effectively distinguish high-risk lithium precipitation aging from simple gas generation aging, so that the risk assessment is not accurate enough to support timely and targeted maintenance decisions. SUMMARY

[0005] In order to solve the above problems, the application provides an unmanned aerial vehicle battery health state online prediction method.

[0006] An unmanned aerial vehicle battery health state online prediction method, comprising the following steps:

[0007] S1, exciting quantum dot tracers located in the unmanned aerial vehicle battery monitoring area to generate an initial fluorescence spectrum containing brightness and wavelength information;

[0008] S2, taking the initial fluorescence spectrum as a reference, identifying the fluorescence quenching phenomenon caused by gas generation by comparing the real-time fluorescence spectrum to generate a preliminary aging optical signal;

[0009] S3, analyze the spectral drift characteristics of the preliminary aging optical signal to determine the aging type, and generate a classified aging diagnosis result containing the aging type;

[0010] S4, trigger acoustic pulse detection based on the classified aging diagnosis result, emit an ultrasonic pulse to the quantum dot tracer, and obtain an original acoustic echo carrying physical deformation information;

[0011] S5, extract the multi-dimensional time-frequency characteristics of the original acoustic echo, and construct a quantitative acoustic fingerprint;

[0012] S6, cross-verify the classified aging diagnosis result and the quantitative acoustic fingerprint, and generate a comprehensive health status report.

[0013] Further schemes of the present application, generating an initial fluorescence spectrum, include the following steps:

[0014] Deploy a monitoring area on the preset position of the surface of the unmanned aerial vehicle battery, and a flexible sensing patch containing a microfluidic channel and a quantum dot tracer is attached in the monitoring area; wherein the quantum dot tracer is a mixed suspension containing oxygen-sensitive A-type quantum dots and temperature-sensitive B-type quantum dots;

[0015] Use a blue light emitting device to irradiate the quantum dot tracer to excite it to emit composite fluorescence;

[0016] Collect and analyze the composite fluorescence by a micro-spectrometer to generate an initial fluorescence spectrum.

[0017] Further schemes of the present application, generating a preliminary aging optical signal, include the following steps:

[0018] Continuously collect real-time fluorescence spectra, and extract the spectral intensity value contributed by the A-type quantum dots;

[0019] Compare the real-time spectral intensity value with the reference intensity value in the initial fluorescence spectrum;

[0020] If the intensity decreases by more than a preset quenching threshold, it is determined that fluorescence quenching occurs, and a preliminary aging optical signal containing aging confirmation and regional positioning information is generated.

[0021] Further schemes of the present application, generating a classified aging diagnosis result, include the following steps:

[0022] Extract the peak wavelength of the B-type quantum dots from the real-time fluorescence spectrum triggering the preliminary aging optical signal;

[0023] Calculate the drift amount of the peak wavelength relative to the reference wavelength in the initial fluorescence spectrum;

[0024] According to whether the drift amount exceeds a preset thermal effect threshold, determine whether the aging type is simple gas production aging or lithium precipitation aging accompanied by local temperature rise;

[0025] A classified aging diagnosis result including region location and aging type is generated.

[0026] The further scheme of the present application obtains the original acoustic echo, comprising the following steps:

[0027] According to the region location information in the classified aging diagnosis result, the ultrasonic transducer in the corresponding region is activated;

[0028] The ultrasonic transducer is controlled to emit ultrasonic pulses of a preset frequency and time length into the microfluidic channel;

[0029] The acoustic signal scattered and reflected in the channel is received, and the original acoustic echo is converted.

[0030] The further scheme of the present application constructs a quantitative acoustic fingerprint, comprising the following steps:

[0031] The transit time and energy attenuation rate of the original acoustic echo are calculated;

[0032] The original acoustic echo is subjected to Fourier transform, and the spectral spread width is calculated;

[0033] The transit time, energy attenuation rate and spectral spread width are combined into a vectorized data packet to construct a quantitative acoustic fingerprint.

[0034] The further scheme of the present application generates a comprehensive health status report, comprising the following steps:

[0035] The physical swelling degree level is decoded according to the quantitative acoustic fingerprint;

[0036] The physical swelling degree level is matched and verified with the aging type in the classified aging diagnosis result;

[0037] Based on the matching result, a comprehensive health status report including risk level, region location and aging type is generated.

[0038] The further scheme of the present application determines the aging type, comprising the following steps:

[0039] The difference between the peak wavelength of the B-type quantum dot in the real-time fluorescence spectrum triggering the preliminary aging optical signal and the reference wavelength of the B-type quantum dot in the initial fluorescence spectrum is calculated to obtain the wavelength drift amount;

[0040] The wavelength drift amount is compared with a preset thermal effect threshold value;

[0041] According to the comparison result, the aging type is determined: if the wavelength drift amount is less than the thermal effect threshold value, it is determined as simple gas production aging; if the wavelength drift amount is greater than or equal to the thermal effect threshold value, it is determined as lithium precipitation aging accompanied by local temperature rise.

[0042] The time of flight is a time interval from emitting the ultrasonic pulse to receiving a main echo peak of the original acoustic echo.

[0043] The original acoustic echo is subjected to Fourier transform to convert the signal from a time domain to a frequency domain to obtain a frequency spectrum.

[0044] In summary, the present application has the following beneficial technical effects:

[0045] 1. By fusing optical diagnosis sensitive to chemical changes and acoustic detection sensitive to physical deformation, the multi-dimensional monitoring capability of the battery aging process is provided, and the specific response of the quantum dot tracer to gas and temperature can capture signals from the chemical level at the early stage of aging and distinguish different aging modes, thereby making up for the deficiency of traditional electrical methods in early local defect identification.

[0046] 2. By constructing a quantitative acoustic fingerprint reflecting the degree of battery micro-expansion, the physical deformation that is difficult to directly measure is converted into a quantifiable acoustic characteristic parameter, which provides a measurement approach based on physical evidence for evaluating the severity of aging, reduces the single dependence on empirical models or capacity attenuation rate, and helps to form a more detailed risk assessment and maintenance strategy.

[0047] 3. The dual information cross-verification mechanism of optical recognition of aging type and acoustic quantification of aging degree is adopted, the signals of different physical principles are verified with each other, the possibility of misjudgment caused by environmental interference or failure of a single type of sensor is reduced, and the reliability of the diagnostic conclusion is improved.

[0048] 4. By integrating the microfluidic channel, quantum dot sensing unit and ultrasonic detection component into the flexible patch, an online monitoring system with compact structure and low invasiveness is formed, the integrated design is convenient for the arrangement of the battery surface, does not affect the original structure and heat dissipation of the battery, and can implement point continuous monitoring for specific areas, and is suitable for unmanned aerial vehicle platforms with limited space and load. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. The drawings are used to provide further understanding of the present application, and those skilled in the art can obtain other drawings from these drawings without creative labor.

[0050] Figure 1A flowchart in the embodiments of the application is disclosed.

[0051] Figure 2 A structural schematic diagram in the embodiments of the application is disclosed. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical scheme and advantages of the embodiments of the application clearer, the technical scheme in the embodiments of the application will be clearly and completely explained below in combination with the drawings in the embodiments of the application. Obviously, the described embodiments are some of the embodiments of the application, rather than all the embodiments of the application. All other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the application.

[0053] The above description is made in combination with the drawings in the embodiments of the application. Figure 1 Figure 2 The preferred detailed description of the application is made.

[0054] The above description is made in combination with the drawings in the embodiments of the application. Figure 1 The application provides an unmanned aerial vehicle battery health state online prediction method, comprising the following steps:

[0055] S1, exciting quantum dot tracers located in the unmanned aerial vehicle battery monitoring area to generate an initial fluorescence spectrum containing brightness and wavelength information;

[0056] S2, taking the initial fluorescence spectrum as a reference, identifying the fluorescence quenching phenomenon caused by gas production by comparing a real-time fluorescence spectrum to generate a preliminary aging optical signal;

[0057] S3, analyzing the spectral drift characteristics of the preliminary aging optical signal to determine the aging type to generate a classified aging diagnosis result containing the aging type;

[0058] S4, triggering acoustic pulse detection based on the classified aging diagnosis result, emitting an ultrasonic pulse to the quantum dot tracer to obtain an original acoustic echo carrying physical deformation information;

[0059] S5, extracting the multi-dimensional time-frequency characteristics of the original acoustic echo to construct a quantitative acoustic fingerprint;

[0060] S6, cross-verifying the classified aging diagnosis result and the quantitative acoustic fingerprint to generate a comprehensive health state report.

[0061] In one of the embodiments of the application, step S1 comprises the following steps:

[0062] ​A monitoring area is arranged at a preset position on the surface of the UAV battery, and a flexible sensing patch containing a microfluidic channel and quantum dot tracers is attached in the monitoring area; wherein the quantum dot tracers are a mixed suspension containing type-A quantum dots sensitive to oxygen and type-B quantum dots sensitive to temperature; a blue light emitting device is used to irradiate the quantum dot tracers to excite them to emit composite fluorescence; and a micro spectrometer is used to collect and analyze the composite fluorescence to generate an initial fluorescence spectrum.

[0063] Specifically, first, a monitoring device is arranged on a UAV battery known to be in good health, and a flexible patch integrated with a microfluidic channel is attached to one or more specific monitoring areas on the surface of the battery to ensure that the patch is in close contact with the surface of the battery. The specific monitoring area of the UAV battery refers to an artificially selected area on the surface of the battery, usually the location where the battery is most likely to produce gas, swell or local overheating during the aging process, such as the geometric center of the cell or near the tab, for arranging the sensing device.

[0064] The microfluidic channel is pre-filled with a quantum dot tracer suspension, and the quantum dot tracer is a mixed suspension containing two types of nanoscale semiconductor crystals that can emit different colored fluorescence when excited by light of a specific wavelength. Among them, type-A quantum dots are quantum dots whose fluorescence intensity is sensitive to changes in the concentration of oxygen in the surrounding environment; type-B quantum dots are quantum dots whose fluorescence peak wavelength regularly shifts with temperature changes.

[0065] A driving voltage is applied to a blue light emitting device located at the entrance end of the microfluidic channel to start it and emit blue light of a specific wavelength and power. For example, the blue light emitting device is a micro-LED that can emit blue light with a center wavelength of 405 nanometers, and the light emitting power is set to 5 milliwatts. This beam of blue light will continuously irradiate and penetrate the microfluidic channel, exciting the quantum dot tracers inside the channel. These quantum dot tracer mixed suspensions, when excited, will emit two different types of fluorescence at the same time.

[0066] At this time, the micro-spectrometer located at the outlet end of the microfluidic channel is activated synchronously, and its acquisition system is aligned with the channel outlet for receiving and analyzing the composite fluorescence transmitted through the suspension. It should be noted that the micro-spectrometer is a miniaturized device capable of real-time decomposition of light signals and measurement of light intensity at different wavelengths, with a measurement range of 450 to 750 nanometers and a resolution of 1 nanometer. The micro-spectrometer will decompose the received light signal into different wavelengths of light in real time and measure the light intensity corresponding to each wavelength, thereby integrating these discrete light intensity and wavelength data points into a continuous curve. This curve is the initial fluorescence spectrum, which records the light signal characteristics of the battery monitoring area in the current state in a two-dimensional coordinate system with wavelength as the horizontal axis and light intensity as the vertical axis, serving as the baseline data for subsequent judgment of battery state changes. It should be understood that the initial fluorescence spectrum is a two-dimensional data set, which is an array containing multiple data points, each data point consisting of a pair of numerical values, i.e., (wavelength, light intensity).

[0067] Among them, the A-type quantum dot data structure is a material with a fixed fluorescence peak at 525 nanometers but variable intensity, and the intensity variation range is set based on covering the complete quenching response curve from standard atmospheric oxygen concentration to pure oxygen environment, which can be obtained based on 1000 laboratory calibration experiments. The B-type quantum dot data structure is a material with relatively stable fluorescence intensity but a peak wavelength that changes with temperature, with a peak wavelength of 655 nanometers at 25 degrees Celsius, and a temperature sensitivity set based on a linear relationship between wavelength drift and temperature change within the range of 10 to 80 degrees Celsius, based on thermal cycle testing of more than 500 batches of quantum dot materials. The setting basis for the light emitting power of the blue light emitting device is to effectively excite both types of quantum dots and avoid excessive energy input causing quantum dot photobleaching or additional thermal effects on the battery. This parameter is determined based on experimental measurement of the quantum dot excitation efficiency curve. The microfluidic channel is a micron-scale pipe etched on a flexible substrate, with a width of 100 microns and a depth of 50 microns, designed to accommodate and guide the quantum dot tracer suspension.

[0068] For example, after the deployment of the unmanned aerial vehicle battery specific monitoring area configuration monitoring device at the geometric center of the brand new unmanned aerial vehicle battery and the start of the blue light emitting device with a power of 5 milliwatts, the quantum dot tracer in the microfluidic channel is excited, and the initial fluorescence spectrum generated by the micro-spectrometer shows that there is a fluorescence peak at a wavelength of 525 nanometers produced by A-type quantum dots, with an intensity reading of 950 units. At the same time, there is a fluorescence peak at a wavelength of 655 nanometers produced by B-type quantum dots, with an intensity reading of 880 units. These two peaks and their intensities together constitute this unique initial fluorescence spectrum, serving as the zero point baseline for subsequent battery health state evaluation.

[0069] In one embodiment of the present application, step S2 comprises the following steps:

[0070] The real-time fluorescence spectrum is continuously collected, and the spectral intensity value contributed by the A-type quantum dots is extracted; the real-time spectral intensity value is compared with the reference intensity value in the initial fluorescence spectrum; if the intensity decreases by more than a preset quenching threshold, it is determined that fluorescence quenching occurs, and a preliminary aging optical signal containing aging confirmation and regional positioning information is generated.

[0071] Specifically, the blue light emitting device and the miniature spectrometer remain in working condition, and the excitation and collection processes are repeated uninterruptedly, thereby continuously generating a series of real-time fluorescence spectra. It should be noted that the real-time fluorescence spectrum refers to the fluorescence spectrum collected at any time point during the operation of the monitoring system after the initial fluorescence spectrum is obtained.

[0072] Each time a new real-time fluorescence spectrum is generated, the built-in comparison and analysis module of the system immediately performs data comparison with the previously stored initial fluorescence spectrum. The core operation of the comparison is to accurately extract and monitor the spectral intensity value corresponding to the A-type quantum dots, that is, the light intensity reading at the characteristic wavelength of the A-type quantum dots. It should be noted that the spectral intensity value corresponding to the A-type quantum dots is a specific numerical value, which represents the intensity of the light measured by the miniature spectrometer at the inherent fluorescence peak wavelength position of the A-type quantum dots.

[0073] The real-time measured intensity value is quantitatively compared with the reference intensity value at the same wavelength recorded in the initial fluorescence spectrum, wherein the reference intensity value can be obtained based on calibration under the initial health state of the battery. A percentage value preset for determining whether the fluorescence intensity decreases significantly is used as the quenching threshold, for example, set to thirty percent. When the real-time intensity value presents a sharp decrease by more than the preset quenching threshold compared with the reference intensity value, it is determined that this event is battery aging gas production, and gas molecules penetrate into the microfluidic channel and cause fluorescence quenching phenomenon. It should be noted that the fluorescence quenching phenomenon refers to the physical and chemical process in which the light emitting ability of the A-type quantum dots is inhibited due to chemical or physical action with gas molecules such as oxygen generated by battery aging, resulting in a decrease in fluorescence intensity. Once the determination is made, the intensity mutation event is recorded, and the physical position information of the event is combined to encapsulate a structured data signal, which is the preliminary aging optical signal, indicating that the battery has shown preliminary signs of aging in the monitoring area.

[0074] The preset quenching threshold is set based on the fact that through more than 200 groups of different aging degree batteries for gas production and fluorescence response correlation experiments, it is found that the fluorescence intensity of the A-type quantum dots generally decreases by more than thirty percent in the case of sufficient abnormal gas production. This quenching threshold can effectively filter out the interference caused by device noise and environmental slight fluctuations, ensuring the accuracy of the judgment.

[0075] The preliminary aging optical signal is a digitized information packet, and the data structure thereof comprises two fields, the first field being aging confirmation, which is a Boolean value, and being set to true when the fluorescence quenching phenomenon occurs; and the second field being region positioning, which is used for storing a unique identifier of a specific monitoring region of the UAV battery where the aging phenomenon occurs, for example, a middle region of the cell or an A tab region.

[0076] For example, the monitoring region at the geometric center position of the UAV battery is continuously monitored. Assuming that at a certain moment when the 52nd hour is running, the collected real-time fluorescence spectrum shows that the spectral intensity value of the A-type quantum dot corresponding to the wavelength of 525 nm is 650 units. Compared with the reference intensity of 950 units recorded in the initial fluorescence spectrum, the intensity decreases by 300 units, with a decrease of 31%. Since 31% is greater than the preset quenching threshold of 30%, it is determined that the fluorescence quenching phenomenon caused by gas production occurs. Then, a preliminary aging optical signal is generated, the content of which is aging confirmation - the field value is true, and region positioning - the field value is the geometric center monitoring region, and the signal is transmitted to the downstream processing step.

[0077] In one embodiment of the present application, step S3 comprises the following steps:

[0078] The peak wavelength of the B-type quantum dot is extracted from the real-time fluorescence spectrum triggering the preliminary aging optical signal; the drift amount of the peak wavelength relative to the reference wavelength in the initial fluorescence spectrum is calculated; and according to whether the drift amount exceeds the preset thermal effect threshold, it is determined whether the aging type is pure gas production aging or lithium precipitation aging accompanied by local temperature rise caused by gas production; and a classified aging diagnosis result comprising region positioning and aging type is generated.

[0079] Specifically, the real-time fluorescence spectrum associated with the preliminary aging optical signal is deeply analyzed. First, the real-time fluorescence spectrum data triggering the alarm is locked, and the fluorescence peak contributed by the B-type quantum dot is located. Then, the peak value addressing algorithm is used to accurately extract the peak wavelength of the fluorescence peak. It should be noted that the peak wavelength is the wavelength value corresponding to the maximum light intensity of the fluorescence spectrum emitted by the B-type quantum dot at a specific temperature. Then, the generated initial fluorescence spectrum is retrieved from the storage, and the reference wavelength of the B-type quantum dot is read therefrom. The reference wavelength specifically refers to the peak wavelength of the B-type quantum dot recorded in the initial fluorescence spectrum, which is the reference zero point for subsequent temperature change measurement.

[0080] Then, the measured peak wavelength is subtracted from the reference wavelength to obtain a difference between the two, which is defined as the wavelength shift. It is noted that the wavelength shift is a calculated peak wavelength change, which is positively correlated with temperature change. The calculated wavelength shift is compared with a preset thermal effect threshold. The thermal effect threshold is a critical wavelength shift value representing temperature rise, for example, set to 2.0 nanometers. If the wavelength shift does not exceed the threshold, the aging event is determined to be pure gas generation aging; otherwise, if the wavelength shift is equal to or exceeds the threshold, it indicates that lithium precipitation aging with local temperature rise occurs during gas generation. Based on the determination conclusion, a classification aging diagnosis result including an explicit aging type is finally generated.

[0081] Wavelength shift calculation:

[0082]

[0083] wherein, represents the wavelength shift, which directly reflects the local temperature rise amplitude of the monitoring point. represents the peak wavelength of the B-type quantum dots extracted from the real-time fluorescence spectrum at the same time when the preliminary aging optical signal is generated, and the value is directly measured by a spectrometer; symbol represents the reference wavelength of the B-type quantum dots recorded in the initial fluorescence spectrum, and the value is calibrated at the initial health state of the battery.

[0084] The spectral shift feature refers to the position change of the fluorescence peak wavelength of the B-type quantum dots relative to its initial reference state, which is directly related to the temperature change of the battery monitoring area. The classification aging diagnosis result is a structured data report, and its data structure includes two core fields, one is the area positioning, which is derived from the preliminary aging optical signal, and the other is the aging type, which is set to one of pure gas generation aging or lithium precipitation aging with local temperature rise.

[0085] The basis for setting the thermal effect threshold is that through a large number of battery thermal failure experimental data calibration, it is found that when the wavelength shift of the B-type quantum dots exceeds 2.0 nanometers, the corresponding battery surface temperature rise exceeds 15 degrees Celsius, which is a warning temperature rise value that may generally induce thermal runaway risk.

[0086] Pure gas generation aging refers to a stage in which the battery aging process only shows internal pressure increase without thermal effect; lithium precipitation aging with local temperature rise is a more dangerous aging mode, which not only generates gas, but also may cause internal micro-short circuit and lead to local temperature rise.

[0087] For example, after receiving the preliminary aged optical signal with the region location as the geometric center monitoring area, the real-time fluorescence spectrum at that moment is analyzed. Assuming that the analysis finds that the peak wavelength of the B-type quantum dots is 657.5 nm. The initial fluorescence spectrum is retrieved, in which the recorded reference wavelength of the B-type quantum dots is 655 nm. The calculated wavelength drift is 657.5 nm minus 655 nm, which is 2.5 nm. Since the wavelength drift of 2.5 nm exceeds the set threshold of 2.0 nm for thermal effects, it is determined that this aging event is lithium precipitation aging accompanied by gas production and local temperature rise. Finally, a classified aging diagnosis result is generated, which indicates that the region location is the geometric center monitoring area and the aging type is lithium precipitation aging accompanied by gas production and local temperature rise.

[0088] In one embodiment of the present application, step S4 includes the following steps:

[0089] According to the region location information in the classified aging diagnosis result, the ultrasonic transducer corresponding to the region is activated; the ultrasonic transducer is controlled to emit an ultrasonic wave pulse of a preset frequency and duration into the microfluidic channel; and the original acoustic echo is obtained by converting the received acoustic signal scattered and reflected in the channel.

[0090] Specifically, after confirming the existence of an aging event, the acoustic detection process for the specific region is triggered according to the region location information in the result, and an activation instruction is sent to the ultrasonic transducer located at the inlet end of the microfluidic channel corresponding to the confirmed aging region. It should be noted that the ultrasonic transducer is a device made of piezoelectric material, which has the ability of bidirectional energy conversion, that is, it can convert electrical signals into ultrasonic waves for emission, and also can convert received ultrasonic waves into electrical signals.

[0091] After receiving the instruction, the ultrasonic transducer is driven to convert electrical energy into high-frequency acoustic energy and emit a precisely controlled ultrasonic wave pulse into the quantum dot tracer liquid in the microfluidic channel. The ultrasonic wave pulse is a short-duration high-frequency acoustic energy packet used to detect the internal structure of the medium. Its emission parameters have a preset frequency and a preset duration, such as a center frequency of 5 MHz and a duration of 2 microseconds. This small acoustic wave propagates forward in the channel filled with liquid. Since battery aging may be accompanied by physical expansion, the slight deformation of the microfluidic channel above the battery caused by the slight expansion of the battery will also cause the generation of microbubbles in the liquid. Therefore, the ultrasonic wave pulse will be scattered by these bubbles and reflected from the deformed channel wall during propagation. These complex interactions change the original propagation path, time, and energy of the acoustic wave.

[0092] Finally, the same ultrasonic transducer used for emission is switched to receive mode, or another transducer pre-positioned at the exit end of the channel is responsible for receiving, to capture the sound signals that have been multiply modulated and changed. The received sound signals are converted back to electrical signals, amplified and digitized, and form a complex waveform data recording the sound intensity change over time, i.e. the raw acoustic echo, which is a one-dimensional time series array recording the voltage signal amplitude output by the receiving transducer over time within the receiving time window. This waveform complexly encodes the physical state information in the channel.

[0093] The preset frequency is set according to the following basis: the frequency is high enough to provide sufficient detection resolution to identify the slight deformation of the microfluidic channel and the bubbles, and at the same time, it also ensures that the sound wave has sufficient penetration distance in the quantum dot tracer liquid. The value can be optimized through acoustic simulation and experimental testing of n different sizes and materials of microfluidic channels. The preset time length is set according to the following basis: to ensure that the emitted pulse is short enough to avoid aliasing with the early echo reflected from the proximal end of the channel, thereby improving the resolution of the timing characteristics of the echo signal.

[0094] It should be noted that the battery micro-expansion extrusion is that the battery slightly expands due to internal gas production and other aging mechanisms, thereby exerting mechanical pressure on the microfluidic channel closely attached to its surface, causing the deformation of the microfluidic channel. Bubble scattering refers to the phenomenon that when the sound wave encounters a bubble with a large difference in acoustic impedance from the liquid medium during transmission, the sound energy is scattered in all directions. Deformed channel wall reflection refers to the phenomenon that when the sound wave hits the deformed inner wall of the channel due to pressure, the direction of propagation changes.

[0095] For example, based on the classification aging diagnosis result, it is assumed that lithium precipitation aging accompanied by local temperature rise occurs in the geometric center monitoring area, the ultrasonic transducer corresponding to the area is activated, and it is instructed to emit an ultrasonic pulse with a frequency of 5 MHz and a time length of 2 microseconds into the microfluidic channel. Since the battery has micro-expanded in this area, the channel is slightly narrowed due to extrusion, and there may be small bubbles inside. The sound wave pulse is captured by the receiving mode of the transducer after propagating and returning in it, generating a raw acoustic echo. Compared with the smooth and simple echo in the healthy state, the waveform of the raw acoustic echo shows multiple chaotic subsequent small peaks and overall time delay, which directly reflects the changes in the physical environment inside the channel.

[0096] In one embodiment of the present application, step S5 includes the following steps:

[0097] The time of flight and the energy decay rate of the original acoustic echo are calculated; the original acoustic echo is subjected to Fourier transform to calculate the spectral spread width; the time of flight, the energy decay rate and the spectral spread width are combined into a vectorized data packet to construct a quantized acoustic fingerprint.

[0098] Specifically, after obtaining the original acoustic echo, the signal processing module is started to analyze this complex waveform data. In order to extract the time domain features, the time of flight of the acoustic wave is first calculated, that is, the time interval from the moment of ultrasonic pulse emission to the reception of the main echo peak. It should be noted that the time of flight is a key time domain parameter for measuring the propagation speed or path length change of the acoustic wave, and the unit is microsecond. At the same time, the energy decay rate of the echo is also calculated, which quantifies the energy loss of the acoustic wave in the propagation process by comparing the total energy of the received echo signal with the initial energy of the transmitted pulse. The energy decay rate is an important parameter reflecting the degree of energy loss of the acoustic wave in the propagation process due to medium absorption, scattering and other effects.

[0099] In order to extract the frequency domain features, Fourier transform is applied to the whole original acoustic echo to convert the signal from time domain to frequency domain, revealing which frequency components the signal is composed of. After obtaining the frequency spectrum of the signal, the spectral spread width is calculated, that is, the width of the frequency range in which the spectral energy is mainly distributed. It should be noted that the spectral spread width is a frequency domain parameter for measuring the concentration degree of the signal spectral energy distribution, and the unit is hertz.

[0100] After the calculation of the above-mentioned multi-dimensional feature parameters is completed, these independent numerical values, including the time of flight, the energy decay rate and the spectral spread width, are combined into a fixed data structure according to a predefined order to form a vectorized data packet. This data packet not only contains various information of the acoustic signal, but also its unique numerical combination can uniquely represent the physical deformation degree and internal gas state of the microfluidic channel at the current time, like a fingerprint. The finally generated vectorized data packet, that is, the quantized acoustic fingerprint, has a data structure of a floating point array, such as [time of flight value, energy decay rate value, spectral spread width value], which uniquely identifies the physical state of the specific monitoring area of the battery in a digital form.

[0101] Energy decay rate calculation:

[0102]

[0103] wherein, represents the energy decay rate, which is a dimensionless normalized value, and its value is between 0 and 1, and the higher the value, the greater the energy loss; represents the total energy of the original acoustic echo, which can be obtained by integrating the square of the echo signal in time; The initial energy of the emitted ultrasonic pulse, measured in joules, is known or measured during the calibration phase.

[0104] It should be noted that multidimensional time-frequency features refer to a set of parameters that can comprehensively describe the characteristics of acoustic signals from two different dimensions: time and frequency. Vectorized data packets are a computer-processable data format that arranges multiple independent feature parameter values ​​into a one-dimensional array or vector in a fixed order, facilitating storage, comparison, and machine learning analysis.

[0105] For example, the signal processing module initially calculates a transit time of 4.5 microseconds, which is longer than the 4.1 microseconds in the healthy state. By calculating the echo signal energy and the initial transmitted energy, the energy attenuation rate is found to be 0.65, higher than 0.2 in the healthy state. After Fourier transform analysis of the echo, its spectral width is found to be 1.2 MHz, which is also greater than 0.8 MHz in the healthy state. Finally, these three characteristic parameters [4.5, 0.65, 1.2] are combined into a vectorized data packet. This data packet is the quantitative acoustic fingerprint constructed for the current state of the geometric center monitoring area, which accurately quantifies the physical deformation and internal gas state caused by lithium plating aging due to gas production and local temperature rise.

[0106] In one embodiment of the present invention, step S6 includes the following steps:

[0107] The physical expansion level is decoded based on the quantitative acoustic fingerprint; the physical expansion level is matched and verified with the aging type in the classification aging diagnosis results; and a comprehensive health status report including risk level, regional location and aging type is generated based on the matching results.

[0108] Specifically, after obtaining the classification aging diagnosis results from optical analysis and the quantitative acoustic fingerprint from acoustic detection, the fusion analysis module is initiated to execute the pre-established analysis logic. It should be noted that the analysis logic is a fixed set of decision rules, and its data structure can be a multidimensional lookup table or a set of if-then logical rules. It defines the correspondence between different aging types and different levels of quantitative acoustic fingerprint feature values.

[0109] This analytical logic first correlates and matches the aging type information indicated in the classification aging diagnosis results with the degree of physical expansion reflected in the quantitative acoustic fingerprint. The degree of physical expansion is an interpretation of the physical meaning implied in the quantitative acoustic fingerprint. It is not a directly measured quantity, but rather an indirect and quantitative assessment of the severity of battery physical deformation through characteristic parameters in the acoustic fingerprint.

[0110] Specifically, each feature parameter value in the quantized acoustic fingerprint is compared with the preset severity grading standard, so as to decode the acoustic fingerprint into a level of physical swelling degree, such as slight, moderate or severe. Then, the level of physical swelling degree is matched and verified with the aging type diagnosed by the optical diagnosis, for example, the lithium precipitation aging type with gas production accompanied by local temperature rise should correspond to moderate or severe physical swelling degree. Through this correlation, the state of the battery can be double-cross verified.

[0111] It should be noted that double-cross verification is a method for improving the reliability of diagnosis, which uses measurement data from two different sources and different principles to confirm the same diagnostic conclusion. First, the verification of whether there is aging, that is, the optical signal confirms the aging, and the acoustic signal should also present a fingerprint different from the healthy state; second, the matching verification of the specific aging type and the aging severity, to ensure that the conclusions drawn by the two independent measurement methods support each other and have no contradictions. After the cross verification is completed, a comprehensive health status report will be generated according to the final confirmed conclusion, integrating all the diagnostic information.

[0112] Among them, the aging severity is the final rating of the danger of the current aging state of the battery, which integrates the information of the aging type and the physical swelling degree. The comprehensive health status report is a text or digital file containing multiple fields, recording the whole process of diagnosis and the final conclusion.

[0113] The precise positioning information of the aging area is directly derived from the area positioning field in the classified aging diagnosis result. The aging mode judgment is the aging type field in the classified aging diagnosis result. The risk level assessment is divided into low risk, moderate risk, high risk levels according to the aging severity. The setting basis of this assessment can be: correlating the diagnostic data of more than 1000 battery samples from normal to failure throughout the life cycle with their actual dissection results, establishing a mapping relationship between diagnostic features and real risks.

[0114] For example, after receiving the classified aging diagnosis result of lithium precipitation aging with gas production accompanied by local temperature rise, and the quantized acoustic fingerprint with content [4.5, 0.65, 1.2]. The analysis logic compares the energy attenuation rate 0.65 of the quantized acoustic fingerprint with the severity grading standard, and evaluates the physical swelling degree as severe. Then, the correlation matching is found that the high-risk type of lithium precipitation aging is highly consistent with the severe physical swelling degree, and the double-cross verification is passed. According to this conclusion, the final output comprehensive health status report is obtained, which clearly indicates that lithium precipitation aging with gas production accompanied by local temperature rise occurs in the geometric center monitoring area, and the risk level assessment is high risk, and the operation suggestion of immediately stopping use and maintenance is attached.

[0115] Referring to the accompanying drawings Figure 2The application further provides a UAV battery health state online prediction system, comprising the following modules.

[0116] An initial fluorescence spectrum capturing module is configured to excite quantum dot tracers located in a UAV battery monitoring area to generate an initial fluorescence spectrum containing brightness and wavelength information.

[0117] A fluorescence quenching recognition module is configured to use the initial fluorescence spectrum as a reference to recognize fluorescence quenching phenomena caused by gas production by comparing real-time fluorescence spectra to generate a preliminary aging optical signal.

[0118] An aging type determination module is configured to analyze spectral drift characteristics of the preliminary aging optical signal to determine an aging type and generate a classified aging diagnosis result containing the aging type.

[0119] An acoustic pulse detection module is configured to trigger acoustic pulse detection based on the classified aging diagnosis result, emit ultrasonic pulses to the quantum dot tracers, and obtain original acoustic echoes carrying physical deformation information.

[0120] A quantitative acoustic fingerprint construction module is configured to extract multi-dimensional time-frequency characteristics of the original acoustic echoes to construct a quantitative acoustic fingerprint.

[0121] A comprehensive health state generation module is configured to cross-verify the classified aging diagnosis result and the quantitative acoustic fingerprint to generate a comprehensive health state report.

[0122] The modules can be realized by software, hardware or a combination thereof, and can be embedded in or independent of a processor of a computer device in the form of hardware, and can be stored in a memory of the computer device in the form of software, so as to be called and executed by the processor to perform operations corresponding to the modules.

[0123] The above embodiments are only used to illustrate the technical solutions of the application, and not to limit the same; although the application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalents; the modifications or replacements do not change the essence of the corresponding technical solutions, and should be included in the protection scope of the application.

Claims

1. A method for online prediction of battery state of health of a UAV, characterized in that, The method comprises the following steps: S1, exciting quantum dot tracers located in the unmanned aerial vehicle battery monitoring area to generate an initial fluorescence spectrum containing brightness and wavelength information; S2, taking the initial fluorescence spectrum as a reference, identifying the fluorescence quenching phenomenon caused by gas production by comparing real-time fluorescence spectra to generate a preliminary aging optical signal; S3, analyzing the spectral drift characteristics of the preliminary aging optical signal to determine the aging type, and generating a classification aging diagnosis result containing the aging type; S4, triggering acoustic pulse detection based on the classification aging diagnosis result, emitting ultrasonic pulses to the quantum dot tracers, and obtaining original acoustic echoes carrying physical deformation information; S5, extracting the multi-dimensional time-frequency characteristics of the original acoustic echoes to construct a quantitative acoustic fingerprint; S6, cross-verifying the classification aging diagnosis result and the quantitative acoustic fingerprint to generate a comprehensive health status report; The method for generating an initial fluorescence spectrum containing brightness and wavelength information comprises the following steps: A monitoring area is arranged at a predetermined position on the surface of the unmanned aerial vehicle battery, and a flexible sensing patch containing a microfluidic channel and quantum dot tracers is attached in the monitoring area; wherein the quantum dot tracers are a mixed suspension containing oxygen-sensitive A-type quantum dots and temperature-sensitive B-type quantum dots, and the mixed suspension is arranged in the microfluidic channel; A blue light emitting device located at the inlet end of the microfluidic channel is used to irradiate the quantum dot tracers located in the microfluidic channel to excite them to emit composite fluorescence; The composite fluorescence is collected and analyzed by a miniature spectrometer located at the outlet end of the microfluidic channel to generate an initial fluorescence spectrum; The method for generating a preliminary aging optical signal comprises the following steps: Gas molecules produced by battery aging penetrate into the microfluidic channel, continuously collect real-time fluorescence spectra, and extract the spectral intensity values contributed by the A-type quantum dots; The real-time spectral intensity values are compared with the reference intensity values in the initial fluorescence spectrum; If the intensity decreases by more than a preset quenching threshold, it is determined that fluorescence quenching has occurred, and a preliminary aging optical signal containing aging confirmation and area positioning information is generated; The method for generating a classification aging diagnosis result containing the aging type comprises the following steps: The peak wavelength of the B-type quantum dots is extracted from the real-time fluorescence spectrum that triggered the preliminary aging optical signal; The drift amount of the peak wavelength relative to the reference wavelength in the initial fluorescence spectrum is calculated; According to whether the drift amount exceeds a preset thermal effect threshold, it is determined that the aging type is pure gas production aging or gas production accompanied by local temperature rise lithium precipitation aging; A classification aging diagnosis result containing area positioning and aging type is generated; The method for obtaining original acoustic echoes carrying physical deformation information comprises the following steps: According to the area positioning information in the classification aging diagnosis result, the ultrasonic transducer in the corresponding area is activated; The ultrasonic transducer is controlled to emit ultrasonic pulses of a preset frequency and time length into the microfluidic channel to identify the micro deformation and bubbles of the microfluidic channel; The original acoustic echoes are obtained by converting the received acoustic signals scattered and reflected in the channel. 2.The UAV battery state of health online prediction method of claim 1, wherein, The method for constructing a quantitative acoustic fingerprint comprises the following steps: The transit time and energy attenuation rate of the original acoustic echoes are calculated; The frequency spectrum width of the original acoustic echoes is calculated by Fourier transform. The time of flight, the energy attenuation rate and the spectral spread width are combined into a vectorized data packet to construct a quantitative acoustic fingerprint. 3.The UAV battery state of health online prediction method of claim 1, wherein, Generating a comprehensive health status report, including the following steps: Decoding the physical swelling degree level according to the quantitative acoustic fingerprint; Matching and verifying the physical swelling degree level with the aging type in the classified aging diagnosis result; Generating a comprehensive health status report containing the risk level, the regional positioning and the aging type based on the matching result. 4.The UAV battery state of health online prediction method of claim 1, wherein, Determining the aging type, including the following steps: Calculating the wavelength drift between the peak wavelength of the B-type quantum dots in the real-time fluorescence spectrum triggering the preliminary aging optical signal and the reference wavelength of the B-type quantum dots in the initial fluorescence spectrum; Comparing the wavelength drift with the preset thermal effect threshold value; According to the comparison result, determining the aging type: if the wavelength drift is less than the thermal effect threshold value, determining it as simple gas production aging; if the wavelength drift is greater than or equal to the thermal effect threshold value, determining it as lithium precipitation aging accompanied by local temperature rise.

5. The unmanned aerial vehicle battery state of health online prediction method of claim 2, wherein, The time of flight is the time interval from transmitting an ultrasonic pulse to receiving its main echo peak; the energy attenuation rate is calculated based on the total energy of the original acoustic echo and the initial energy of the ultrasonic pulse, quantifying the energy loss of the sound wave during propagation.

6. The unmanned aerial vehicle battery state of health online prediction method of claim 2, wherein, The original acoustic echo is subjected to Fourier transform to convert the signal from the time domain to the frequency domain, obtaining its frequency spectrum; the spectral spread width is calculated based on the frequency spectrum, wherein the spectral spread width is the width of the frequency range in which the spectral energy is mainly distributed.

Images