Self-calibration MEMS ultrasonic flowmeter based on AI algorithm

Abstract

The invention relates to the technical field of gas flow measurement, in particular to a self-calibration MEMS (micro-electromechanical system) ultrasonic flowmeter based on an AI (artificial intelligence) algorithm, which adopts a high-frequency MEMS ultrasonic probe, a low-power-consumption driving circuit and an acoustic / electrical impedance matching structure to realize efficient transmission and stable excitation of ultrasonic signals; a low-power-consumption microcontroller and an AI compensation module are arranged in the flowmeter, temperature, pressure, time of flight (TOF) and actual sound velocity measured in a static gas state are fused, dynamic correction and intelligent analysis of gas flow are achieved, and the measurement precision in a low-flow-velocity section and under complex working conditions is remarkably improved. The device supports an event-driven low-power-consumption working mode, and long-term battery power supply and modular deployment can be realized; compared with a traditional PZT flowmeter, the scheme has better integration, stability and adaptability while reducing system cost and power consumption, and is suitable for new-generation intelligent gas meters and industrial gas flow monitoring scenes.

Description

Technical Field

[0001] This invention relates to the field of gas flow measurement technology, and in particular to a self-calibrating MEMS ultrasonic flow meter based on AI algorithms. Background Technology

[0002] With the widespread application of particulate fluids in chemical, pharmaceutical, energy, and food processing industries, flow velocity measurement, as a key parameter, is of great significance for process optimization and safety control. Currently, the mainstream methods for measuring particulate fluid flow velocity mainly include:

[0003] 1. Electromagnetic induction flow meter

[0004] This type of sensor utilizes the Faraday principle of electromagnetic induction and is suitable for conductive liquids, but it is ineffective for measuring non-conductive fluids (such as some oils or particulate gas-solid two-phase flows), and the sensor is easily interfered with in high-temperature and high-dust environments.

[0005] 2. Doppler ultrasonic flow meter

[0006] This method determines speed by analyzing echo frequency shift and is suitable for fluids containing particles or bubbles. However, it lacks sensitivity for working conditions with small particle size or few particles, and the signal-to-noise ratio is limited, resulting in a significant decrease in accuracy at low speeds.

[0007] 3. Differential pressure flow meters (such as orifice plates and venturi flow meters)

[0008] The flow velocity is estimated by measuring the pressure difference across the fluid. This method is simple in structure, but it is susceptible to clogging by particle deposition, and the relationship between pressure difference and flow velocity is no longer linear in the low Reynolds number range, making it unsuitable for dynamically changing particulate flows.

[0009] 4. Mechanical impeller flow meter

[0010] The flow rate is sensed by rotating components, but in high-wear, high-dust or strong-impact flow fields, the lifespan is short, the response is sluggish, and maintenance is frequent, so it is gradually being phased out.

[0011] These traditional methods generally suffer from the following problems: poor response to low flow velocities (<0.5m / s) and discontinuous flow fields; particle disturbances and pipe wall interference lead to measurement instability or distortion; sensor structures are prone to wear and blockage, requiring frequent maintenance; and they cannot effectively integrate the multiple physical characteristics of the fluid for intelligent compensation.

[0012] Therefore, there is an urgent need for a flow velocity measurement technology that can stably sense the behavior of particulate fluids without structural contact, especially for unstable flow fields and particulate disturbance scenarios, to improve the overall system's stability, versatility, and low flow velocity accuracy. Summary of the Invention

[0013] The purpose of this invention is to provide a self-calibrating MEMS ultrasonic flow meter based on AI algorithms, which solves the problems of traditional flow meters having low frequency and high driving voltage, resulting in high overall system cost, high power consumption, large size, and being unfavorable for long-term battery power supply and modular design.

[0014] To achieve the above objectives, the present invention provides a self-calibrating MEMS ultrasonic flow meter based on an AI algorithm, comprising:

[0015] MEMS ultrasonic probes are equipped with a dedicated acoustic impedance matching layer and acoustic coupling structure.

[0016] The MOSFET single-ended drive circuit module provides a stable drive voltage for the MEMS ultrasonic probe.

[0017] An electrical impedance matching network element for the drive circuit is disposed between the MOSFET single-ended drive circuit module and the MEMS ultrasonic probe to enhance the ultrasonic signal transmission efficiency of the MEMS ultrasonic probe.

[0018] A low-power microcontroller performs real-time ultrasonic time-of-flight measurement, temperature and pressure compensation calculation, and correction based on the actual gas sound velocity;

[0019] The temperature and pressure sensor module monitors the temperature and pressure of the gas in real time and dynamically corrects the propagation speed of the ultrasonic signal.

[0020] The compensation mechanism element based on the actual gas sound velocity automatically measures the actual gas sound velocity and performs dynamic compensation when the gas is stationary or at a low flow rate.

[0021] The low-power comparator circuit module automatically wakes up the low-power microcontroller to start measurement when a gas flow event is detected;

[0022] The AI ​​algorithm module is built into the low-power microcontroller, enabling the low-power microcontroller to analyze ultrasonic signals, temperature, pressure and real-time sound velocity data in real time.

[0023] AI compensation module, based on Transformer structure or LSTM neural network;

[0024] The fusion processing module supports a dynamic weight adjustment mechanism;

[0025] A low-power monitoring module is used to acquire and process signals triggered by events.

[0026] The environmental monitoring module collects environmental characteristics such as temperature, humidity, and particle concentration.

[0027] The acoustic impedance matching layer and acoustic coupling structure of the MEMS ultrasonic probe are made of thin-layer polymer material or special resin layer.

[0028] The MEMS ultrasonic probe operates at a frequency of 1MHz, meaning it transmits and receives ultrasonic signals at a frequency of 1MHz.

[0029] The driving voltage provided by the MOSFET single-ended driving circuit module is 20V.

[0030] The low-power microcontroller automatically enters a deep sleep mode when no gas flow signal is detected.

[0031] The low-power microcontroller has a built-in floating-point unit (FPU) for real-time ultrasonic time-of-flight measurement, temperature and pressure compensation calculation, and correction based on the actual gas sound speed.

[0032] The temperature and pressure sensor module is connected to the low-power microcontroller via a digital I2C connector.

[0033] The present invention also provides a flow velocity measurement method based on passive acoustics and AI compensation, applied to the aforementioned self-calibrating MEMS ultrasonic flow meter based on AI algorithms, comprising the following steps:

[0034] Feature extraction steps (time domain, frequency domain);

[0035] AI modeling and deployment steps;

[0036] Data fusion steps;

[0037] Output steps and visualization steps.

[0038] The beneficial effects of the self-calibrating MEMS ultrasonic flow meter based on AI algorithm of the present invention are as follows: MEMS ultrasonic technology significantly reduces cost and power consumption, facilitating product miniaturization, mass production, and market promotion; the MOSFET single-ended drive circuit element and impedance matching structure optimize electrical signal transmission efficiency, effectively reducing cost and improving measurement accuracy; the event-driven low-power design extends battery life; the built-in AI algorithm module and dynamic compensation mechanism based on actual gas sound velocity enable real-time intelligent analysis of signal characteristics and environmental changes, improving measurement accuracy and stability; the acoustic impedance matching and coupling structure of the MEMS ultrasonic probe significantly improves ultrasonic energy transfer efficiency, further enhancing system reliability and accuracy; the compensation mechanism element based on actual gas sound velocity can effectively cope with different gas compositions, environmental fluctuations, or sensor errors, improving measurement consistency under complex operating conditions. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a schematic diagram of the MEMS ultrasonic gas flow measurement module of the present invention.

[0041] Figure 2 This is a schematic diagram of the single-ended MOSFET driving circuit and impedance matching network of the present invention.

[0042] Figure 3 This is a schematic diagram of the event-driven low-power workflow of the present invention.

[0043] Figure 4 This is a flowchart of the AI ​​algorithm and dynamic sound speed compensation processing of the present invention. Detailed Implementation

[0044] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0045] Please see Figure 1 ,in Figure 1 This is a schematic diagram of the MEMS ultrasonic gas flow measurement module structure of the present invention. The present invention provides a self-calibrating MEMS ultrasonic flow meter based on an AI algorithm, including the MEMS ultrasonic probe, the MOSFET single-ended drive circuit module, the drive circuit electrical impedance matching network element, the low-power microcontroller, the temperature and pressure sensor module, the compensation mechanism element based on the actual gas sound velocity, the low-power comparator circuit module, and the AI ​​algorithm module.

[0046] The ultrasonic flow meter has at least one pair of ultrasonic transducers, each of which has a MEMS ultrasonic probe for generating an ultrasonic signal. The ultrasonic signal travels along a path between the two ultrasonic transducers. The MEMS ultrasonic probe is single-ended driven by a MOSFET single-ended drive circuit module, which provides a stable 20V drive voltage. The MEMS ultrasonic probe operates at a frequency of 1MHz and features miniaturized size and low power consumption. It is equipped with an acoustic impedance matching layer and acoustic coupling structure using a thin layer of polymer material or a special resin layer to enhance the ultrasonic flow. The propagation efficiency of the ultrasonic signal in the gas medium is improved. The electrical impedance matching network element of the drive circuit adds an electrical impedance matching element between the drive circuit and the MEMS ultrasonic probe, including an impedance matching resistor or an LC low-pass filter network, to optimize the drive waveform, reduce high-frequency reflections and signal loss, and enhance the transmission efficiency of the ultrasonic signal. The low-power microcontroller in the ultrasonic flowmeter performs real-time ultrasonic time measurement through its built-in floating-point arithmetic unit, and, combined with the AI ​​algorithm module and the compensation mechanism element based on the actual gas sound velocity, analyzes and processes the ultrasonic signal, temperature and pressure data, and the actual sound velocity measured when the gas is stationary in real time. The temperature and pressure sensor module uses digital I / O... 2 The C-connector is connected to the low-power microcontroller for real-time monitoring of gas temperature and pressure, and dynamic correction of ultrasonic signal propagation speed. The compensation mechanism element based on actual gas sound velocity is used to automatically measure the actual sound velocity of the gas in the field when a stationary state is detected, including low flow rate or zero flow rate, for sound velocity correction in subsequent flow measurement process, further improving the measurement accuracy of the ultrasonic flow meter. The low-power comparator circuit module is connected to the low-power comparator, and automatically wakes up the low-power microcontroller to start measurement when a gas flow event is detected. When there is no gas flow, the low-power comparator enters a sleep state, which can reduce the overall power consumption of the ultrasonic flow meter. The low-power comparator also has the built-in AI algorithm module, which is used to enable the low-power microcontroller to analyze ultrasonic signals, temperature, pressure and real-time sound velocity data in real time, dynamically compensate for changes in gas composition, and improve the measurement accuracy and stability of the ultrasonic flow meter.

[0047] Please see Figures 2 to 4 , Figure 2 This is a schematic diagram of the single-ended MOSFET driving circuit and impedance matching network of the present invention. Figure 3 This is a schematic diagram of the event-driven low-power workflow of the present invention. Figure 4This is a flowchart of the AI ​​algorithm and dynamic sound velocity compensation processing of this invention. This invention provides a self-calibrating MEMS ultrasonic flow meter based on an AI algorithm, employing a low-power microcontroller (MCU) + AI algorithm module architecture. By collecting raw data from the MEMS ultrasonic probe and the temperature and pressure sensors, combined with on-site sound velocity measurement under static gas conditions, and utilizing time-series modeling and dynamic compensation strategies, it achieves high-precision prediction and real-time feedback of gas flow. This invention also provides a flow velocity measurement method based on passive acoustics and AI compensation, applied to the aforementioned self-calibrating MEMS ultrasonic flow meter based on an AI algorithm, including the following steps:

[0048] Feature extraction steps (time domain, frequency domain);

[0049] AI modeling and deployment steps;

[0050] Data fusion steps;

[0051] Output steps and visualization steps.

[0052] The AI ​​algorithm flow of a flow velocity measurement method based on passive acoustics and AI compensation specifically includes the following steps:

[0053] First, the data acquisition and control process:

[0054] 1. Acquisition Equipment and Architecture: The ultrasonic signal acquisition consists of the MEMS sensor, the MOSFET single-ended drive circuit module, and the electrical impedance matching network element of the drive circuit. The signal is input to the high-speed ADC module, such as AD7380, via a low-noise preamplifier circuit to ensure high-resolution signal sampling. The low-power microcontroller (MCU) takes over the buffering and preliminary processing of the ADC data. When higher precision ultrasonic time-of-flight (TOF) measurement is required, it is recommended to use an FPGA to implement the cross-correlation / subsampling interpolation algorithm.

[0055] 2. Temperature and pressure data acquisition and synchronization: The temperature and pressure sensor module is connected via I... 2 C is connected to the low-power microcontroller (MCU), and the sampling frequency is configurable, with a default range of 2Hz to 10Hz. The low-power microcontroller (MCU) performs soft timestamp synchronization to ensure pairing with the ultrasonic data. During the sampling process, it is determined whether to enter the "flow velocity stable / stationary" mode based on the actual working conditions. If the stationary condition is met (ΔTOF stable, <0.01μs / s), a "real-time sound velocity remeasurement" is triggered.

[0056] 3. Sound velocity measurement triggering mechanism: The low-power microcontroller (MCU) periodically (e.g., every 10 seconds) or event-driven (detecting a stationary state) enables bidirectional TOF measurement logic to calculate the sound velocity of the gas in the field and use it as a reference for subsequent measurement compensation.

[0057] 4. Data Acquisition Control Flow: The entire acquisition process is controlled by an event-driven mechanism. By default, the low-power microcontroller (MCU) is in a low-power sleep state. Once the low-power comparator circuit module detects a gas flow event, it immediately wakes up the low-power microcontroller (MCU) to perform a complete measurement and analysis cycle. The acquired data is stored in an external SPIFlash and transmitted in batches to the AI ​​algorithm module for historical backtracking.

[0058] Secondly, the detailed steps of data preprocessing:

[0059] 1. Raw signal acquisition: Use a high-speed ADC to sample the data acquired by the receiving channel at equal intervals, and the sampling rate is recommended to be no less than 1 to 5 MSPS; each sampling should cover a complete set of ultrasound pulses (up / down), and it is recommended to buffer no less than 256 to 1024 points.

[0060] 2. Digital Filtering (Noise Suppression): Select an appropriate filter type. For low-power microcontrollers (MCUs) with high real-time requirements, IIR filters (such as 5th-order Butterworth or Chebyshev filters) are recommended. For FPGAs / offline processing with high accuracy requirements, FIR filters (window design method, such as Hamming window) can be used. Bandwidth recommendation: Set the filter bandwidth to ±20% of the center frequency. Implementation method: Supports direct convolution with fixed coefficient tables (FIR) or real-time calculation based on difference equations (IIR). Filtering result verification: Observe whether the signal-to-noise ratio (SNR) improves after noise suppression.

[0061] 3. Baseline correction (zero drift compensation): For each frame of sampled signal, take the average of the first N points (e.g., the first 20 to 30 points) as the baseline drift estimate B; calculate the difference of the entire waveform, S_corrected[n] = S_raw[n] - B; optional, a sliding window can be used to perform dynamic zero drift compensation for different time periods.

[0062] 4. Wavelet denoising (optional): Use Discrete Wavelet Transform (DWT) to decompose the signal into multiple scales; perform soft / hard thresholding on the high-frequency subband (to remove spike interference); reconstruct the signal to obtain a high-fidelity denoised version; note that when resources are limited on the low-power microcontroller (MCU) side, it can be replaced by bandpass filtering + mean smoothing.

[0063] Then, the feature extraction steps include:

[0064] 1. Ultrasonic Time-of-Flight (TOF) Extraction:

[0065] Use cross-correlation functions to match uplink and downlink signals (Cross-Correlation):

[0066]

[0067] Where Rxy(τ) is the cross-relation function between signals x[n] and y[n], representing the similarity of the two signals under lag τ; x[n] is the discrete sampled value sequence of the upstream received signal; y[n] is the discrete sampled value sequence of the downstream received signal; τ is the number of sample points for the lag time, in units of sample points (not seconds); N is the sampling length, representing the total sampling length of the signal.

[0068] Find the peak position of cross-correlation, τ max Converted to time difference:

[0069] TOF diff =τ max ×T s

[0070] Where, τ max T is the number of delayed sample points corresponding to the maximum value of the cross-correlation function; s It is the sampling period (e.g., 1µs represents a sampling rate of 1MHz); TOF diff The time-of-flight difference is measured in seconds or nanoseconds. Parabolic interpolation can be performed near the peak point to improve time resolution (upgrading to subsampling accuracy, <10ns).

[0071] 2. Amplitude characteristics:

[0072] Extract the maximum amplitude Amax or root mean square amplitude RMS of the signal:

[0073] Maximum amplitude:

[0074] Amax = max(|x[n]|)

[0075] Root mean square amplitude:

[0076]

[0077] Where Amax is the maximum instantaneous amplitude of the received signal; RMS is the root mean square amplitude of the received signal, reflecting the signal energy or average intensity; and x[n] is the sampling sequence value of the ultrasonic received signal.

[0078] 3. Spectral Feature Extraction:

[0079] Perform an FFT on the filtered signal (N = 256 or 512):

[0080] X(f) = FFT{S[n]}

[0081] Extract the center frequency of the spectrum:

[0082]

[0083] Among them, f c It is the center frequency, representing the frequency at which the centroid of the spectrum is located; f i is the i-th frequency point, in Hz; x(fi) is the complex amplitude of the i-th frequency point in the Fast Fourier Transform (FFT) result.

[0084] Calculate the 3dB bandwidth (used to determine signal quality);

[0085] It can extract indicators such as spectral energy and frequency drift as indicators of health status.

[0086] 4. Construct signal feature vectors:

[0087] The final output feature vector is as follows, which can be used for subsequent fusion in the AI ​​model:

[0088] X feature =[TOF diff A max ,RMS,f C BW 3db SNR est ,R peak ]

[0089] Among them, TOF diff It is the propagation time difference between upstream and downstream signals; A max It is the maximum amplitude of the ultrasound signal (reflecting the coupling condition); RMS is the root mean square amplitude (reflecting signal intensity); f C It is the center frequency, indicating the frequency at which the centroid of the spectrum is located; BW 3db It is -3dB bandwidth (representing spectral width, a measure of signal clarity); SNR est It is the signal-to-noise ratio estimate (used to determine data quality); R peak It is the ratio of the main cross-correlation peak to the secondary peak (to identify reflected interference or waveform distortion).

[0090] Meanwhile, the steps for modeling the fusion of temperature, pressure, and sound velocity include:

[0091] 1. Theoretical speed of sound calculation (for reference or deviation correction):

[0092] The expression for the speed of sound derived from the ideal gas law:

[0093]

[0094] Used to estimate the ideal speed of sound at the current temperature T and pressure P; compared with the measured actual speed of sound C. real Differences are used for dynamic correction.

[0095] 2. On-site measured speed of sound C realThe real-time sound velocity obtained through upstream and downstream TOF calculations can be used as a benchmark for AI compensation.

[0096] 3. Construct a dynamic feature matrix (AI model input):

[0097] F=[TOF,Amplitude,Freq,Features,T,P,C real ]

[0098] By integrating ultrasonic signals, temperature and pressure information, and on-site sound velocity, a complete "dynamic feature" input is formed.

[0099] Among them, C theory γ is the theoretical speed of sound (derived from the formula for calculating the state of a gas), γ is the specific heat ratio of the gas (approximately 1.4 for air); R is the ideal gas constant, approximately 8.314 J / (mol·K); T is the temperature (unit: K); M is the molar mass of the gas (approximately 28.97 g / mol for air); C real It is the actual measured speed of sound (inferred from the upstream and downstream TOF difference); F is the dynamic feature vector, used as input for the AI ​​model.

[0100] Furthermore, the steps of model inference (AI gas flow prediction) include:

[0101] 1. Use a lightweight LSTM / Transformer timing model;

[0102] 2. Based on the feature matrix F, predict the current gas flow rate:

[0103] v Pred =AI_Model(F)

[0104] 3. The AI ​​prediction results are time-sensitive to changes in flow velocity trends, making them suitable for scenarios involving pipeline disturbances and multi-component gas mixing.

[0105] 4. Can be deployed on the low-power microcontroller (MCU) platform containing the floating-point arithmetic unit (FPU).

[0106] Among them, v Pred This is the predicted gas flow rate output by the AI ​​model (unit: m / s); AI_Model represents the trained LSTM / Transformer; F is the dynamic feature vector, including multi-dimensional features such as TOF, T, and P.

[0107] Finally, the calculation results are transmitted via UART / I 2 C / Wireless communication output to the gas metering main control terminal; stores daily / hourly total flow, AI model prediction accuracy, and sensor health status logs; supports power outage data retention and abnormal operating condition alarm recording.

[0108] This invention has the following technical innovations:

[0109] Self-calibrating sound velocity measurement mechanism: Automatically triggers TOF interchange measurement in a stationary state and dynamically corrects the sound velocity;

[0110] Event-driven + AI fusion algorithm: It can still achieve high-frequency data perception and intelligent compensation under extremely low power consumption;

[0111] Integrated MEMS transducer driver and impedance matching network improve transmission efficiency:

[0112] The built-in FPU floating-point unit supports AI model execution, enhancing embedded real-time computing capabilities.

[0113] The above description discloses only one preferred embodiment of the present invention, and should not be construed as limiting the scope of the present invention. Those skilled in the art will understand that all or part of the processes of the above embodiments can be implemented, and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A self-calibrating MEMS ultrasonic flow meter based on AI algorithm, characterized in that, include: MEMS ultrasonic probes are equipped with a dedicated acoustic impedance matching layer and acoustic coupling structure. The MOSFET single-ended drive circuit module provides a stable drive voltage for the MEMS ultrasonic probe. An electrical impedance matching network element for the drive circuit is disposed between the MOSFET single-ended drive circuit module and the MEMS ultrasonic probe to enhance the ultrasonic signal transmission efficiency of the MEMS ultrasonic probe. A low-power microcontroller performs real-time ultrasonic time-of-flight measurement, temperature and pressure compensation calculation, and correction based on the actual gas sound velocity; The temperature and pressure sensor module monitors the temperature and pressure of the gas in real time and dynamically corrects the propagation speed of the ultrasonic signal. The compensation mechanism element based on the actual gas sound velocity automatically measures the actual gas sound velocity and performs dynamic compensation when the gas is stationary or at a low flow rate. The low-power comparator circuit module automatically wakes up the low-power microcontroller to start measurement when a gas flow event is detected; The AI ​​algorithm module is built into the low-power microcontroller, enabling the low-power microcontroller to analyze ultrasonic signals, temperature, pressure and real-time sound velocity data in real time. AI compensation module, based on Transformer structure or LSTM neural network; The fusion processing module supports a dynamic weight adjustment mechanism; A low-power monitoring module is used to acquire and process signals triggered by events. The environmental monitoring module collects environmental characteristics such as temperature, humidity, and particle concentration.

2. The self-calibrating MEMS ultrasonic flow meter based on AI algorithm as described in claim 1, characterized in that, The acoustic impedance matching layer and acoustic coupling structure of the MEMS ultrasonic probe are made of thin-layer polymer material or special resin layer.

3. The self-calibrating MEMS ultrasonic flow meter based on AI algorithm as described in claim 1, characterized in that, The MEMS ultrasonic probe operates at a frequency of 1MHz, meaning it transmits and receives ultrasonic signals at a frequency of 1MHz.

4. The self-calibrating MEMS ultrasonic flow meter based on AI algorithm as described in claim 3, characterized in that, The driving voltage provided by the MOSFET single-ended driving circuit module is 20V.

5. The self-calibrating MEMS ultrasonic flow meter based on AI algorithm as described in claim 1, characterized in that, The low-power microcontroller automatically enters a deep sleep mode when no gas flow signal is detected.

6. The self-calibrating MEMS ultrasonic flow meter based on AI algorithm as described in claim 1, characterized in that, The low-power microcontroller has a built-in floating-point unit (FPU) for real-time ultrasonic time-of-flight measurement, temperature and pressure compensation calculation, and correction based on the actual gas sound speed.

7. The self-calibrating MEMS ultrasonic flow meter based on AI algorithm as described in claim 1, characterized in that, The temperature and pressure sensor module uses digital I / O. 2 The C connector is connected to the low-power microcontroller.

8. A flow velocity measurement method based on passive acoustics and AI compensation, applied to the self-calibrating MEMS ultrasonic flow meter based on AI algorithm as described in claims 1-7, characterized in that, Includes the following steps: Feature extraction steps (time domain, frequency domain); AI modeling and deployment steps; Data fusion steps; Output steps and visualization steps.

Images