An ultrasonic-based insect counting detection method

Abstract

This invention discloses an ultrasonic-based insect counting and detection method, belonging to the field of pest monitoring technology. The method includes three steps: Step 1: Setting up a detection channel in the monitoring area, and arranging at least one set of ultrasonic transmitting and receiving units along the detection channel to form an ultrasonic detection field covering the channel cross-section; Step 2: The transmitting units selectively transmit continuous wave ultrasonic signals and periodic pulse ultrasonic signals, with the center frequency of the transmitted signals ranging from 40kHz to 500kHz; Step 3: The receiving units receive in real time the transmitted wave signals and reflected wave signals after being acted upon by insects within the detection channel. The signals include amplitude changes caused by the insects, transit time shifts, scattering noise, and Doppler frequency shift information. This ultrasonic-based insect counting and detection method does not rely on optical pathways, can work stably in completely dark environments, and can resist strong light interference, solving the problems of poor nighttime performance and susceptibility to changes in light in traditional optical counting schemes.

Description

Technical Field

[0001] This invention relates to the field of pest monitoring technology, specifically to an ultrasonic-based method for counting and detecting insects. Background Technology

[0002] Real-time, automated monitoring of flying insect populations is crucial for agricultural pest and disease early warning and ecological research. Existing technologies mainly include: optical trapping and image recognition, which are greatly affected by lighting conditions, have poor performance at night, and consume high energy for image processing; microwave radar detection, which can penetrate certain obstacles, but has limited ability to distinguish high-frequency Doppler signals generated by the tiny wing movements of insects, and is also costly; and passive acoustic detection, which can only listen to the spontaneous sounds of insects, has a short range, and is easily affected by environmental noise. Ultrasonic waves, on the other hand, are mechanical waves with frequencies higher than the range of human hearing. When ultrasonic waves encounter moving objects, they produce a significant Doppler effect. When insects fly, their wings flap at specific frequencies (usually tens to thousands of hertz). This periodic micro-movement periodically modulates the incident continuous ultrasonic waves, forming characteristic sidebands or discrete harmonics on the echo spectrum with intervals of wing-flapping frequencies, providing a unique "acoustic fingerprint" for insect identification.

[0003] For example, in prior art 1 (Chinese patent application number CN202110657174.5, application date 2021-06-11), a method for insect classification and counting based on convolutional neural networks integrates deep learning and image processing technologies. Deep learning is used to train a semantic segmentation model to classify pests trapped by insect-attracting boards and obtain pest region features. This is then combined with morphological processing in image processing to achieve pest counting. This method is particularly suitable for agricultural pest control. Prior art 2 (application number CN202110657174.5, application date 2021-06-11) Chinese Patent No. CN202110181799.9 (application date: 2021-02-08) presents a rapid method for counting the number of small insects. This method overcomes the shortcomings of existing methods for counting leaf-infested pests, such as poor accuracy and low inspection efficiency. By clamping the leaf to be tested between two filter papers using two glass plates, the insect body breaks, and the insect's body fluid seeps into the filter paper, forming spots. The number of these spots is then counted using a CCD machine vision recognition and positioning system. This method has the advantages of high accuracy, high inspection efficiency, and significantly reduced workload.

[0004] However, separating weak insect modulation signals from complex environmental echoes requires high signal-to-noise ratio processing technology. At the same time, it is difficult to distinguish and count individual insect events from a continuous spectrum stream, which easily leads to repetition or omission, resulting in poor practicality. Therefore, an ultrasonic-based insect counting and detection method has been proposed to effectively solve the above problems. Summary of the Invention

[0005] The purpose of this invention is to provide an ultrasonic-based insect counting and detection method to solve the problems currently existing in the market as described in the background.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an insect counting and detection method based on ultrasound, comprising the following steps:

[0007] Step 1: Set up a detection channel in the monitoring area, and arrange at least one set of ultrasonic transmitting and receiving units along the detection channel to form an ultrasonic detection field covering the channel cross section. Two sets of ultrasonic detection units can be arranged in the front and back along the axial direction of the detection channel to form a dual-channel detection structure with a spacing of L (the value ranges from 10mm to 30mm, and is dynamically adjusted according to the average body length and movement speed of the target insect), which provides a hardware foundation for insect movement direction discrimination and identification of adhering insect groups.

[0008] Step 2: The transmitting unit selectively transmits continuous wave ultrasonic signals or periodic pulse ultrasonic signals. The center frequency of the transmitted signals is 40kHz–500kHz, preferably 100kHz–300kHz (this frequency range balances detection resolution and penetration capability). The continuous wave mode is used to capture the Doppler frequency shift of the wing flapping motion of flying insects, while the pulse mode is used to detect acoustic field disturbances of crawling insects. The transmission parameters of the periodic pulse ultrasonic signals can be dynamically adjusted. The pulse width is set to 100μs–500μs (adapting to the detection requirements of insects with a body length of 1.5mm–10mm), and the repetition period is set to 5ms–20ms. The detection sensitivity and power consumption are balanced by optimizing the pulse parameters.

[0009] Step 3: The receiving unit receives the transmitted or reflected wave signals after the insect has acted on the detection channel in real time. The signals include amplitude changes, transit time shifts, scattering noise and Doppler frequency shift information caused by the insect. Specifically, they include amplitude changes, transit time shifts, scattering noise enhancement and Doppler frequency shift, which provide rich data support for subsequent feature extraction.

[0010] Step 4: The received signal is sequentially amplified, filtered, and converted, with simultaneous extraction of time-domain and frequency-domain feature parameters. Time-domain feature parameters include amplitude attenuation, signal disturbance duration, and transit time offset. Frequency-domain feature parameters include the fundamental frequency of the wingbeats, harmonic components, and symmetrical sideband spacing. Amplitude attenuation is a key parameter in this process. ( To preset the minimum effective amplitude attenuation threshold, a value ranging from 5% to 30% is set based on the size of the target insect and the detection environment. (Signal disturbance duration...) ( The minimum effective disturbance duration is preset, with a value ranging from 50μs to 100μs; The maximum effective perturbation duration is preset, with a value ranging from 1ms to 5ms, all set according to the theoretical time range of the target insect passing through the detection field; a short-time Fourier transform is used to perform time-frequency analysis on the signal to generate a dynamic spectrum, in which the harmonic components satisfy... ( =2,3,..., (This is the fundamental frequency of wing flapping); the symmetrical sideband spacing is equal to the fundamental frequency of wing flapping.

[0011] Step 5: Based on preset thresholds, a dynamic baseline model, and multi-feature fusion criteria, select effective perturbation signals caused by the insect body. The dynamic baseline model updates the baseline signal in real time using a sliding window algorithm (window length 1s–5s), as shown in the formula.

[0012] (in for Time baseline value, This represents the number of sampling points within the window. The sampling interval is... (The signal is a historical sampled signal), which effectively suppresses environmental drift interference; the multi-feature fusion criteria must meet the following requirements: the signal features are significantly different from the interference features caused by environmental airflow and vibration (verified by variance analysis, the ratio of the variance of the interference signal to the variance of the insect signal is ≤0.3), and the matching degree between the frequency domain feature parameters and the preset insect wing flapping frequency database is ≥ a preset threshold (the preset threshold ranges from 80% to 95%).

[0013] Step Six: Perform event correlation, jitter removal, and counting statistics on the effective perturbation signals, and output the cumulative number and target classification information. Event correlation processing includes: tracking the independent event trajectory formed by the signal characteristics corresponding to a single insect in the time-frequency domain, and combining the signal intensity change trend (the signal intensity rises from the initial value to the peak value and then gradually decreases to a complete trajectory cycle) to avoid repeated counting of a single insect during its stay in the detection field. Using a dual-channel detection structure, the insect's movement direction is determined by comparing the time sequence of the effective perturbation signals of the two channels. Only insects moving in one direction are counted to avoid repeated counting caused by traveling back and forth.

[0014] Preferably, in step two, the center frequency of the transmitted signal is preferably 100kHz–300kHz, wherein the continuous wave mode is used to capture the Doppler frequency shift of the wing flapping motion of the flying insect, and the pulse mode is used to detect the acoustic field disturbance of the crawling insect.

[0015] Preferably, the detection channel is arranged with two sets of ultrasonic detection units arranged back and forth along the axial direction to form a dual-channel detection structure with a spacing of L. By comparing the time sequence of the effective disturbance signals of the two channels, the direction of insect movement is determined, and the sticky insect swarm is identified by combining the duration and intensity changes of the signal disturbance.

[0016] Preferably, the time-domain feature parameters extracted in step four include amplitude attenuation, signal disturbance duration, and transit time offset; the frequency-domain feature parameters include the fundamental frequency of the flapping fin, harmonic components, and symmetrical sideband spacing, wherein the amplitude attenuation is ≥ ,in To preset the minimum effective amplitude attenuation threshold, a value ranging from 5% to 30% is set based on the size of the target insect and the detection environment, and the duration of signal disturbance is also preset. ,in The minimum effective disturbance duration is preset, and its value range is [value range missing]. , The maximum effective disturbance duration is preset, with a value ranging from 1ms to 5ms, all set according to the theoretical time range for the target insect to pass through the detection field.

[0017] Preferably, the multi-feature fusion criterion in step five needs to satisfy the following: the signal features are significantly different from the interference features caused by environmental airflow and vibration, and the matching degree between the frequency domain feature parameters and the preset insect wing flapping frequency database is ≥ a preset threshold, wherein the preset threshold ranges from 80% to 95%.

[0018] Preferably, it also includes an environmental compensation step: collecting environmental data in real time using temperature and humidity sensors, and based on a formula Correcting the speed of sound and compensating for the effects of temperature changes on transit time and Doppler shift calculations; among which The value represents the corrected speed of sound. 331.3 is the baseline value of the speed of sound at 0℃ under standard atmospheric pressure. 0.606 is the coefficient of sound speed as a function of temperature. T is the ambient temperature collected in real time.

[0019] Preferably, when multiple sets of overlapping frequency components are detected, a blind source separation algorithm (such as independent component analysis, ICA) is used to separate the signal features of different insects, and a lightweight convolutional neural network model is used to analyze the signal spectrum and output the probability distribution of the number and types of targets.

[0020] Preferably, the event association processing in step six includes: tracking the independent event trajectory formed by the signal features corresponding to a single insect in the time-frequency domain, and combining the signal intensity change trend, wherein the signal intensity rises from the initial value to the peak value and then gradually decreases to a complete trajectory cycle, so as to avoid the single insect being counted repeatedly during its stay in the detection field.

[0021] Preferably, the transmission parameters of the periodic pulsed ultrasonic signal can be dynamically adjusted, and the pulse width is set to... It is suitable for detecting insects with a body length of 1.5mm–10mm, with a repetition period of 5ms–20ms, balancing detection sensitivity and power consumption, and adapting to the detection needs of insects of different sizes.

[0022] Preferably, the preset insect wingbeat frequency database can be updated locally or in the cloud, and includes storage pests, including rice weevils (wingbeat frequency 80Hz–120Hz), grain beetles (wingbeat frequency 100Hz–150Hz), greenhouse flying insects including aphids (wingbeat frequency 120Hz–180Hz), and fruit flies (wingbeat frequency 200Hz–300Hz), as well as the wingbeat frequency and harmonic characteristic range of common targets. It supports the customization of adding or deleting target species data according to the application scenario.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0024] (1) It does not rely on optical pathways and can work stably in completely dark environments (such as inside grain warehouses and greenhouses at night). It can also resist strong light interference, solving the problems of poor nighttime performance and easy influence of light changes in traditional optical counting schemes. It can penetrate thin dust and is not affected by the color and transparency of insects, avoiding the detection failure caused by dust coverage and transparent insects in infrared beam technology. At the same time, it is not affected by environmental humidity fluctuations, overcoming the defects of easy drift and humidity sensitivity of capacitive counting schemes. The detection process is not affected by airflow disturbances or slight vibrations. Through multi-feature fusion criteria, it can effectively distinguish insect signals from environmental interference signals. It is suitable for various complex application scenarios such as grain warehouses, greenhouses, and ventilation ducts.

[0025] (2) It can detect both crawling and flying insects. It can capture the sound field disturbance of small and medium-sized storage crawling pests such as rice weevils and grain beetles through pulse wave mode, and capture the wing-beat Doppler frequency shift of flying insects such as aphids and fruit flies through continuous wave mode. It is suitable for various target insects with a body length ≥1.5mm. Through the dual-channel time sequence orientation structure, it can effectively avoid repeated counting caused by insects traveling back and forth. With the help of signal disturbance duration analysis and blind source separation algorithm, it can accurately identify multiple targets in sticky insect groups and high-density scenes, reducing counting omissions and misjudgments. Combined with the significant signal changes caused by acoustic impedance differences, multi-feature fusion criteria and advanced algorithm optimization, the counting accuracy is ≥90%, providing accurate data support for insect monitoring.

[0026] (3) Overall power consumption is less than 10mW, supports battery power mode, and the average current can be controlled in some scenarios. Within a certain range, it does not require frequent power supply replacements, can meet the needs of long-term unattended monitoring, has a simple structure and does not rely on optical components, does not require frequent cleaning and maintenance, can effectively reduce later operating costs, and is suitable for the decentralized and low-cost monitoring needs in agricultural production.

[0027] (4) It can be flexibly integrated into various equipment or scenarios such as conveyor belts, traps, ventilation ducts, and greenhouse entrance frames to form a "through" detection node, adapting to the installation requirements of different monitoring scenarios. It can not only count insects, but also output insect type classification labels (such as "flower-visiting insects" and "potential pests") through wing vibration frequency database matching, providing refined data support for ecological management and pest control. At the same time, it can be connected to the Internet of Things by communication module to realize remote data upload and centralized management. The ultrasonic power used is low and the frequency exceeds the range of biological hearing, so it will not interfere with insects, crops and ecological environment. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the detection process of the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] like Figure 1 As shown, the present invention provides the following technical solution: a method for counting and detecting insects based on ultrasound.

[0031] Example 1: Grain Warehouse Rice Weevil Monitoring (Pulse Wave Mode + Dual-Channel Detection):

[0032] 1. Detection Scenario Setup: A detection channel is set up inside the grain warehouse ventilation duct. ABS pipe with an inner diameter of 5mm is used as the main body of the detection channel, with black sound-absorbing cotton lined the inner wall to suppress multiple ultrasonic wave reflections and interference. Two sets of ultrasonic transceiver units are arranged along the channel axis, with a spacing L of 15mm, forming a dual-channel detection structure. Based on the average length of the rice weevil (2-3mm) and its movement speed (0.5-1mm / s), the detection is performed using the formula... (in This represents the theoretical time difference for the worm to pass through the two channels. The calculation of the insect's movement speed shows a theoretical time difference of 15-30ms, providing a sufficient time window for direction determination.

[0033] 2. Ultrasonic Emission: The transmitting unit uses a piezoelectric ceramic transducer with a center frequency of 100kHz, selects pulse wave transmission mode, sets the pulse width to 200μs, and the repetition period to 10ms, based on the ultrasonic propagation characteristic formula. (in This refers to the wavelength of ultrasound. The speed of sound at standard atmospheric pressure and 20°C is 343 m / s. (The emission frequency is used), and the calculated wavelength is approximately 3.43 mm, which matches the size of the rice weevil, allowing for precise capture of sound field disturbances caused by the insect; through the formula... (in The average transmit power, Peak power, The pulse width. To optimize power consumption for repetitive cycles, the average power is reduced to less than 5mW.

[0034] 3. Signal Reception and Processing: When the rice elephant passes through the detection channel, the receiving unit captures the transmitted wave signal, amplifies it through a preamplifier (amplification factor set to 40dB), filters it through a 40kHz-500kHz bandpass filter, and then digitizes it through an analog-to-digital converter (sampling rate set to 1MHz); the signal is then processed using the formula... (in The baseline signal amplitude is the signal amplitude when there are no insects present. The amplitude attenuation was calculated based on the signal amplitude when the insect passed through. The measured amplitude attenuation was 15%-25% when the rice weevil passed through.

[0035] 4. Valid signal filtering: based on preset thresholds The dynamic baseline model (with a sliding window length of 2 seconds) filters valid signals; the dynamic baseline model uses the formula... (in for Time baseline value, This represents the number of sampling points within the window. The sampling interval is... (The historical sampling signal) is updated in real time, effectively filtering out high-frequency jitter interference caused by airflow.

[0036] 5. Direction Determination and Counting: By comparing the trigger times of the valid signals of the two channels, if the trigger time of channel 1 is earlier than that of channel 2 and the time difference is within the range of 15-30ms, it is determined that the weevil is moving in the direction of channel 1→channel 2, and only this direction is counted; the valid signal is subjected to 50μs of jitter removal to avoid repeated counting caused by signal jitter; 100 weevils were tested continuously, and 92 counts were successful, with an accuracy of 92%.

[0037] Example 2: Greenhouse aphid monitoring (continuous wave mode + temperature and humidity compensation):

[0038] 1. Setting up the detection scenario: Set up a detection channel on the greenhouse ventilation window frame, and deploy a set of transceiver ultrasonic transducers with beams covering the channel cross section; install a DS18B20 temperature and humidity sensor with a sampling period of 1 second to collect ambient temperature T and relative humidity H in real time.

[0039] 2. Ultrasonic Emission: The transmitting unit uses an ultrasonic transducer with a center frequency of 200kHz, selects continuous wave transmission mode, and continuously emits a single-frequency ultrasonic signal; according to the Doppler frequency shift formula... (in For Doppler frequency shift, The wingbeat speed of aphids is typically taken as 0.5 m / s. This is the speed of sound after real-time correction. (where 'frequency' is the transmission frequency), the calculated frequency shift is approximately 583Hz, falling within the preset 0.1Hz-2kHz signal processing bandwidth, which allows for precise extraction of wing flapping characteristics.

[0040] 3. Signal Reception and Processing: The echo signal modulated by the flapping motion of aphids is captured by the receiving unit and mixed with the local oscillator signal (200kHz) of the transmitting source to obtain an intermediate frequency signal. After being filtered by a 0.1Hz-2kHz bandpass filter, it is digitized by an analog-to-digital converter (sampling rate set to 4kHz). Short-time Fourier transform is used to perform time-frequency analysis on the signal, with a window length of 256ms and an overlap rate of 50%, to generate a dynamic spectrum diagram from which frequency domain features such as symmetrical sideband spacing are extracted.

[0041] 4. Environmental compensation: based on the formula The sound velocity is corrected in real time to compensate for the impact of temperature and humidity changes on the detection. For example, when the greenhouse temperature is 30℃ and the relative humidity is 60%, the calculated sound velocity is about 349.5m / s, which is about 1.9% different from the sound velocity under standard conditions. After correction by this formula, the transit time calculation error is controlled within ±0.5%.

[0042] 5. Feature Matching and Counting: The extracted sideband interval features are matched with the feature range (120Hz-180Hz) of aphids in the preset insect body wingbeat frequency database. If the matching degree is ≥85%, it is determined to be a valid target. By tracking the independent trajectory formed by the features in the time and frequency domain, combined with the signal strength change trend (rising to the peak and then falling), the counting of a single aphid is completed. It adopts an intermittent working mode, wakes up once every 5 seconds for detection, with an average current of <50μA, and supports continuous operation for more than 6 months powered by lithium battery.

[0043] Example 3: Monitoring of high-density mixed populations:

[0044] 1. Detection scenario setup: A detection channel is set up at the exit of the insect behavior research device, and a set of ultrasonic transceiver units with a center frequency of 300kHz are arranged, equipped with a high-performance digital signal processor to support complex algorithm calculations; the inner diameter of the detection channel is set to 8mm to accommodate mixed insects (including rice weevils, fruit flies, aphids, etc.) with a body length of 1.5mm-10mm.

[0045] 2. Ultrasonic transmission and reception: The transmitting unit supports switching between continuous wave and pulse wave modes. To address the characteristics of mixed populations, a time-division transmission strategy is adopted (continuous wave works for 1 second, pulse wave works for 1 second alternately). The receiving unit synchronously acquires transmitted and reflected wave signals, and fuses the features of the two signals through a signal synthesis algorithm to improve multi-target detection capabilities.

[0046] 3. Multi-target signal separation: When multiple sets of overlapping frequency components are detected, an independent component analysis blind source separation algorithm is used, through the formula... (in It is a mixed signal matrix. It is a mixed matrix. The algorithm decomposes the mixed signal (using the source signal matrix) to separate the wing-beating characteristics and sound field disturbance characteristics of different insects. For example, when rice weevils (amplitude attenuation characteristics in pulse wave mode) and fruit flies (200Hz-300Hz wing-beating fundamental frequency in continuous wave mode) are detected at the same time, the algorithm can accurately separate the two types of signals.

[0047] 4. Machine Learning Classification and Counting: The separated signal spectrograms (normalized to 256×256 pixels) are input into a lightweight convolutional neural network model, which uses the cross-entropy loss function. (in Number of insect species For real labels, The model is trained to predict probabilities and outputs the probability distribution of the number and types of targets; the model inference time is <100ms, which meets the requirements of real-time detection.

[0048] 5. Handling and Outputting Adhesive Insect Swarms: When the duration of the signal disturbance t > T_max (set to 5ms) and the amplitude exhibits multi-peak variations, it is identified as an adhesive insect swarm, and peak segmentation algorithm is used to process it. (in For the set of peaks, for Point signal amplitude, The system splits the target into independent targets (based on the peak threshold); combines the direction discrimination results of the dual-channel structure to avoid duplicate counting; outputs the cumulative count and the proportion of various insect species in real time, and uploads them to the cloud management platform via the LoRa module, with a data transmission latency of <1s.

[0049] The above is the entire counting and detection process of the method, and all contents not described in detail in this specification are existing technologies known to those skilled in the art.

[0050] The contents not described in detail in this specification are existing technologies known to those skilled in the art. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for counting and detecting insects based on ultrasound, characterized in that, Includes the following steps: Step 1: Set up a detection channel in the monitoring area, and arrange at least one set of ultrasonic transmitting and receiving units along the detection channel to form an ultrasonic detection field covering the channel cross section; Step 2: The transmitting unit selectively transmits continuous wave ultrasonic signals and periodic pulse ultrasonic signals, with the center frequency of the transmitted signals being 40kHz–500kHz. Step 3: The receiving unit receives the transmitted wave signal and reflected wave signal after the insect body in the detection channel in real time. The signal includes the amplitude change caused by the insect body, the transit time offset, the scattering noise and the Doppler frequency shift information. Step 4: Amplify, filter, and convert the received signal sequentially, and extract time-domain and frequency-domain feature parameters simultaneously. Step 5: Based on preset thresholds, dynamic baseline models, and multi-feature fusion criteria, filter out effective perturbation signals caused by insects; Step 6: Perform event correlation, de-jitter processing, and counting statistics on the valid disturbance signals, and output the cumulative number and target classification information.

2. The method for counting and detecting insects based on ultrasound according to claim 1, characterized in that: In step two, the center frequency of the transmitted signal is preferably 100kHz–300kHz, wherein the continuous wave mode is used to capture the Doppler frequency shift of the wing flapping motion of the flying insect, and the pulse mode is used to detect the acoustic field disturbance of the crawling insect.

3. The insect counting and detection method based on ultrasound according to claim 1, characterized in that: The detection channel is arranged with two sets of ultrasonic detection units arranged back and forth along the axial direction to form a dual-channel detection structure with a spacing of L. By comparing the time sequence of the effective disturbance signals of the two channels, the direction of insect movement is determined, and the sticky insect swarm is identified by combining the duration and intensity changes of the signal disturbance.

4. The insect counting and detection method based on ultrasound according to claim 1, characterized in that: The time-domain feature parameters extracted in step four include amplitude attenuation, signal disturbance duration, and transit time offset; the frequency-domain feature parameters include the fundamental frequency of the flapping fin, harmonic components, and symmetrical sideband spacing, wherein the amplitude attenuation is ≥ ,in To preset the minimum effective amplitude attenuation threshold, a value ranging from 5% to 30% is set based on the size of the target insect and the detection environment, and the duration of signal disturbance is also preset. ,in The minimum effective disturbance duration is preset, and its value range is [value range missing]. , The maximum effective disturbance duration is preset, with a value ranging from 1ms to 5ms, all set according to the theoretical time range for the target insect to pass through the detection field.

5. The insect counting and detection method based on ultrasound according to claim 1, characterized in that: The multi-feature fusion criterion in step five must meet the following conditions: the signal features are significantly different from the interference features caused by environmental airflow and vibration, and the matching degree between the frequency domain feature parameters and the preset insect wing flapping frequency database is greater than or equal to the preset threshold, wherein the preset threshold ranges from 80% to 95%.

6. The insect counting and detection method based on ultrasound according to claim 1, characterized in that: It also includes an environmental compensation step: real-time collection of environmental data via temperature and humidity sensors, based on formulas. Correcting the speed of sound and compensating for the effects of temperature changes on transit time and Doppler shift calculations; among which The value represents the corrected speed of sound. 331.3 is the baseline value of the speed of sound at 0℃ under standard atmospheric pressure. 0.606 is the coefficient of sound speed as a function of temperature. T is the ambient temperature collected in real time.

7. The method for counting and detecting insects based on ultrasound according to claim 1, characterized in that: When multiple sets of overlapping frequency components are detected, blind source separation algorithms (such as independent component analysis, ICA) are used to separate the signal features of different insects. A lightweight convolutional neural network model is then used to analyze the signal spectrum and output the probability distribution of the number and types of targets.

8. The method for counting and detecting insects based on ultrasound according to claim 1, characterized in that: The event association processing in step six includes: tracking the independent event trajectory formed by the signal features corresponding to a single insect in the time-frequency domain, and combining the signal intensity change trend, wherein the signal intensity rises from the initial value to the peak value and then gradually decreases to a complete trajectory cycle, so as to avoid the single insect being counted repeatedly during its stay in the detection field.

9. The method for counting and detecting insects based on ultrasound according to claim 1, characterized in that: The transmission parameters of the periodic pulsed ultrasonic signal can be dynamically adjusted, and the pulse width is set to... It is suitable for detecting insects with a body length of 1.5mm–10mm, with a repetition period of 5ms–20ms, balancing detection sensitivity and power consumption, and adapting to the detection needs of insects of different sizes.

10. The method for counting and detecting insects based on ultrasound according to claim 5, characterized in that: The preset insect wingbeat frequency database can be stored locally and updated in the cloud. It includes storage pests, such as rice weevils (wingbeat frequency 80Hz–120Hz), grain beetles (wingbeat frequency 100Hz–150Hz), greenhouse flying insects including aphids (wingbeat frequency 120Hz–180Hz), and fruit flies (wingbeat frequency 200Hz–300Hz). It also includes the wingbeat frequency and harmonic characteristic range of common targets and supports the addition or deletion of target species data according to the application scenario.

Images