Large-feed grain recognition method and system based on multi-scale wavelet transform

Abstract

The disclosure provides a large feeding amount grain identification method and system based on multi-scale wavelet transform, relating to the technical field of grain loss measurement, a grain loss sensor acquires a grain collision signal, and a grain collision voltage signal converted by a signal processing circuit; the time-frequency energy density of the grain signal feature frequency domain is calculated by wavelet transform under different scales, the time-frequency energy densities under multiple scales are superimposed, and the total energy density is obtained; the grain energy interval constraint layer energy threshold is constructed, and it is judged whether the energy at a certain moment is greater than the grain energy interval constraint layer energy threshold; when the energy at the moment is greater than the grain energy interval constraint layer energy threshold, an event signal is obtained; after obtaining the event signal, a morphological opening and closing operation is performed, a time window threshold is set, and it is judged whether the event signal duration is greater than the time window threshold, if greater, a counting operation is performed, the above steps are repeated until all grain counting is completed, and the disclosure provides a detection accuracy.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of kernel loss measurement, and in particular to a large feeding amount kernel identification method and system based on multi-scale wavelet transform. BACKGROUND

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute the prior art.

[0003] In the corn harvesting link, the application of mechanization and intelligent technology is the key way to reduce labor costs, reduce harvesting losses and improve operation efficiency, which is crucial to food security. With the development of agricultural production towards large-scale, large feeding amount combine harvesters are becoming more and more indispensable in agricultural production. However, as the feeding amount of the combine harvester becomes larger, the precision of the kernel loss measurement sensor of the large feeding amount harvester under high frequency material impact also needs to be further improved. With the development of combine harvesters towards larger feeding amount, the demand for developing high-precision loss sensors suitable for large flow working conditions is particularly urgent.

[0004] The existing scheme mainly uses various methods such as acoustic-electric sensing, pressure sensing and image processing to distinguish kernels from impurities in the cleaned mixture. Sensors based on piezoelectric effect (especially PVDF piezoelectric film) have been successfully applied to kernel loss monitoring of rice, rapeseed, wheat, chickpea and other crops. The principle is that different materials (kernels and impurities) produce different distortions when hitting the piezoelectric film due to differences in physical properties (such as hardness and elasticity), resulting in a proportional polarization charge on the surface of the film. Through signal processing circuit to modulate and analyze the charge signal, use the voltage frequency characteristics in the time domain signal to identify different materials and evaluate the loss amount. However, the existing scheme still has the following problems:

[0005] (1) Circuit complexity and flexibility problem. The signal processing circuit relies on a large number of hardware elements (such as filters, amplifiers, voltage comparators, envelope detectors) for signal conditioning and conversion, resulting in high energy consumption and poor stability of the circuit. Once the signal processing circuit is designed and finalized, it is difficult to flexibly adjust key parameters such as filter parameters, and the adaptability is limited;

[0006] (2) Attenuation time bottleneck and precision / speed limitation: the core problem is that the signal attenuation time is long. The RC time constant of the envelope detector of the traditional sensor is crucial. If the RC is too large, the signal attenuation is slow, which increases the attenuation time and easily causes the pulse of the kernel to be missed under high-speed impact; if the RC is too small, the single kernel signal may be repeatedly counted. Too many capacitive elements in the signal processing circuit of the traditional sensor further exacerbate the problem of slow discharge. Moreover, the signal attenuation time of the traditional sensor is too long, which seriously restricts the detection precision and upper limit of the monitored flow of the sensor. When the kernel flow increases, the impact interval becomes shorter, and the residual voltage (not fully attenuated) of the previous signal will be superimposed on the subsequent signal, causing the duration of the input voltage square wave to be abnormally prolonged, the counting accuracy to be sharply decreased, and the error to be significantly increased. This makes it difficult for the existing sensor to meet the demand of high-flow and high-precision monitoring of large feeding amount combine harvesters.

[0007] (3) Although some studies have tried to shorten the attenuation time (such as to 2-3 ms) by optimizing the material of the sensitive plate or adding elastic damping materials, the damping performance of such materials is often significantly affected by temperature, resulting in unstable performance, and the attenuation problem caused by the circuit itself has not been fundamentally solved.

[0008] (4) The existing piezoelectric thin film type loss sensor adopts a single sensor design, and the whole is a metal substrate-elastic body composite structure, which leads to:

[0009] 1) The broadband damping effect of the elastic body (rubber pad) attenuates the high-frequency components of the corn kernel impact signal, causing the frequency spectrum to shift to low frequencies, and the degree of overlap with the frequency spectrum of the corn cob impact signal to increase;

[0010] 2) The substrate conducts vibration to cause global crosstalk;

[0011] 3) Thermal mechanical stress causes the film to debond. Although attempts have been made to install the film in different regions, the substrate coupling bottleneck has not been broken. SUMMARY

[0012] To solve the above problems, the present disclosure proposes a large feeding amount kernel recognition method and system based on multi-scale wavelet transform. From the physical layer, a hierarchical buffer topology is designed to isolate frequency domain pollution and mechanical crosstalk, improve the signal processing circuit, and break through the attenuation time limitation caused by capacitive load; combined with the kernel recognition algorithm, the energy at different scales is integrated, and the kernel signal is determined by the energy threshold and the energy signal duration.

[0013] According to some embodiments, the present disclosure adopts the following technical solutions:

[0014] A large-feed grain identification system based on multi-scale wavelet transform includes a grain loss sensor, a signal processing circuit, and a processor. The piezoelectric sensing array of the grain loss sensor includes a PVDF piezoelectric film, an array of independent impact conduction units, a vibration decoupling and energy dissipation layer, and a supporting substrate layer. The PVDF piezoelectric film is bonded to the upper surface of the array of independent impact conduction units, and the vibration decoupling and energy dissipation layer is bonded to the lower surface of the array of independent impact conduction units. The supporting substrate layer is disposed below the vibration decoupling and energy dissipation layer.

[0015] The grain loss sensor is connected to the signal processing circuit, which includes a DC-coupled charge amplifier. The PVDF piezoelectric film of the grain loss sensor deforms under the impact force of the grain, and its electrode terminals gather equal amounts of heterogeneous charges. The charges are converted into voltage signals through the signal processing circuit and transmitted to the processor.

[0016] Furthermore, the array-type independent impact conduction unit is a rectangular sensor array composed of multiple electrically isolated stainless steel plates, the interval gap between each stainless steel plate is controlled to be 0.05mm~0.2mm, and a PVDF piezoelectric film is directly bonded to the upper surface of each stainless steel plate by a highly conductive mechanism type epoxy resin glue.

[0017] Furthermore, a vibration decoupling and energy dissipation layer is bonded to the lower surface of the arrayed independent impact conduction unit through a high-temperature vulcanization process. The vibration decoupling and energy dissipation layer is a hydrogenated nitrile rubber cushioning pad. The lower surface of the hydrogenated nitrile rubber cushioning pad is coplanarly bonded to a supporting base layer, and the supporting base layer is an integral stainless steel substrate.

[0018] Furthermore, the signal processing circuit includes a cable capacitor, a DC-coupled charge amplifier, a DC-coupled charge amplifier input resistor, a DC-coupled charge amplifier input capacitor, a feedback capacitor, and a feedback resistor. The PVDF piezoelectric film deforms under the impact force of the grain. Due to the piezoelectric effect, the sensor electrode terminals will accumulate equal amounts of heterogeneous charges. The voltage signal is generated by charging and discharging the feedback capacitor. The decay time of the voltage signal is equivalent to the capacitor discharge time. The smaller the time constant, the faster the voltage decays and the faster the circuit response.

[0019] According to some embodiments, the present disclosure adopts the following technical solutions:

[0020] The recognition method of the large-feed grain recognition system based on multi-scale wavelet transform includes:

[0021] The grain loss sensor obtains the grain collision signal and processes it using the signal processing circuit;

[0022] obtaining a grain collision voltage signal converted by a signal processing circuit;

[0023] The time-frequency energy density of the grain signal characteristic frequency domain is calculated by wavelet transform at different scales, and the time-frequency energy density at multiple scales is superimposed to obtain the total energy density.

[0024] Constructing the energy threshold of the grain energy interval constraint layer to determine whether the energy at a certain moment is greater than the energy threshold of the grain energy interval constraint layer; when the energy at that moment is greater than the energy threshold of the grain energy interval constraint layer, obtaining an event signal;

[0025] After obtaining the event signal, a morphological opening and closing operation is performed, a time window threshold is set, and it is determined whether the duration of the event signal is greater than the time window threshold. If it is greater, a counting operation is performed, and the above steps are repeated until all grains are counted.

[0026] According to some embodiments, the present disclosure adopts the following technical solutions:

[0027] A computer program product comprises a computer program, wherein when the computer program is executed by a processor, the computer program realizes the identification method of the large-feed grain identification system based on multi-scale wavelet transform.

[0028] According to some embodiments, the present disclosure adopts the following technical solutions:

[0029] A non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the recognition method of the large-feed grain recognition system based on multi-scale wavelet transform is implemented.

[0030] According to some embodiments, the present disclosure adopts the following technical solutions:

[0031] An electronic device comprises: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device implements the recognition method of the large-feed grain recognition system based on multi-scale wavelet transform.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] The disclosed large-feed grain recognition system based on multi-scale wavelet transform proposes a hierarchical buffering topology of piezoelectric film → independent micro-steel plate → shock-absorbing rubber → integral substrate, which isolates frequency domain pollution and mechanical crosstalk from a physical level, providing an industrial-grade reliable solution for corn grain loss monitoring; by improving the signal processing circuit, reducing the signal decay time, and improving the resolution of the grain signal, the problems of pulse omission and repeated counting caused by voltage residual superposition under high grain flow are solved, while avoiding the sensitivity drop and temperature drift uncontrollable caused by the existing decay time optimization scheme.

[0034] The disclosed method for identifying grains with large feed amounts based on multi-scale wavelet transform inputs the frequency domain range of a specific signal through a processor, performs multi-scale wavelet transform on the signal within the frequency domain range, integrates the energy at different scales, and performs dual determination of the grain signal through dual determination of the energy threshold and the energy signal duration, thereby improving the flexibility of the monitoring device; superimposes the time-frequency energy density at multiple scales to obtain the sum of the energy density at all scales, and the energy superposition enhances the energy peak of the grain signal (compared with the ambient noise), and the wide-band fusion avoids frequency deviation missed detection due to individual differences in sensors.

[0035] The disclosed method for identifying large feed rates of grains based on multi-scale wavelet transform solves the problem of broken and sticky pulses using morphological closing / opening operations, improves the counting completeness rate, and can maintain an accuracy rate of >92% when the feed rate varies from 12 kg / s to 22 kg / s (the grain flow rate is between 50 grains / s and 90 grains / s). BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings, which constitute a part of the present disclosure, are used to provide a further understanding of the present disclosure. The exemplary embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation to the present disclosure.

[0037] Figure 1 This is a schematic structural diagram of a grain loss sensor according to an embodiment of the present disclosure;

[0038] Figure 2 Schematic diagram of a signal processing circuit according to an embodiment of the present disclosure;

[0039] Figure 3 is a waveform diagram of a measured signal response according to an embodiment of the present disclosure;

[0040] Figure 4 This is a flow chart of the identification method of the large-feed grain identification system based on multi-scale wavelet transform according to an embodiment of the present disclosure;

[0041] Figure 5 The accuracy of the feed rate of the disclosed embodiment is measured when the flow rate of the grains is 50 grains / s;

[0042] Figure 6The accuracy of the feed rate of the disclosed embodiment is measured when the flow rate of the grains is 70 grains / s;

[0043] Figure 7 This is the measured accuracy when the flow rate of the feeding amount of the grains in the embodiment of the present disclosure is 90 grains / s.

[0044] Among them, 101 is an array of independent impact conduction units; 102 is a PVDF piezoelectric film; 103 is a vibration decoupling and energy dissipation layer; and 104 is a supporting base layer. DETAILED DESCRIPTION

[0045] The present disclosure will be further described below with reference to the accompanying drawings and embodiments.

[0046] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present disclosure belongs.

[0047] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0048] Example 1

[0049] In one embodiment of the present disclosure, a large-feed grain recognition system based on multi-scale wavelet transform is provided to address the problems of low grain monitoring accuracy and difficulty in distinguishing grain signals during the existing grain loss measurement process as the feed volume of combine harvesters increases. The present disclosure designs a grain loss sensor and a signal processing circuit, the specific contents of which are as follows:

[0050] like Figure 1 The figure shows the structure of the grain loss sensor. The piezoelectric sensing array of the grain loss sensor adopts a rigid-flexible coupled hierarchical buffer structure, including a PVDF piezoelectric film, an array of independent impact conduction units, a vibration decoupling and energy dissipation layer, and a supporting base layer. The upper surface of the array of independent impact conduction units is bonded to the PVDF piezoelectric film, and the lower surface of the array of independent impact conduction units is bonded to the vibration decoupling and energy dissipation layer. The supporting base layer is disposed below the vibration decoupling and energy dissipation layer.

[0051] Specifically, the array of independent impact transmission units 101 is a rectangular sensor array composed of multiple electrically isolated 304 stainless steel plates, each measuring 80 mm (length) x 12 mm (width) x 1 mm (thickness). Each 304 stainless steel plate is isolated from the others, with the gap between adjacent plates precisely controlled to 0.05-0.2 mm, preferably 0.1 mm in this disclosure. This achieves mechanical decoupling and electromagnetic shielding, eliminating signal crosstalk caused by simultaneous impacts of multiple seeds. The array of independent impact transmission units efficiently transmits the discrete seed impact kinetic energy to the underlying PVDF piezoelectric film. The high rigidity of the 304 stainless steel ensures lossless transmission of impact stress.

[0052] The PVDF piezoelectric film 102 is placed on the upper surface of the array of independent shock conduction units. Specifically, the PVDF piezoelectric film is directly bonded to the upper surface of each stainless steel plate using a highly conductive epoxy resin adhesive. This design is a signal fidelity design. The rigid fixed interface eliminates the stress buffering effect of the traditional soft adhesive layer, ensuring zero delay transmission of the shock stress wave (response time ≤ 1μs). The thickness of the PVDF piezoelectric film is 30μm ± 5% (piezoelectric constant d 33 ≥43pC / N), matching the impact frequency domain of agricultural machinery (typical bandwidth 0.1~50kHz), achieving full-frequency dynamic response fidelity.

[0053] The vibration decoupling and energy dissipation layer 103 is a multi-physical field isolation structure. The vibration decoupling and energy dissipation layer is bonded to the lower surface of the arrayed independent impact conduction unit through a high-temperature vulcanization process. The vibration decoupling and energy dissipation layer is a hydrogenated nitrile rubber (HNBR) shock-absorbing pad, which adopts a parametric shock absorption design. The rubber Shore hardness is 40~60HA, and 52HA is preferred in this disclosure. The thickness is 0.8mm (preferably 1.0mm), forming a dissipation system with a critical damping ratio ζ≈0.7; the interface peel strength is ≥3kN / m (ASTMD903 standard), ensuring that the vibration energy is converted into heat energy through friction within the rubber molecular chain (loss factor tanδ≥0.3).

[0054] The lower surface of the hydrogenated nitrile rubber cushion is coplanarly bonded with a support base layer 104. The support base layer is a monolithic stainless steel substrate and an integrated load-bearing platform with dimensions of 100mm×100mm×1mm. The support base layer is mechanically reinforced and uses a rigid support boundary (deformation tolerance ≤0.01mm) to constrain the lateral creep of the rubber cushion. During application, positioning holes (Ø5mm×4) are reserved at the bolt assembly interface with the agricultural machinery housing to achieve shear-resistant installation. The thermal expansion coefficient of the base (CTE≈17.3×10 -6 / ℃) to match the upper structure and eliminate temperature drift stress.

[0055] like Figure 2As shown, a signal processing circuit is improved. The present disclosure breaks through the attenuation time limitation caused by traditional capacitive loads by reconstructing the topology architecture of the signal processing circuit and selecting components. The specific implementation method is as follows:

[0056] The signal processing circuit includes a cable capacitor, a DC-coupled charge amplifier, a DC-coupled charge amplifier input resistor, a DC-coupled charge amplifier input capacitor, a feedback capacitor, and a feedback resistor. The DC-coupled charge amplifier uses an ultra-low bias current operational amplifier to construct a charge-to-voltage conversion stage to reduce capacitive components.

[0057] As an embodiment, the PVDF piezoelectric film deforms under the impact force of the grain. Due to the piezoelectric effect, the sensor electrode terminals will gather equal amounts of heterogeneous charges, which will pass through the signal processing circuit. In the signal processing circuit, including the capacitor and resistors , the sensor can be considered as a capacitor and resistors ; Also includes cable capacitance , DC coupled charge amplifier input resistance and input capacitance , feedback capacitor , the feedback capacitor realizes the conversion of charge to voltage signal, It is the feedback resistor, which provides a discharge path for the feedback capacitor. With feedback capacitor Parallel, impedance Take the form of complex numbers. The real part is the resistance and the imaginary part is the capacitance. Analyzing the circuit, we know that:

[0058] (1)

[0059] in, is plural;

[0060] When the feedback resistor When it is big enough, .

[0061] The total circuit charge is the sum of the charges on each capacitor:

[0062] (2)

[0063] in, is the parasitic charge of the piezoelectric film, is the cable charge, is the operational amplifier input capacitor charge, is the charge of the feedback capacitor.

[0064] When the open-loop gain When it is big enough, ,

[0065] (3)

[0066] (4)

[0067] in, is the input voltage on both sides of the operational amplifier, is the output voltage of the amplifier circuit, is the total charge, is the open-loop gain, is the feedback capacitor.

[0068] It can be seen from this that the capacitance of the DC coupled charge amplifier itself and the length of the cable will not affect the output of the charge amplifier. The output voltage depends only on the input charge. And it is related to the feedback capacitor. So the feedback capacitor The choice of size determines the voltage.

[0069] Since the voltage signal is generated through the feedback capacitor Charging and discharging occur, so the decay time of the electrical signal can be equivalent to the capacitor discharge time Time constant The smaller it is, the faster the voltage decays and the faster the circuit responds. The circuit can be analyzed as an RC circuit:

[0070] (5)

[0071] (6)

[0072] in, and are the capacitor and resistor currents (vector).

[0073] From the initial Available,

[0074] (7)

[0075] (8)

[0076] Among them, formula (7) is the voltage obtained by integrating formula (6): The expression of Over time The expression of the change. Formula (8) is obtained by taking the time derivative of Formula (7). Represents the voltage at time 0.

[0077] exist At this moment, the initial voltage decay rate can be obtained:

[0078] (9)

[0079] The time required for the voltage to decay to 0 at the initial rate:

[0080] (10)

[0081] Among them, formula (10) is obtained by the decay rate of formula (9), using the initial voltage Divide by the time constant obtained from formula (9).

[0082] exist When the voltage is:

[0083] (11)

[0084] The time constant is effective. The faster the signal decays, the shorter the time it takes for each signal to respond and recover to stability, and the higher the upper limit of the grain monitoring rate. The internal structure of the charge amplifier forms a high-pass filter. Using formula (12), the cutoff frequency is calculated to be 1.6kHz. This can suppress low-frequency noise (such as mechanical vibration interference) while retaining the grain and corncob signals that need to be monitored, meeting the test grain signal frequency band range.

[0085] (12)

[0086] This paper reduces the time constant of the feedback loop to achieve rapid signal attenuation and reset (typical value ≤ 1ms), significantly improving the dynamic performance of the system. The simulation process is specifically reflected as follows:

[0087] When the kernel impact frequency When the traditional circuit does not decay the signal in time, it will cause pulse overlap and generate missed counting errors. Decrease, signal attenuation rate Lift to ensure the interval between two impacts Experiments show that when the grain flow rate is between 12 kg / s and 20 kg / s (50 grains / s to 90 grains / s), the average counting accuracy is above 92%.

[0088] Short time constant Pass the capacitor charge Quick release, The residual voltage drops to 1.8% , eliminating the interference of the long-tail effect on subsequent signals, and improving the detection rate of weak grain collision signals.

[0089] The present invention meets the upper limit of the grain impact frequency when the combine harvester is operating at a speed of 8 to 12 km / h. Grains / s, avoiding signal saturation and distortion. Under corn harvesting conditions with a feed rate of 12kg / s to 20kg / s (grain flow rate of 50 to 90 grains / s), the effective signal capture rate is improved.

[0090] Furthermore, as an embodiment, the signal processing circuit of the present disclosure outputs a voltage signal to a processor, which inputs a specific signal frequency range, performs a multi-scale wavelet transform on the signal within the frequency range, integrates the energy at different scales, and counts and detects the grain signal using a dual determination of energy threshold and energy signal duration. The steps include:

[0091] The grain loss sensor obtains the grain collision signal and processes it using the signal processing circuit;

[0092] Obtaining the seed collision voltage signal converted by the signal processing circuit and screening the characteristic frequency domain of the seed signal;

[0093] The time-frequency energy density is calculated by wavelet transform at different scales, and the time-frequency energy density at multiple scales is superimposed to obtain the sum of the energy density at all scales;

[0094] Constructing the energy threshold of the grain energy interval constraint layer and judging whether the total energy density is greater than the energy threshold of the grain energy interval constraint layer; when the total energy density is greater than the energy threshold of the grain energy interval constraint layer, obtaining an event signal;

[0095] After obtaining the event signal, a morphological opening and closing operation is performed, a time window threshold is set, and it is determined whether the duration of the event signal is greater than the time window threshold. If it is greater, a counting operation is performed, and the above steps are repeated until all grains are counted.

[0096] Example 2

[0097] In one embodiment of the present disclosure, a method for identifying a large-feed grain system based on multi-scale wavelet transform is provided, and the specific steps include:

[0098] Step 1: The grain loss sensor obtains the grain collision signal and processes it using the signal processing circuit;

[0099] Step 2: Obtain the grain collision voltage signal converted by the signal processing circuit and filter the characteristic frequency domain of the grain signal;

[0100] Step 3: Calculate the time-frequency energy density through wavelet transform at different scales, superimpose the time-frequency energy density at multiple scales, and obtain the sum of the energy density at all scales;

[0101] Step 4: Construct the energy threshold of the grain energy interval constraint layer and determine whether the total energy density is greater than the energy threshold of the grain energy interval constraint layer; when the total energy density is greater than the energy threshold of the grain energy interval constraint layer, obtain an event signal;

[0102] Step 5: After obtaining the event signal, perform morphological opening and closing operations, set the time window threshold, and determine whether the duration of the event signal is greater than the time window threshold. If it is greater, perform the counting operation and repeat the above steps until all grains are counted.

[0103] As an embodiment, the specific implementation process of the identification method of the large-feed grain identification system based on multi-scale wavelet transform disclosed in the present invention is as follows:

[0104] Step 1: The grain loss sensor obtains the grain collision signal, processes it using the signal processing circuit and transmits it to the processor, and the processor obtains the grain collision voltage signal converted by the signal processing circuit.

[0105] Specifically, the PVDF piezoelectric film obtains the grain collision signal and converts it into a charge signal. The charge signal outputs a voltage signal through the charge amplifier circuit. The voltage signal enters the AD7606 signal collector, is transmitted to the FPGA development board, is converted into a digital signal, and then is transmitted to the host computer for processing.

[0106] Step 2: The processor obtains the voltage signal, calculates the time-frequency energy density through wavelet transform at different scales, and superimposes the time-frequency energy density at multiple scales to obtain the sum of the energy density at all scales;

[0107] Specifically, wavelet transform can transform a signal as a time function into another time and frequency domain in order to better interpret the original signal in the time domain. The scaling and time shift of the signal Perform time-frequency analysis:

[0108] (13)

[0109] in, is the scale parameter, is the translation (shifting position over time) parameter. is the conjugate of the complex Morlet wavelet. The time-frequency energy density is:

[0110] (14)

[0111] The energy at K scales is superimposed to enhance the significance of the target frequency band. By integrating the total energy at all scales:

[0112] (15)

[0113] The Gaussian window of Morlet wavelet optimizes the Heisenberg uncertainty principle constraint and achieves a time resolution of ≤ 0.1 ms in the characteristic frequency band of grain collision.

[0114] The energy superposition disclosed in the present invention enhances the energy peak of the grain signal (compared with the ambient noise), and the wide-band fusion avoids missed detection of frequency deviations caused by individual differences of sensors.

[0115] Step 3: Construct the energy threshold of the grain energy interval constraint layer and determine whether the total energy density is greater than the energy threshold of the grain energy interval constraint layer; when the total energy density is greater than the energy threshold of the grain energy interval constraint layer, obtain an event signal;

[0116] Specifically, determine whether the total energy density is greater than the energy threshold of the grain energy interval constraint layer:

[0117] (16)

[0118] Energy thresholds constrained by grain energy intervals The signal points that meet the conditions are screened out, and the conditions are combined through Boolean operations to encode and classify the event types, that is, the points with energy greater than the energy threshold are 1, and those with energy less than the energy threshold are 0.

[0119] This threshold maintains a detection rate of >97% during harvester speed change operation.

[0120] Step 4: After obtaining the event signal, perform morphological opening and closing operations, set the time window threshold, and determine whether the duration of the event signal is greater than the time window threshold. If it is greater, perform the counting operation and repeat the above steps until all grains are counted.

[0121] Specifically, the morphological closing operation is used to fill the tiny gaps in the event region as follows:

[0122] (17)

[0123] The morphological opening operation is used to eliminate interference caused by external noise, as follows:

[0124] (18)

[0125] Use closing and opening operations to expand the previous signal. Represents the expansion operation on the signal, Represents the corrosion operation on the signal. The two operations are to fill the gaps in the signal to reduce noise and signal glitches.

[0126] The closed operation disclosed in the present invention repairs the broken pulse caused by signal jitter to achieve pulse integrity, and the open operation eliminates the wide pulse tailing to achieve precise boundary positioning.

[0127] Finally, the joint energy threshold and time window threshold ( ) Determine a valid seed event. A time threshold is applied to the duration of the event signal that meets the energy threshold. If the event duration is greater than the time threshold, the seed count is incremented by one; otherwise, it is not counted.

[0128] In the large-feed grain recognition system based on multi-scale wavelet transform disclosed in the present invention, the micro-gap isolation design (0.1mm) of the array-type independent impact conduction unit in the grain loss sensor enables the spatial resolution to reach The unit, combined with the high rigidity of 304 stainless steel (yield strength ≥ 205MPa), adjacent channel signal isolation to achieve crosstalk suppression, and zero-loss stress transmission, has an impact kinetic energy conduction efficiency of ≥ 97%.

[0129] PVDF piezoelectric film rigid-flexible direct coupling interface achieves ultrafast response: stress wave transmission delay ≤1μs, frequency band expansion, covering the characteristic spectrum of corn kernels; sensitivity leap: charge collection efficiency is improved (d 33 =43pC / N, 10% higher than conventional design).

[0130] The parameterized HNBR damping layer (52HA hardness, 1.0mm thickness, tanδ ≥ 0.3) in the vibration decoupling and energy dissipation layer forms a directional energy dissipation channel. Vibration isolation features attenuation of agricultural machinery broadband vibration (10–800Hz) by ≥26dB. Impact resistance: impact tests have shown no plastic deformation of the rubber pad (GB / T 12832 standard). The supporting base layer is a monolithic stainless steel substrate (using a thermal expansion matching design and anti-shear mounting) to ensure: temperature drift suppression: output drift ≤±0.5% FS within the -40°C to +85°C range; and mechanical deformation resistance: substrate deflection <0.01mm.

[0131] Innovative circuit design to reduce the bottleneck of signal decay time:

[0132] (1) Signal chain anti-interference innovation: output voltage Completely independent of cable length and parasitic capacitance, the transmission error of a 20m cable is compressed to 0.1%;

[0133] (2) Breakthrough of dynamic response limit: initial voltage decay rate Large, supporting ultra-high seed flow monitoring of up to 1000 grains / s per piezoelectric film in theory (compared to 40 grains / s of traditional circuits);

[0134] In addition, the identification method of the large-feed grain recognition system disclosed in the present invention based on multi-scale wavelet transform adopts Morlet wavelet time-frequency focusing combined with multi-scale energy fusion; morphological closing / opening operations solve the problems of broken and sticky pulses, and the counting completeness rate is improved; and the accuracy rate is maintained at >92% when the feed rate varies from 12kg / s to 22kg / s (the grain flow rate is between 50 grains / s and 90 grains / s).

[0135] Example 3

[0136] In one embodiment of the present disclosure, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the computer program implements the identification method of the large-feed grain identification system based on multi-scale wavelet transform.

[0137] Example 4

[0138] In one embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, which is used to store computer instructions. When the computer instructions are executed by a processor, the recognition method of the large-feed grain recognition system based on multi-scale wavelet transform is implemented.

[0139] Example 5

[0140] In one embodiment of the present disclosure, an electronic device is provided, comprising: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the recognition method of the large-feed grain recognition system based on multi-scale wavelet transform.

[0141] The present disclosure is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0142] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate a computer implemented process, so that the instructions executed on the computer or other programmable data processing devices provide a process for implementing the functions specified in the flowchart Figure 1 one flow or multiple flows and / or the functions specified in the block Figure 1 one flow or multiple flows and / or the functions specified in the block

[0143] Although the specific embodiments of the present disclosure are described above with reference to the drawings, the description is not a limitation on the scope of protection of the present disclosure, and those skilled in the art should understand that various modifications or changes made on the basis of the technical solutions of the present disclosure without creative labor are still within the scope of protection of the present disclosure.

Claims

1. A large-feed grain recognition system based on multi-scale wavelet transform, characterized in that: The invention comprises a grain loss sensor, a signal processing circuit and a processor. The piezoelectric sensing array of the grain loss sensor comprises a PVDF piezoelectric film, an array of independent impact conduction units, a vibration decoupling and energy dissipation layer and a supporting base layer. The upper surface of the array of independent impact conduction units is bonded to the PVDF piezoelectric film, the lower surface of the array of independent impact conduction units is bonded to the vibration decoupling and energy dissipation layer, and the supporting base layer is disposed below the vibration decoupling and energy dissipation layer. The array-type independent impact conduction unit is a rectangular sensor array composed of multiple electrically isolated stainless steel plates, the spacing between each stainless steel plate is controlled to be 0.05mm-0.2mm, and a PVDF piezoelectric film is directly bonded to the upper surface of each stainless steel plate using a highly conductive epoxy resin adhesive; the lower surface of the array-type independent impact conduction unit is bonded to a vibration decoupling and energy dissipation layer through a high-temperature vulcanization process, and the vibration decoupling and energy dissipation layer is a hydrogenated nitrile rubber cushion pad, and the lower surface of the hydrogenated nitrile rubber cushion pad is coplanarly bonded to a supporting base layer, and the supporting base layer is a monolithic stainless steel substrate; The seed loss sensor is connected to the signal processing circuit, which includes a DC-coupled charge amplifier. The PVDF piezoelectric film of the seed loss sensor deforms under the impact force of the grain, and its electrode terminals gather equal amounts of heterogeneous charges. The signal processing circuit converts the charges into voltage signals and transmits them to the processor. The specific implementation method of the large-feed grain recognition system of the multi-scale wavelet transform includes: The grain loss sensor obtains the grain collision signal and processes it using the signal processing circuit; obtaining a grain collision voltage signal converted by a signal processing circuit; The time-frequency energy density of the grain signal characteristic frequency domain is calculated by wavelet transform at different scales, and the time-frequency energy density at multiple scales is superimposed to obtain the total energy density. Construct the energy threshold of the grain energy interval constraint layer to determine whether the energy at a certain moment is greater than the energy threshold of the grain energy interval constraint layer; when the energy at that moment is greater than the energy threshold of the grain energy interval constraint layer, obtain the event signal; when the sum of the energy density is greater than the energy threshold of the grain energy interval constraint layer, that is , through the energy threshold of the grain energy interval constraint layer Filter out the signal points that meet the conditions, that is, obtain the event signal; After obtaining the event signal, a morphological opening and closing operation is performed, a time window threshold is set, and it is determined whether the duration of the event signal is greater than the time window threshold. If it is greater, a counting operation is performed, and the above steps are repeated until all grains are counted.

2. The identification method of the large-feed grain recognition system based on multi-scale wavelet transform is specifically based on the large-feed grain recognition system based on multi-scale wavelet transform according to claim 1, characterized in that: include: The grain loss sensor obtains the grain collision signal and processes it using the signal processing circuit; obtaining a grain collision voltage signal converted by a signal processing circuit; The time-frequency energy density of the grain signal characteristic frequency domain is calculated by wavelet transform at different scales, and the time-frequency energy density at multiple scales is superimposed to obtain the total energy density. Constructing the energy threshold of the grain energy interval constraint layer to determine whether the energy at a certain moment is greater than the energy threshold of the grain energy interval constraint layer; when the energy at that moment is greater than the energy threshold of the grain energy interval constraint layer, obtaining an event signal; Wavelet transform transforms the signal as a time function into another time and frequency domain in order to better explain the original signal in the time domain. Continuous wavelet transform is done by using the mother wavelet function Morlet wavelet. The scaling and time shift of the signal Perform time-frequency analysis, superimpose the energy at K scales, and integrate the total energy at all scales. When the total energy density is greater than the energy threshold of the grain energy interval constraint layer, that is, , through the energy threshold of the grain energy interval constraint layer Filter out the signal points that meet the conditions, that is, obtain the event signal; After obtaining the event signal, a morphological opening and closing operation is performed, a time window threshold is set, and it is determined whether the duration of the event signal is greater than the time window threshold. If it is greater, a counting operation is performed, and the above steps are repeated until all grains are counted.

3. The identification method of the large-feed grain identification system based on multi-scale wavelet transform according to claim 2, characterized in that: Wavelet transform transforms the signal as a time function into another time and frequency domain to interpret the original signal in the time domain. Continuous wavelet transform performs time-frequency analysis on the signal through scaling and time shifting of the mother wavelet function Morlet wavelet. A decision function is constructed to determine the grain signal through threshold decision making. The grain energy interval constraint layer is used to screen out signal points that meet the conditions. The conditions are combined through Boolean operations to encode and classify the event types. Specifically, the energy threshold of the grain energy interval constraint layer is The signal points that meet the conditions are screened out, and the conditions are combined through Boolean operations to encode and classify the event types, that is, the points with energy greater than the energy threshold are 1, and those with energy less than the energy threshold are 0.

4. The identification method of the large-feed grain identification system based on multi-scale wavelet transform according to claim 2, characterized in that: The morphological closing operation is used to fill the tiny gaps in the event area, and the morphological opening operation is used to eliminate noise. The time threshold judgment is performed on the duration events of the event signal that meet the energy threshold. If the event duration is greater than the time threshold, the grain count is increased by one, otherwise it is not counted.

5. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the recognition method of the large-feed grain recognition system based on multi-scale wavelet transform according to any one of claims 2 to 4 is implemented.

6. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the recognition method of the large-feed grain recognition system based on multi-scale wavelet transform as described in any one of claims 2 to 4 is implemented.

7. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to execute the recognition method of the large-feed grain recognition system based on multi-scale wavelet transform as described in any one of claims 2 to 4.

Images