Lost grain identification system and method based on energy differentiation method and regression model

Abstract

The invention discloses a lost grain identification system and method based on energy discrimination method and regression model, wherein the system includes a signal acquisition device, a signal processing circuit and a lost grain detection and counting system; the signal acquisition device includes a sensing plate and a piezoelectric element; the signal processing circuit includes a common-phase proportional amplifier circuit, an envelope detection circuit and a common-phase integration circuit; the signal processing circuit also includes three voltage comparator circuits connected in parallel, and the voltage thresholds corresponding to the three voltage comparator circuits are different; the lost grain detection and counting system includes a single-chip microcomputer; the single-chip microcomputer is used to read the signals output by the three voltage comparator circuits, determine the type of the object according to the number of falling edges of the three signals, and count the objects of each type. The invention can realize the accurate discrimination and detection of grains and debris, and provide a reference for adjusting the working parameters of the cleaning system in the actual production of rapeseed combine harvesters.

Description

Technical Field

[0001] The invention relates to the technical field of detecting seeds lost in cleaning during rapeseed harvesting, and in particular to a lost seed recognition system and method based on an energy discrimination method and a regression model. Background Art

[0002] Rapeseed is the largest oil crop in my country and plays an important role in the security of my country's edible oil supply. However, the current problems of high rapeseed mechanized harvesting losses are prominent. Harvest loss is not only an important indicator for measuring the performance of harvesting equipment, but also an important basis for adjusting the relevant working parameters of the harvester. One of the main reasons for the high mechanized rapeseed losses in production is the lack of a relatively accurate harvest loss detection method, which leads to a lack of basis for operators to adjust the working parameters in production.

[0003] In the prior art, patent CN 108370712 A discloses an integrated multi-block double-layer cleaning loss monitoring device with a draft-by-draft comb. In this patent, a draft-by-draft comb and a double-layer monitoring sensor below are arranged at the end of the vibrating screen. During operation, the comb discharges the stems, and the cleaning airflow blows out the light impurities. The grains and a small amount of short stems fall through the gap to the guide plate and finally slide onto the double-layer sensor. The signals generated by the upper and lower layers of the piezoelectric element are amplified, and the machine vibration signal is offset by an adder to retain the material signal; then, the impurity interference is removed by filtering, the grain signal is obtained, and converted into a standard square wave voltage signal. The high-speed counter in the secondary instrument counts these square wave signals, and combines the preset cleaning loss model to calculate and display the cleaning loss in real time. This design ensures that only the grain signal is captured and analyzed to achieve the monitoring of cleaning loss. In this patented method, on the one hand, the short stems and pod shells discharged by the comb teeth will affect the statistics of the grains, resulting in inaccurate grain counting, and the patent requires the use of a pre-built cleaning loss software model to count the grains, which requires a high degree of accuracy in the software model. On the other hand, in actual operation, there is a certain overlap between the frequency domain range of rapeseed grains and the frequency domain range of the remainder, the threshold range is difficult to determine, and the cutoff frequency calculation and circuit design of the filter circuit are relatively complicated. When the cutoff frequency of the signal to be processed changes, the circuit often needs to be redesigned and the parameters adjusted, which is time-consuming and labor-intensive.

[0004] In addition, patent CN115130512A discloses a method and system for detecting grain loss rate based on principal component analysis and neural network, which includes extracting features from the digital quantity of voltage signals when grains and impurities impact the sensing plate, performing principal component analysis on the features, obtaining p principal components, and dividing the evaluation samples where the p principal components are located into a training data set and a test data set. The training data set trains the BP neural network, and calculates the error between the output layer output and the expected output. When the error is greater than the set error, the weight and threshold are adjusted until the error is less than or equal to the set error. The test data set tests the trained BP neural network, outputs the labels of grains and impurities, and calculates the grain loss rate of cereals using the number of grain labels. This method relies on software algorithms such as neural networks, has high requirements on the computing power of computers, and also has high requirements on the accuracy of software models. It is difficult to implement and has high maintenance costs in the later stage. Summary of the invention

[0005] Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides a lost grain identification system and method based on energy differentiation method and regression model, which has the advantages of fast processing speed, strong environmental adaptability, high efficiency, low cost, etc. The present invention aims to achieve accurate differentiation and detection of grains and debris in the cleaning material, and provide a reference for adjusting the working parameters of the cleaning system in the actual production of rapeseed combine harvesters.

[0006] Technical solution: To achieve the above-mentioned purpose, the lost grain identification system based on energy differentiation method and regression model of the present invention comprises a signal acquisition device, a signal processing circuit and a lost grain detection and counting system;

[0007] The signal acquisition device includes a sensing plate and a piezoelectric element;

[0008] The signal processing circuit comprises an in-phase proportional amplifier circuit, an envelope detection circuit and an in-phase integration circuit which are arranged in sequence;

[0009] The signal processing circuit also includes three voltage comparator circuits connected in parallel, and the voltage thresholds corresponding to the three voltage comparator circuits are different; the integrated voltage value output by the in-phase integration circuit is used as the input signal of the three voltage comparator circuits;

[0010] The lost grain detection and counting system includes a single-chip microcomputer; the single-chip microcomputer is used to read the signals output by the three voltage comparator circuits, determine the type of the object according to the number of falling edges of the three signals, and count each type of object; the types of objects include three categories: silicone shells, grains, and short stems, and the characteristics of the signals corresponding to the three types of objects are: 1 falling edge, 2 falling edges, and 3 falling edges, respectively.

[0011] Furthermore, the signal acquisition device also includes a support frame, on which a plurality of the sensing plates are mounted; the four corners of the sensing plates are respectively connected to the support frame by bolts, and a rubber shock absorber is provided between the sensing plates and the support frame; the sensing plates are arranged to be inclined relative to the ground.

[0012] Furthermore, the in-phase proportional amplification circuit includes an input interface and a two-stage amplification circuit, and each stage of the amplification circuit includes a filter capacitor and an operational amplifier; the electrical signal collected by the signal acquisition device is input by the input interface and is processed by the two-stage amplification circuit in sequence.

[0013] Furthermore, the envelope detection circuit includes a detection diode D3, a capacitor C18 and a resistor R12; the amplified signal processed by the in-phase proportional amplifier circuit is input into the anode of the detection diode D3, and the capacitor C18 and the resistor R12 are respectively connected between the cathode of the detection diode D3 and the ground.

[0014] Further, the in-phase integration circuit includes an operational amplifier U4, a resistor R14, a resistor R16, a capacitor C20, a capacitor C23 and a resistor R17, and the resistance values ​​of the resistor R14 and the resistor R16 are equal, and the capacitance values ​​of the capacitor C20 and the capacitor C23 are equal;

[0015] The signal processed by the envelope detection circuit is connected to the positive input terminal of the operational amplifier U4 through the resistor R14; the resistor R16 is connected between the negative input terminal of the operational amplifier U4 and GND;

[0016] The resistor R17 and the capacitor C23 are connected in parallel between the negative input terminal and the output terminal of the operational amplifier U4.

[0017] Furthermore, the voltage comparator circuit includes a voltage comparator, a potentiometer and a voltage stabilizing circuit; the potentiometer is connected to the inverting input terminal of the voltage comparator and is used to set different voltage thresholds; the integrated voltage value output by the in-phase integration circuit is input to the in-phase input terminal of the voltage comparator; and the output terminal of the voltage comparator is connected to the single-chip microcomputer via the voltage stabilizing circuit.

[0018] The lost grain identification method based on energy differentiation method and regression model is based on the above-mentioned lost grain identification system based on energy differentiation method and is implemented by the single chip microcomputer, and the method comprises:

[0019] The signals output by the three voltage comparator circuits are read respectively and recorded as signal A, signal B, and signal C, respectively, wherein the voltage threshold of the voltage comparator circuit corresponding to signal A is the largest, and the voltage threshold of the voltage comparator circuit corresponding to signal C is the smallest;

[0020] Perform rising edge detection and falling edge detection on signal C. When a rising edge is detected, the variable DC corresponding to signal C is marked as true, and the capture process of signal A and signal B is started. The captured signals corresponding to signal A and signal B are saved in variables DA and DB respectively.

[0021] When the falling edge of signal C is detected, the states of variables DA and DB are determined:

[0022] If both variables DA and DB are false, it means the object is a silique shell, and the corresponding silique shell counter increases by 1;

[0023] If the variable DA is true and the variable DB is false, it means the object is a grain, and the corresponding grain counter increases by 1;

[0024] If both variables DA and DB are true, it means that the object is a short stem, and the corresponding short stem counter increases by 1.

[0025] Furthermore, the method for constructing the fan speed-differentiation threshold regression model includes:

[0026] Acquire test data obtained through bench experiments, wherein the test data records the integrated voltage values ​​output by the corresponding in-phase integration circuits of multiple groups of impact tests on the induction plate at different impact speeds using three types of materials, namely, silicone shells, seeds, and short stems;

[0027] Based on the test data, a scatter distribution diagram of the integrated voltage values ​​corresponding to the three types of materials is obtained;

[0028] Calculating optimal discrimination thresholds corresponding to different impact velocities based on the scatter point distribution diagram, wherein the optimal discrimination thresholds include a seed-shell discrimination threshold and a seed-short stalk discrimination threshold;

[0029] According to the correspondence between the fan speed and the impact speed, the numerical correspondence between the fan speed and the optimal distinction threshold is obtained;

[0030] Data fitting is performed based on the numerical correspondence between the fan speed and the optimal distinction threshold value to obtain the fan speed-discrimination threshold value regression model.

[0031] Furthermore, among the three voltage thresholds corresponding to the three voltage comparator circuits, two voltage thresholds with larger values ​​are determined based on the following steps:

[0032] Determine a seed-silique shell distinction threshold and a seed-short stalk distinction threshold based on the current fan speed and the fan speed-discrimination threshold regression model;

[0033] The seed-shell distinction threshold and the seed-short stalk distinction threshold are assigned to two voltage thresholds respectively.

[0034] Beneficial effects: The lost grain identification system and method based on the energy differentiation method of the present invention have the following beneficial effects:

[0035] (1) Based on the fact that the amplitude characteristics of seed grains and residual signals are significantly different from each other through multiple experiments, the present invention innovatively proposes an energy differentiation method to solve the limitation of the existing frequency domain differentiation method in distinguishing rapeseed grains from residuals. This method innovatively proposes a method of converting the impact signals of rapeseed grains and residuals into energy values ​​by designing a hardware integration circuit, based on the significant differences in the amplitude characteristics of the impact signals of rapeseed grains and residuals. The energy differentiation threshold of the impact signals of rapeseed grains and residuals is found through experiments.

[0036] (2) In the hardware integration circuit, when the energy of the seed and debris impact signals changes due to changes in crop status and field operation conditions, the energy distinction threshold can be adjusted by a potentiometer without redesigning the signal processing circuit. Based on the difference in the impact energy of the husk, seed, and short stem on the sensor plate, the hardware integration circuit designed in the present invention can clearly distinguish between seeds and debris through the energy of different signals in actual production, has strong practicality, and is more time-saving and labor-saving than the traditional filter circuit design (frequency domain distinction method).

[0037] (3) Compared with the comparative documents cited in the background technology, the present invention processes the electrical signal generated by the piezoelectric unit through a hardware circuit and utilizes the computing power of a single-chip microcomputer to achieve accurate statistics on the number of lost grains in the cleaned material. It does not rely on computer computing power and constructed computer software models, has low implementation and maintenance costs, is less difficult to implement, and is more conducive to large-scale applications.

[0038] (4) After bench tests, the optimal discrimination thresholds at different impact speeds were obtained. Based on the bench test results, a regression model of fan speed and discrimination threshold was constructed. This regression model can provide a reference for the optimal threshold for distinguishing grains at different fan speeds for grain loss identification during field harvesting. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the composition of the lost grain identification system based on the energy differentiation method and regression model;

[0040] Figure 2 This is a front structural diagram of the signal acquisition device;

[0041] Figure 3 This is a back structural diagram of the signal acquisition device;

[0042] Figure 4 The structure diagram of the rapeseed combine harvester with the loss grain recognition system based on energy differentiation method and regression model;

[0043] Figure 5 It is the circuit principle diagram of the in-phase proportional amplifier circuit;

[0044] Figure 6 It is the circuit schematic diagram of the envelope detection circuit;

[0045] Figure 7 It is the circuit principle diagram of the in-phase integration circuit;

[0046] Figure 8 is a circuit schematic diagram of a three-way voltage comparator circuit;

[0047] Figure 9(a), Figure 9(b), and Figure 9(c) are the results of the impact signals of the husk, seed, and short stalk collected by an oscilloscope from the amplification, detection, and integration parts of the circuit;

[0048] Figures 10(a), 10(b), and 10(c) are the output results obtained by collecting the integral output signals corresponding to the husk, the seed, and the short stalk from the voltage comparator circuit part of the circuit through an oscilloscope;

[0049] Fig.11 It is a flow chart of the method for detecting and counting lost rapeseed grains based on the energy differentiation method;

[0050] Fig.12 It is the structural schematic diagram of the test device;

[0051] Figure 13(a), Figure 13(b), Figure 13(c), and Figure 13(d) are the distribution diagrams of the integrated voltage values ​​of the seeds and impurities when the impact speed is 2.5m / s, 3m / s, 3.5m / s, and 4m / s, respectively;

[0052] Figure 14(a) is a curve diagram showing the relationship between the fan speed and the seed-shell distinction threshold C1;

[0053] FIG14( b ) is a curve diagram showing the relationship between the fan speed and the seed-short stem distinction threshold C2.

[0054] In the figure: 1- rapeseed combine harvester; 2- cleaning screen; 3- signal acquisition device; 4- display screen; 5- support frame; 6- sensor plate; 7- bolt; 8- piezoelectric element; 9- rubber shock absorber; 10- conveyor belt; 11- adjustment frame. DETAILED DESCRIPTION

[0055] The present invention will be further described below in conjunction with the accompanying drawings.

[0056] like Figure 1 The lost grain identification method based on energy differentiation method and regression model shown includes a signal acquisition device 3, a signal processing circuit and a lost grain detection and counting system.

[0057] like Figure 2 and Figure 3 As shown, the signal acquisition device includes a sensing plate 6 and a piezoelectric element 8 fixed at the center point of the back side of the sensing plate 6; it also includes a support frame 5, on which a plurality of sensing plates 6 are installed, so that the signal acquisition device can collect material signals in a wider range; the four corners of the sensing plate 6 are connected to the support frame 5 by bolts 7, and a rubber shock absorber 9 is provided between the sensing plate 6 and the support frame 5, and the rubber shock absorber 9 is mainly used to isolate the influence of environmental noise such as machine vibration; the sensing plate 6 is tilted relative to the ground. In this embodiment, the sensing plate 6 forms an angle of 45° with the ground. According to tests, at this time, the seeds and impurities bounce and move away from the sensing plate 6 after impacting the sensing plate 6, and no secondary impact interference or other problems will occur.

[0058] In this embodiment, the number of induction plates 6 is 4, and the induction plates 6 are 304 stainless steel plates with a length, width and thickness of 150 mm, 150 mm and 0.5 mm respectively. The piezoelectric element 8 is a PZT piezoelectric ceramic, which is fixed to the center point of the back side of the induction plate 6 by epoxy resin glue.

[0059] In actual use, such as Figure 4 As shown, the signal acquisition device is installed at the lower end of the cleaning screen 2 in the rapeseed combine harvester 1. The lost grains and debris are discharged from the tail of the cleaning screen and fall onto the induction plate 6 to impact the induction plate 6.

[0060] The signal processing circuit comprises an in-phase proportional amplifier circuit, an envelope detection circuit and an in-phase integration circuit which are arranged in sequence;

[0061] The signal processing circuit also includes three voltage comparator circuits connected in parallel with each other, the voltage thresholds corresponding to the three voltage comparator circuits are different, and the voltage threshold is also the energy distinction threshold; the integrated voltage value output by the in-phase integration circuit is used as the input signal of the three voltage comparator circuits;

[0062] The lost grain detection and counting system includes a single-chip microcomputer; the single-chip microcomputer is used to read and detect the signals output by the three voltage comparator circuits, determine the type of the object according to the number of falling edges of the three signals, and count each type of object; the types of objects include three categories: silicone shells, grains, and short stems, and the characteristics of the signals corresponding to the three types of objects are: 1 falling edge, 2 falling edges, and 3 falling edges, respectively.

[0063] During operation, the cleaning material hits the sensing plate 6, causing the piezoelectric element 8 to generate an electrical signal of impact vibration. The signal obtained after the electrical signal is processed by the in-phase proportional amplifier circuit, the envelope detection circuit and the in-phase integration circuit can represent the energy of the object that hits the sensing plate 6. The above signal is compared with the energy distinction threshold between rapeseed grains and impurities established based on the hardware integration circuit through a three-way voltage comparator circuit, and the signals corresponding to rapeseed grains, short stems and pod shells are identified and converted into three different types of pulse signals. In addition to the single-chip microcomputer, the loss grain detection and counting system also includes a display screen 4. The single-chip microcomputer identifies and counts various types of objects based on the pulse signal. The display screen 4 is used to display the number of detected grains and different impurities, as well as to display the cleaning loss situation in real time.

[0064] The above-mentioned lost grain identification method is to establish the energy distinction threshold between rapeseed grains and impurities through a hardware integration circuit, so as to accurately identify and count lost grains from the cleaning materials, and can also identify and count husks and short stems. This method is conducive to the subsequent accurate calculation of the cleaning loss situation, and plays a guiding role in the composition analysis of grains and impurities in the cleaning materials.

[0065] Specifically, Figure 5 As shown, the above-mentioned in-phase proportional amplifier circuit includes an input interface P3 and a two-stage amplifier circuit, and each stage of the amplifier circuit includes a filter capacitor and an operational amplifier; the electrical signal collected by the signal acquisition device is input by the input interface P3 and is processed by the two-stage amplifier circuit in turn.

[0066] Specifically, the first-stage amplifier circuit includes a filter capacitor C11, an operational amplifier U5, a grounding resistor R3, an input resistor R4, a feedback resistor R5, and a balancing resistor R6. In addition, it also includes a 9V power supply for the operational amplifier U5, a coil L5, a capacitor C12, and a capacitor C13 arranged on the power supply line. Among them: the grounding resistor R3 is connected between the input signal path and the ground, so that the input signal is referenced to GND. The input resistor R4 is connected to the non-inverting input terminal (+) of the operational amplifier U5 to set the input impedance; the feedback resistor R5 is connected between the output terminal and the inverting input terminal (-) of the operational amplifier U5 to help stabilize the gain and reduce distortion; the balancing resistor R6 is connected between the inverting input terminal (-) and GND to balance the bias current of the two input terminals of the operational amplifier U5 and reduce the error caused by the input bias current. The amplification factor A1 of the first-stage amplifier circuit is 1+R6 / R5.

[0067] The second-stage amplifier circuit includes a filter capacitor C14, an operational amplifier U9, a grounding resistor R7, an input resistor R8, a feedback resistor R10, and a balancing resistor R9. In addition, it also includes a 9V power supply for the operational amplifier U5, a coil L6, a capacitor C15, and a capacitor C16 arranged on the power supply line. Since the structures of the two-stage amplifier circuits are exactly the same, the functions of the resistors and capacitors at the corresponding positions in the two-stage amplifier circuits are also the same, which will not be described in detail here. The amplification factor of the second-stage amplifier circuit is A2=1+R10 / R9. The overall amplification factor of the in-phase proportional amplifier circuit is the product of A1 and A2.

[0068] Since the probability of ensuring perfect consistency between the physical components produced by the manufacturer and the idealized devices is low, after the filter capacitor C11 filters out the DC component in the input electrical signal for the first time, there is still a weak DC component that enters the operational amplifier U5 to be amplified, and then the amplified DC component can be filtered out by the filter capacitor C14. It can be seen that by setting up a common-phase proportional amplifier circuit composed of two-stage amplifier circuits, the amplified signal can be filtered step by step, and the actual error can be reduced as much as possible through filtering, and finally the DC component can be filtered out. In order to further filter out the DC component, a capacitor C17 and a resistor R11 for pulling the signal to GND as the reference are also set after the second-stage amplifier circuit.

[0069] like Figure 6 As shown, the envelope detection circuit includes a detection diode D3, a capacitor C18 and a resistor R12; the amplified signal processed by the in-phase proportional amplifier circuit is input into the anode of the detection diode D3, and the capacitor C18 and the resistor R12 are respectively connected between the cathode of the detection diode D3 and the ground.

[0070] The above-mentioned envelope detection circuit can perform amplitude detection operation and demodulate the low-frequency signal from the amplitude modulation signal. If there is no detection diode D3 and low-pass filtering is performed directly, the positive and negative envelope lines will be offset, so that the low-frequency signal cannot be detected. When the amplified signal is positive and exceeds the forward conduction voltage of the diode (0.24V), the diode is turned on, and the signal is charged to the capacitor C18 through the diode D3. At this time, the voltage at point TP6 increases with the increase of the charging voltage; when the amplified signal voltage begins to decrease and is less than the diode conduction voltage, the diode is reversely cut off, and the charging to the capacitor C18 is stopped at this time, and the discharge is carried out through the resistor R12. The voltage at point TP6 decreases with the discharge. Since the capacitor C18 is charged through the detection diode D3 while being discharged to the resistor R12, considering the difference in charging and discharging time, the parameters of the capacitor C18 and the resistor R12 are adjusted to obtain a suitable and gentle envelope detection signal.

[0071] like Figure 7As shown, the in-phase integration circuit includes an operational amplifier U4, a resistor R14, a resistor R16, a capacitor C20, a capacitor C23 and a resistor R17, and the resistance value of the resistor R14 is equal to that of the resistor R16, and both resistance values ​​are R, and the capacitance value of the capacitor C20 is equal to that of the capacitor C23, and both capacitance values ​​are C, so that the circuit balance can be ensured;

[0072] The signal processed by the envelope detection circuit is connected to the positive input terminal of the operational amplifier U4 through the resistor R14; the resistor R16 is connected between the negative input terminal of the operational amplifier U4 and GND;

[0073] The resistor R17 and the capacitor C23 are connected in parallel between the negative input terminal and the output terminal of the operational amplifier U4.

[0074] In addition, the in-phase integration circuit also includes a capacitor C19, a resistor R13 and a resistor R15. The capacitor C19 is located before the resistor R14 and is used to isolate the signal from direct current and pass alternating current. The resistor R13 is used to pull the signal to the reference GND, and the resistor R15 is connected between the positive input terminal of the operational amplifier U4 and GND. In addition, it also includes a +9V power supply for powering the operational amplifier U4, and a coil L7, a capacitor C21 and a capacitor C22 arranged on the power supply line.

[0075] like Figure 8 As shown in FIG. 1 , the in-phase integration circuit performs in-phase integration operation on the output voltage signal of the detection circuit to obtain the waveform area of ​​the voltage signal, that is, calculate the energy contained in the waveform, and finally obtain the integral signal waveform. The purpose of designing the in-phase integration circuit is to find out the energy difference between rapeseed grains and impurities. U i and U o The relationship between them is: , where t is the time.

[0076] At the same time, the resistance values ​​R of the two resistors R14 and R16 or the capacitance values ​​C of the two capacitors C20 and C23 are adjusted. The larger the resistance value R and the capacitance value C, the slower the integration speed will be, that is, the peak of the integral waveform will appear later, so that the effective area obtained by the integration can be increased. Since the peak of the integral waveform appears too late, it will overlap with the second integral waveform, which may cause the loss detection system to miss and misjudge the loss of grains. Therefore, the integral peak time is appropriately adjusted by continuously changing the R and C parameters. The designed in-phase integration circuit can quickly and continuously identify rapeseed grain signals, with a short integration time and high speed, and can adapt to continuous operation conditions.

[0077] Preferably, the voltage comparator circuit includes a voltage comparator, a potentiometer and a voltage stabilizing circuit; the potentiometer is connected to the inverting input terminal of the voltage comparator and is used to set different voltage thresholds to distinguish different voltage ranges after integration of grains and residual signals; the integrated voltage value output by the in-phase integration circuit is input to the in-phase input terminal of the voltage comparator; the output terminal of the voltage comparator is connected to the single-chip microcomputer via the voltage stabilizing circuit.

[0078] Taking the first voltage comparator circuit as an example, it includes a voltage comparator U3 and a potentiometer R19. The voltage comparator U3 is powered by a +9V power supply, and the power supply connection end of the voltage comparator U3 is grounded through a capacitor C25. The two fixed ends of the potentiometer R19 are connected to the +9V power supply and the ground respectively; the movable end of the potentiometer R19 is grounded through a resistor C24, and is connected to the inverting input end of the voltage comparator. With the above connection structure, the voltage threshold of the voltage comparator U3 is adjusted by adjusting the movable end of the potentiometer R19. The voltage stabilizing circuit includes a resistor R21 and a zener diode D2. The resistor R21 is connected between the output end of the voltage comparator U3 and the pin DC1 of the single-chip computer. The cathode of the zener diode D2 is connected to the pin DC1, and the other end is grounded; in addition, the pin DC1 is grounded through a capacitor C26. The output end of the voltage comparator U3 is also connected to a resistor R20 and a light-emitting diode OUT1. The light-emitting diode OUT1 serves as an indicator light. When the light-emitting diode OUT1 is illuminated, it indicates that there is a voltage input to the voltage comparator U3. Capacitors C24, C25 and C26 are used for filtering to prevent the circuit from interference. When the voltage at the non-inverting input terminal (pin 1) of the voltage comparator U3 is higher than the voltage at the inverting input terminal (pin 3), its output terminal (pin 4) outputs a voltage of 9V, otherwise there is no output at the output terminal. When the output terminal outputs a voltage, the voltage regulator diode D2 reduces the voltage to 3.3V and transmits it to pin DC1.

[0079] Similarly, the second voltage comparator circuit has a voltage comparator U7 and a potentiometer R22, the voltage stabilizing circuit includes a resistor R24 ​​and a voltage stabilizing diode D4, and the output end of the voltage comparator U7 is connected to the pin DC2 of the single-chip computer; in addition, it also includes a capacitor C27, a capacitor C28, and a capacitor C29 for filtering (corresponding to the capacitor C24, the capacitor C25, and the capacitor C26 of the first voltage comparator, respectively), and a resistor R23 and a light-emitting diode OUT2 for indicating whether there is a voltage input.

[0080] The third voltage comparator circuit has a voltage comparator U8 and a potentiometer R25, the voltage stabilizing circuit includes a resistor R27 and a voltage stabilizing diode D5, and the output end of the voltage comparator U8 is connected to the pin DC3 of the microcontroller; in addition, it also includes a capacitor C30, a capacitor C31, and a capacitor C32 for filtering (corresponding to the capacitor C24, the capacitor C25, and the capacitor C26 of the first voltage comparator, respectively), and a resistor R26 and a light-emitting diode OUT3 for indicating whether there is a voltage input.

[0081] The three voltage comparator circuits are connected in parallel and do not interfere with each other. The voltage thresholds of the voltage comparator U3, the voltage comparator U7 and the voltage comparator U8 decrease from front to back, and the threshold voltages of the three are set according to the impact energy of the short stem, the grain and the husk. Therefore, if the voltage value output after integration is greater than the voltage threshold of the voltage comparator U8 but less than the threshold voltage of the voltage comparator U7, the pulse signal characteristic received by the single-chip microcomputer has only one falling edge, and the signal is formed by the husk impacting the induction plate 6; if the voltage value output after integration is greater than the voltage threshold of the voltage comparator U7 but less than the threshold voltage of the voltage comparator U3, the pulse signal characteristic received by the single-chip microcomputer has two falling edges, and the signal is formed by the grain impacting the induction plate 6; if the voltage value output after integration is greater than the voltage threshold of the voltage comparator U3, the pulse signal characteristic received by the single-chip microcomputer has three falling edges, and the signal is formed by the short stem impacting the induction plate 6.

[0082] Figure 9(a) , 9(b) 9(c) are the results of the impact signals generated by the husk, seed and short stem on the induction plate 6 collected by the oscilloscope from the in-phase proportional amplifier circuit, envelope detection circuit and in-phase integration circuit of the circuit. Among them, the signal results displayed by the oscilloscope's four channels (CH1, CH2, CH3, CH4) are the original impact signal and the results obtained by the three module circuits of amplification, detection and integration in sequence, which are displayed in orange, blue, purple and green respectively.

[0083] Figure 10(a) , 10(b) 10(c) are the output signals of the in-phase integration circuits of the husk, grain and short stem respectively, which are the output results obtained by collecting the voltage comparison module part of the circuit through the oscilloscope. Among them, the signal results displayed by the three channels of the oscilloscope are: the output result of only displaying the yellow pulse signal indicates that the input signal of the voltage comparison module is the husk impact signal processed by the integration circuit, the output result of displaying the yellow and blue pulse signals indicates that the input signal of the voltage comparison module is the grain impact signal processed by the integration circuit, and the output result of displaying the yellow, blue and purple pulse signals indicates that the input signal of the voltage comparison module is the short stem impact signal processed by the integration circuit.

[0084] The lost grain identification method based on energy differentiation method and regression model is based on the lost grain identification system based on energy differentiation method and regression model and is implemented by the single chip microcomputer. The flow chart is as follows: Fig.11 As shown, specifically, the method includes the following steps S101-S105:

[0085] Step S101, respectively reading the signals output by the three voltage comparator circuits (the three circuits respectively include the voltage comparator U3, the voltage comparator U7 and the voltage comparator U8), and recording them as signal A, signal B and signal C, respectively, wherein the voltage threshold of the voltage comparator circuit corresponding to signal A is the largest, and the voltage threshold of the voltage comparator circuit corresponding to signal C is the smallest;

[0086] Step S102, performing rising edge detection and falling edge detection on signal C. When a rising edge is detected, the variable DC corresponding to the mark signal C is true, and the capture process of signal A and signal B is entered, and the captured signals corresponding to signal A and signal B are saved in variables DA and DB respectively;

[0087] Step S103, when the falling edge of signal C is detected, the states of variables DA and DB are determined:

[0088] If both variable DA and variable DB are false, the three pulse signals only generate one falling edge, indicating that the object is a silique shell, and the corresponding silique shell counter increases by 1;

[0089] If the variable DA is true and the variable DB is false, the three-way pulse signal generates two falling edges, indicating that the object is a grain, and the corresponding grain counter increases by 1;

[0090] If both variable DA and variable DB are true, the three pulse signals generate three falling edges, indicating that the object is a short stem, and the corresponding short stem counter increases by 1.

[0091] In the above judgment process, the variable is true, indicating that the falling edge is detected, and the variable is false, indicating that the falling edge is not detected. Based on the above method, the three-way signal can be effectively processed to effectively count the number of falling edges generated by the three-way signal at the same time, and accurately determine whether the object impacting the sensing plate 6 is the husk, the seed or the short stem, and the statistical result is accurate.

[0092] The method for constructing the fan speed-discrimination threshold regression model includes the following steps S201-S205:

[0093] Step S201, obtaining test data obtained through bench experiments, wherein the test data records multiple groups of impact tests on the induction plate using three types of materials, namely, silicone shells, seeds, and short stems, at different impact speeds, and the corresponding integral voltage values ​​output by the in-phase integration circuit;

[0094] Step S202, based on the test data, obtaining a scatter plot of the integrated voltage values ​​corresponding to the three types of materials;

[0095] Step S203, calculating the optimal discrimination thresholds corresponding to different impact velocities based on the scatter distribution diagram, wherein the optimal discrimination thresholds include a seed-shell discrimination threshold and a seed-short stalk discrimination threshold;

[0096] Step S204, obtaining a numerical correspondence between the fan speed and the optimal distinction threshold value according to the correspondence between the fan speed and the impact speed;

[0097] Step S205 , performing data fitting based on the numerical correspondence between the fan speed and the optimal distinction threshold value to obtain the fan speed-discrimination threshold value regression model.

[0098] Among the three voltage thresholds corresponding to the three voltage comparator circuits, the two voltage thresholds with larger values ​​(i.e., the voltage thresholds of the voltage comparator U3 and the voltage comparator U7) are determined based on the following steps S301-S302:

[0099] Step S301, determining a seed-shell distinction threshold and a seed-short stalk distinction threshold based on the current fan speed and the fan speed-discrimination threshold regression model;

[0100] Step S302, assigning the seed-silique shell distinction threshold to the voltage threshold corresponding to the voltage comparator U7, and assigning the seed-short stalk distinction threshold to the voltage threshold corresponding to the voltage comparator U3.

[0101] In the present invention, the bench test is based on Fig.12 The test device shown in the figure is used to transport the cleaned material to the induction plate through the conveyor belt 10. The material is laid flat on the conveyor belt. The material placement width is consistent with the induction plate width. The end of the conveyor belt and the center point of the induction plate are in a straight line in the vertical direction, so that the material can fall to the center point of the induction plate. The piezoelectric ceramic is fixed on the back of the center point of the induction plate. The fixed angle of the induction plate is 45 degrees to the horizontal direction, and the installation height is adjusted by adjusting the frame 11 to simulate different grain impact speeds.

[0102] In order to determine the threshold for distinguishing lost grains from debris, a bench test at different impact speeds was designed under the working conditions that meet the actual operation of the harvester. The impact speed is determined based on the speed at which the blown material hits the induction plate at different fan speeds of the rapeseed harvester cleaning system. Through calibration, it can be seen that when the fan speed range of the cleaning system is 700~1200r / min, the speed range of the material impacting the induction plate is 2.503-3.980m / s, and the fan speed is linearly positively correlated with the speed at which the material impacts the induction plate. The specific corresponding relationship is shown in Table 1 below.

[0103] Table 1 Comparison of the speed of the material colliding with the induction plate at different fan speeds

[0104] Fan speed of cleaning system The speed of the material at 0.4 meters below the discharge port 700 2.503 800 2.768 900 3.056 1000 3.433 1100 3.665 1200 3.980

[0105] The height of the free fall of the material in the bench test was determined based on the material impact speed of 2.5-4m / s (the height is the vertical height from the end of the conveyor belt to the center point of the induction plate below), and the height range is 32cm-81.5cm. Four sets of bench tests with different impact speeds were designed. Each set used 25 pod shells, 50 seeds and 25 short stems to impact the induction plate. The integrated voltage values ​​generated by different materials impacting the induction plate were recorded respectively, and scatter plots corresponding to the integrated voltage values ​​of the three types of materials were drawn to show the distribution law of the three types of signals. The test results are shown in Figure 2. Figure 13(a) to Figure 13(d) As shown, the four figures are the distribution diagrams of the integrated voltage values ​​of seeds and impurities at four impact speeds of 2.5m / s, 3m / s, 3.5m / s, and 4m / s. It can be found that the distribution of the integrated voltage values ​​of silique shells, seeds and short stems generally shows a trend that the silique shells are smaller than the seeds and smaller than the short stems. Therefore, the silique shell-seed distinction threshold C1 and the seed-short stem distinction threshold C2 are established, and the enumeration method is used to conduct a comprehensive search of the experimentally measured integrated voltage values. The optimal distinction thresholds under different impact speeds are obtained as shown in Table 2 below. As can be seen from Table 2, the distinction accuracy under different impact speeds is generally good, all above 90%, and the highest distinction accuracy is 93.9%. Figure 13(a) to Figure 13(d) In the figure, the orange straight line corresponds to the seed-shell distinction threshold C1, and the blue straight line corresponds to the seed-short stalk distinction threshold C2.

[0106] Table 2 Comparison table of optimal discrimination thresholds at different speeds

[0107] C1 C2 Accuracy 2.5m / s 0.840 2.690 0.904 3 m / s 1.010 2.970 0.910 3.5 m / s 1.020 3.280 0.919 4 m / s 1.100 3.490 0.939

[0108] According to the test results in Table 1 and Table 2, the relationship curves (regression models) between the fan speed and the seed-shell distinction threshold C1 and the seed-short stalk distinction threshold C2 were plotted, as shown in Figure 14(a) and Figure 14(b). By fitting the main data in Figure 14(a) and Figure 14(b), the calculation formulas for the optimal distinction threshold of the rapeseed harvester cleaning system at various fan speeds can be obtained, which are: y 1=0.00049 x +0.52760, y 2=0.00163 x +1.56280, where y 1 is the seed-shell distinction threshold, y 2 is the seed-short stem distinction threshold, x is the fan speed. The fan speed is linear with the seed-shell distinction threshold C1. The R 2 The values ​​are 0.9062 and 0.99439 respectively. Substituting the formula obtained from the fitting model into any fan speed value of the rapeseed harvester cleaning system can obtain the corresponding discrimination threshold.

[0109] In order to verify the accuracy of the loss detection system, the impact velocity was selected as 4 m / s, the threshold interval of seeds and impurities was (1.100, 3.490), and calibration tests with different seed numbers, different impurities and different mixtures were carried out. Each group of tests was repeated 3 times. The number of seeds was set to 50-200, and 4 different calibration mixtures were prepared according to the proportion of silicone, short stem and seeds in the mixture. The results are shown in Table 3.

[0110] Table 3 Material calibration test results

[0111] Calibration Materials Average Monitoring accuracy (%) Seeds (50 pieces) 49.0 98.0 Seeds (100 pieces) 97.8 97.8 Seeds (150 pieces) 141.8 94.5 Seeds (200 pieces) 183.2 91.6 Siliqua shell (10g) 5 - Short stems (10g) 0 - Seeds (100 pieces) + husks (10g) 96.8 96.8 Seeds (100 pieces) + short stems (10g) 95.2 95.2 Seeds (100) + husk (10g) + short stem (10g) 95.5 95.5 Seeds (100) + husk (5g) + short stem (5g) 96.2 96.2

[0112] As can be seen from Table 3, the loss detection system has a good ability to distinguish impurities overall, among which the silique shell and short stem have a certain impact on the accuracy of the loss detection system. For the case of pure grains, the detection accuracy of the loss detection system can reach 98%. When the number of grains is between 50 and 150, the impact strength of the grains is low, the interaction between the grains is also small, and the detection accuracy is high, reaching more than 95%. When the grain flow exceeds 150 grains and reaches 200 grains, the accuracy decreases. This is because a large number of grains collide with the sensor surface in a short period of time, causing two or more grains to hit the sensing plate at the same time, resulting in missed detection of grains, but the error is still 10%.

[0113] In the test with only impurity materials, it can be seen that when a single impact test was conducted on 10g of silique shell, a small amount of silique shell was misdetected as grains, while 10g of short stems were basically not misdetected as grains. Analysis of the biological characteristics of silique shells and short stems revealed that silique shells are generally light in weight, not much different from the weight of grains, while short stems are generally heavier in weight, much different from grains. For mixed materials with different proportions of grains and impurities, the detection accuracy can reach 95.2%-96.8%. And as the proportion of impurities in the mixed materials decreases, the recognition accuracy of the loss detection system continues to improve.

[0114] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A lost grain identification system based on energy differentiation method and regression model, which includes a signal acquisition device, a signal processing circuit and a lost grain detection and counting system; The signal acquisition device includes a sensing plate and a piezoelectric element; The signal processing circuit comprises an in-phase proportional amplifier circuit, an envelope detection circuit and an in-phase integration circuit which are arranged in sequence; Features: The signal processing circuit also includes three voltage comparator circuits connected in parallel with each other, the three voltage comparator circuits corresponding to the three voltage thresholds are different, and the voltage thresholds are determined based on the fan speed-discrimination threshold regression model; the integrated voltage value output by the in-phase integration circuit is used as the input signal of the three voltage comparator circuits; The lost grain detection and counting system includes a single-chip microcomputer; the single-chip microcomputer is used to read the signals output by the three voltage comparator circuits, determine the type of the object according to the number of falling edges of the three signals, and count each type of object; the types of objects include three categories: silicone shells, grains, and short stems, and the characteristics of the signals corresponding to the three types of objects are: 1 falling edge, 2 falling edges, and 3 falling edges, respectively.

2. The lost grain identification system based on energy differentiation method and regression model according to claim 1, characterized in that: The signal acquisition device also includes a support frame, on which a plurality of the sensing plates are mounted; the four corners of the sensing plates are respectively connected to the support frame by bolts, and a rubber shock absorber is provided between the sensing plates and the support frame; the sensing plates are arranged tilted relative to the ground.

3. The lost grain identification system based on energy differentiation method and regression model according to claim 1, characterized in that: The in-phase proportional amplifier circuit includes an input interface and a two-stage amplifier circuit, and each stage of the amplifier circuit includes a filter capacitor and an operational amplifier; the electrical signal collected by the signal acquisition device is input by the input interface and processed by the two-stage amplifier circuit in turn.

4. The lost grain identification system based on energy differentiation method and regression model according to claim 1, characterized in that: The envelope detection circuit includes a detection diode D3, a capacitor C18 and a resistor R12; the amplified signal processed by the in-phase proportional amplifier circuit is input to the anode of the detection diode D3, and the capacitor C18 and the resistor R12 are respectively connected between the cathode of the detection diode D3 and the ground.

5. The lost grain identification system based on energy differentiation method and regression model according to claim 1, characterized in that: The in-phase integration circuit includes an operational amplifier U4, a resistor R14, a resistor R16, a capacitor C20, a capacitor C23 and a resistor R17, and the resistance values ​​of the resistor R14 and the resistor R16 are equal, and the capacitance values ​​of the capacitor C20 and the capacitor C23 are equal; The signal processed by the envelope detection circuit is connected to the positive input terminal of the operational amplifier U4 through the resistor R14; the resistor R16 is connected between the negative input terminal of the operational amplifier U4 and GND; The resistor R17 and the capacitor C23 are connected in parallel between the negative input terminal and the output terminal of the operational amplifier U4.

6. The lost grain identification system based on energy differentiation method and regression model according to claim 1, characterized in that: The voltage comparator circuit includes a voltage comparator, a potentiometer and a voltage stabilizing circuit; the potentiometer is connected to the inverting input terminal of the voltage comparator and is used to set different voltage thresholds; the integrated voltage value output by the in-phase integration circuit is input to the in-phase input terminal of the voltage comparator; the output terminal of the voltage comparator is connected to the single-chip microcomputer via the voltage stabilizing circuit.

7. A method for identifying lost grains based on energy differentiation and regression model, which is based on the system for identifying lost grains based on energy differentiation and regression model described in any one of claims 1 to 6 and is implemented by the single chip microcomputer, characterized in that the method include: The signals output by the three voltage comparator circuits are read respectively and recorded as signal A, signal B, and signal C, respectively, wherein the voltage threshold of the voltage comparator circuit corresponding to signal A is the largest, and the voltage threshold of the voltage comparator circuit corresponding to signal C is the smallest; Perform rising edge detection and falling edge detection on signal C. When a rising edge is detected, the variable DC corresponding to signal C is marked as true, and the capture process of signal A and signal B is started. The captured signals corresponding to signal A and signal B are saved in variables DA and DB respectively. When the falling edge of signal C is detected, the states of variables DA and DB are determined: If both variables DA and DB are false, it means the object is a silique shell, and the corresponding silique shell counter increases by 1; If the variable DA is true and the variable DB is false, it means the object is a grain, and the corresponding grain counter increases by 1; If both variables DA and DB are true, it means that the object is a short stem, and the corresponding short stem counter increases by 1.

8. The method for identifying lost grains based on energy differentiation and regression model according to claim 7, characterized in that: The method for constructing the fan speed-discrimination threshold regression model includes: Acquire test data obtained through bench experiments, wherein the test data records the integrated voltage values ​​output by the corresponding in-phase integration circuits of multiple groups of impact tests on the induction plate at different impact speeds using three types of materials, namely, silicone shells, seeds, and short stems; Based on the test data, a scatter distribution diagram of the integrated voltage values ​​corresponding to the three types of materials is obtained; Calculating optimal discrimination thresholds corresponding to different impact velocities based on the scatter distribution diagram, wherein the optimal discrimination thresholds include a seed-shell discrimination threshold and a seed-short stem discrimination threshold; According to the correspondence between the fan speed and the impact speed, the numerical correspondence between the fan speed and the optimal distinction threshold is obtained; Data fitting is performed based on the numerical correspondence between the fan speed and the optimal distinction threshold value to obtain the fan speed-discrimination threshold value regression model.

9. The method for identifying lost grains based on energy differentiation and regression model according to claim 8, characterized in that: Among the three voltage thresholds corresponding to the three voltage comparator circuits, the two voltage thresholds with larger values ​​are determined based on the following steps: Determine a seed-silique shell distinction threshold and a seed-short stem distinction threshold based on the current fan speed and the fan speed-discrimination threshold regression model; The seed-shell distinction threshold and the seed-short stem distinction threshold are assigned to two voltage thresholds respectively.

Images