A self-adaptive filtering based combined harvester cleaning loss rate detection system and method
By combining differential piezoelectric sensors and LMS adaptive filtering algorithms, the problems of low efficiency and insufficient accuracy in the detection of cleaning loss rate in combine harvesters are solved, and real-time and accurate online detection is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANJING AGRI MECHANIZATION INST MIN OF AGRI
- Filing Date
- 2026-04-13
- Publication Date
- 2026-06-12
AI Technical Summary
The current method of detecting the cleaning loss rate of combine harvesters relies on manual inspection, which is inefficient and labor-intensive. Furthermore, the fixed filtering parameters of existing sensors cannot adapt to complex field environments, resulting in low detection accuracy, susceptibility to interference, and inability to achieve real-time online detection.
A differential piezoelectric sensor is used to acquire signals. Combined with a signal preprocessing circuit and a microcontroller, the LMS adaptive filtering algorithm is used for adaptive noise suppression and dynamic threshold pulse capture to achieve accurate detection of grain impact signals.
It significantly improved the signal-to-noise ratio of grain impact signals, enhanced the system's adaptability, improved detection accuracy and reliability, and enabled real-time online automated cleaning loss rate monitoring.
Smart Images
Figure CN122192813A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of combine harvester loss rate detection, and in particular to a combine harvester cleaning loss rate detection system and method based on adaptive filtering. Background Technology
[0002] The cleaning loss rate of a combine harvester is a key indicator of its operational performance. Accurate and real-time detection of the cleaning loss rate is crucial for improving harvesting efficiency and reducing grain loss. Currently, the detection of the cleaning loss rate in combine harvesters largely relies on manual methods. This method is not only inefficient but also labor-intensive and time-consuming, failing to meet the actual needs of online, real-time detection during harvester operations in the field.
[0003] To achieve automated detection, existing technologies have developed matching cleaning loss sensors. These sensors are typically installed at the rear of the harvester and collect vibration signals during operation to identify grain impact characteristics. The amount of grain detected by the sensor is used as the basis for judging the loss rate; the less grain detected, the lower the harvester's loss rate, and vice versa. However, the filtering stages of existing cleaning loss sensors generally use multi-circuit filtering or fixed digital filtering methods. The filtering parameters cannot be dynamically adjusted, making it difficult to adapt to the complex and changing field operating environment. During operation, they are easily affected by vibrations from the engine, transmission system, and other non-grain-related sources, resulting in a low signal-to-noise ratio of the grain impact vibration signal. This makes it impossible to accurately separate the characteristic frequency components related to the grain, ultimately affecting the accuracy of grain impact vibration characteristic identification. Consequently, the cleaning loss rate detection results are significantly inaccurate and fail to meet the requirements for precise detection.
[0004] In summary, there is an urgent need for a combine harvester cleaning loss rate detection system that can adapt to complex field operation environments, has strong anti-interference capabilities, accurate detection, and can achieve real-time online detection, in order to solve the aforementioned technical problems of existing detection methods and sensors. Summary of the Invention
[0005] The purpose of this application is to provide a system and method for detecting the cleaning loss rate of a combine harvester based on adaptive filtering. This system can solve the problems of low efficiency and high labor intensity in traditional manual detection of the cleaning loss rate of combine harvesters, as well as the technical defects of existing fixed sensor filtering methods that are easily affected by non-grain vibration and have low detection accuracy. This system enables real-time online accurate detection of the cleaning loss rate of combine harvesters.
[0006] To achieve the above objectives, this application provides the following solution: In a first aspect, this application provides a combine harvester cleaning loss rate detection system based on adaptive filtering, comprising: The differential piezoelectric sensor installed at the tail of the combine harvester is used to collect a main signal including grain impact signal and mechanical vibration noise signal, and a reference signal including mechanical vibration noise signal. A signal preprocessing circuit, connected to the differential piezoelectric sensor, is used to amplify, filter, and boost the main signal and the reference signal to obtain a preprocessed main signal and a preprocessed reference signal. A microcontroller is connected to the signal preprocessing circuit to synchronously acquire the preprocessed main signal and the preprocessed reference signal, and sequentially perform bandpass filtering, adaptive noise suppression based on the LMS adaptive filtering algorithm, seed impact signal pulse capture under dynamic threshold, and cleaning loss rate calculation.
[0007] Secondly, this application provides a method for detecting the cleaning loss rate of a combine harvester based on adaptive filtering. The method is applied to the aforementioned combine harvester cleaning loss rate detection system based on adaptive filtering. The method includes: Acquire preprocessed main signal and reference signal; the main signal includes grain impact signal and mechanical vibration noise signal, and the reference signal includes mechanical vibration noise signal. Bandpass filtering is performed on the preprocessed main signal and reference signal to obtain the denoised main signal and denoised reference signal; The LMS adaptive filtering algorithm is used to dynamically estimate and suppress the mechanical vibration noise signal in the denoised main signal using the denoised reference signal, so as to obtain a pure grain impact signal. A dynamic threshold is set based on the idling noise, and the pure grain impact signal is pulse captured based on the dynamic threshold to count the number of valid grain impacts. The cleaning loss rate is calculated based on the effective number of grain impacts.
[0008] According to the specific embodiments provided in this application, this application has the following technical effects: (1) Improve the signal-to-noise ratio of the grain impact signal: This application uses a differential piezoelectric sensor to collect the main signal (including grain + vibration) and the reference signal (vibration only) respectively, and introduces the LMS adaptive filtering algorithm for noise suppression, which can identify and dynamically cancel mechanical vibration noise in real time, thereby significantly improving the signal-to-noise ratio of the grain impact signal; (2) Enhance the system’s adaptability to complex field environments: Since the LMS adaptive filtering algorithm can adjust the filter weights online according to the reference signal, the system can adapt to non-steady-state disturbances such as engine speed changes and transmission system disturbances, overcoming the defects of “fixed filtering parameters and inability to adapt to changes in the working environment” in the existing technology, and improving detection stability; (3) Improve the accuracy and reliability of cleaning loss rate detection: Through the multi-level processing chain of "bandpass filtering + adaptive noise suppression + dynamic threshold pulse capture", the real grain impact signal is effectively separated, avoiding false counting or missed counting caused by mechanical vibration noise interference, thereby significantly improving the accuracy of cleaning loss rate calculation and meeting the monitoring needs of precision agriculture for low loss operation. (4) Realize real-time online detection: The entire signal acquisition, processing and cleaning loss rate calculation are automatically completed by the microcontroller without manual intervention. This solves the problems of low efficiency, high labor intensity and inability to provide real-time feedback in manual detection, and realizes automated, continuous and online cleaning loss monitoring. Attached Figure Description
[0009] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0010] Figure 1 A schematic diagram of the functional modules of a combine harvester cleaning loss rate detection system based on adaptive filtering, provided in an embodiment of this application; Figure 2 This is a schematic diagram of the process for capturing grain impact signal pulses. Detailed Implementation
[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0012] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0013] In one exemplary embodiment, such as Figure 1 As shown, a combine harvester cleaning loss rate detection system based on adaptive filtering is provided, including a differential piezoelectric sensor installed at the tail of the combine harvester, a signal preprocessing circuit, and a microcontroller. Each component is described in detail below.
[0014] (1) Differential piezoelectric sensor.
[0015] Differential piezoelectric sensors are used to acquire a main signal, including grain impact signals and mechanical vibration noise signals, and a reference signal, including mechanical vibration noise signals. The differential piezoelectric sensor adopts a double-layer differential structure. The signal collected by the upper sensitive plate is used as the main signal, including the grain impact signal and the mechanical vibration noise signal. The signal collected by the lower sensitive plate is used as the reference signal, which mainly includes the mechanical vibration noise signal.
[0016] main signal : ; Reference signal : ; in, The seed impact signal is collected by the upper sensing plate at time n. Let n be the mechanical vibration noise signal collected by the upper sensitive plate at time n. The signal represents the mechanical vibration noise collected by the lower layer sensitive plate at time n.
[0017] (2) Signal preprocessing circuit.
[0018] A signal preprocessing circuit, connected to the differential piezoelectric sensor, is used to amplify, filter, and boost the main signal and the reference signal to obtain a preprocessed main signal and a preprocessed reference signal.
[0019] like Figure 1 As shown, the signal preprocessing circuit includes a charge amplifier circuit, an RC low-pass filter circuit, and a DC boost circuit connected in sequence.
[0020] 1) Charge amplifier circuit.
[0021] The charge amplifier circuit is connected to the differential piezoelectric sensor and is used to amplify the main signal and the reference signal.
[0022] The charge amplifier circuit is used to convert the weak charge signal output by the differential piezoelectric sensor into a voltage signal and amplify it initially.
[0023] 2) RC low-pass filter circuit.
[0024] The RC low-pass filter circuit is connected to the output of the charge amplifier circuit and is used to perform preliminary filtering on the amplified main signal and reference signal.
[0025] The RC filter circuit is mainly used to filter out electromagnetic interference in the signal, prevent high-frequency noise from aliasing during microcontroller acquisition, and ensure the purity of the input signal.
[0026] 3) DC lift circuit.
[0027] The current boosting circuit is connected to the output of the RC low-pass filter circuit and is used to boost the pre-filtered main signal and reference signal to the acquisition range of the microcontroller to obtain the pre-processed main signal and pre-processed reference signal.
[0028] The DC boost circuit is used to boost the AC voltage signal to within the acquisition range of the microcontroller's ADC, preventing the negative half-cycle of the signal from being truncated and ensuring that the complete waveform can be accurately captured by subsequent digital processing stages.
[0029] (3) Microcontroller.
[0030] A microcontroller is connected to the signal preprocessing circuit to synchronously acquire the preprocessed main signal and the preprocessed reference signal, and sequentially perform bandpass filtering, adaptive noise suppression based on the LMS adaptive filtering algorithm, seed impact signal pulse capture under dynamic threshold, and cleaning loss rate calculation.
[0031] The microcontroller includes a dual-channel ADC synchronous acquisition module, a pre-filter module, an LMS adaptive filter module, an effective grain impact count statistics module, and a loss rate calculation module.
[0032] 1) Dual-channel ADC synchronous acquisition module.
[0033] The dual-channel ADC synchronous acquisition module is connected to the signal preprocessing circuit and is used to synchronously acquire the preprocessed main signal and the preprocessed reference signal.
[0034] After the aforementioned signal preprocessing, the main signal and reference signal are fed into the microcontroller. The dual-channel ADC synchronous acquisition module is activated first to simultaneously acquire the preprocessed main signal and reference signal, ensuring time consistency between the two signals. The acquired continuous data stream is stored in a memory buffer, with a sampling rate set to 10kHz to meet the requirements for capturing transient signals.
[0035] 2) Pre-filter module.
[0036] The pre-filter module is used to perform bandpass filtering on the preprocessed main signal and the preprocessed reference signal to obtain the denoised main signal and the denoised reference signal.
[0037] Because the combine harvester generates a large amount of non-grain vibration interference during operation, a pre-filter is required before adaptive filtering to prevent such signals from overshadowing the grain signal. This filtering function is built into the processor and uses a fourth-order Butterworth bandpass filter with a passband frequency set from 1000Hz to 4000Hz. This effectively suppresses low-frequency mechanical noise signals and electromagnetic noise with frequencies higher than 10kHz, resulting in a denoised main signal and a denoised reference signal.
[0038] 3) LMS adaptive filtering module.
[0039] The LMS adaptive filtering module is used to dynamically estimate and suppress mechanical vibration noise signals in the denoised main signal using the denoised reference signal, so as to obtain a pure grain impact signal.
[0040] The pre-filter module inputs the processed data into the LMS adaptive filter module. The LMS adaptive filter module uses the Least Mean Squares (LMS) adaptive filtering algorithm to continuously adjust the filter coefficients based on the error between the reference signal and the main signal, so as to dynamically eliminate the interference of background noise and improve the recognizability of the grain impact signal.
[0041] The LMS adaptive filtering algorithm dynamically adjusts the filter weights based on the denoised reference signal, constructing in real time an estimate of the mechanical vibration noise signal similar to the noise components in the main signal. This estimate is then subtracted from the main signal to output a clean grain impact signal. The process is illustrated by the following formula: The formula for calculating the estimated value of mechanical vibration noise signal is as follows: in, .
[0042] Pure grain impact signal .
[0043] Based on the minimum mean square error criterion, the filter weights are updated using gradient descent to more accurately estimate the noise in the next time step. The update formula is as follows: in, For the updated weight vector, The step size factor determines the convergence speed and steady-state error of the algorithm.
[0044] By updating the filter weights, the filter continuously "learns" the transfer function relationship between the reference signal and the main signal. Whenever the machine vibration changes, it drives the filter weights to be updated and adjusted, achieving "adaptive" filtering.
[0045] 4) Effective kernel impact count statistics module.
[0046] The effective grain impact count module is used to set a dynamic threshold based on idling noise, and to perform pulse capture on the pure grain impact signal based on the dynamic threshold to count the effective grain impact count.
[0047] After filtering by the LMS adaptive filtering module, a clean grain impact signal with mechanical vibration noise removed is obtained. Since the vibration frequencies of different combine harvesters vary, the grain capture threshold is difficult to determine. Therefore, this application designs a noise self-checking function. After self-checking, a threshold floating above the background noise can be obtained. This threshold can prevent noise from triggering falsely, while also ensuring sensitivity to weak grain signals.
[0048] Background noise statistics: The noise self-check function requires idling before the harvester starts operating. At this time, it will calculate the error signal between the idling noise output by the LMS adaptive filtering algorithm and the main signal. And calculate the average amplitude of the absolute value of the error signal. .
[0049] Dynamic threshold setting: Set a safety factor. (Take a value of 1.5~2.0), and the average amplitude Multiply to obtain the dynamic threshold. .
[0050] The pure grain impact signal is compared with a dynamic threshold, and the specific rules are as follows: Figure 2 As shown: Step 0: Initialize counter N=0 and timer T=0, then execute step 1. Step 1: Read the pure seed impact signal, convert the signal into an absolute value and compare it with a dynamic threshold. If the absolute value is greater than the dynamic threshold, proceed to Step 2; otherwise, return to Step 1.
[0051] Step 2: Read the maximum value of the pure grain impact signal to ensure that the highest point of impact energy is captured, thereby improving the accuracy of feature extraction.
[0052] Step 3: To prevent false detections caused by grain rebound on the differential piezoelectric sensor, a short silence period (2ms) is set after a grain is detected.
[0053] Step 4: Design a second-level timer to continuously accumulate the number of valid pulses generated within a 1-second period. (i.e., the number of effective kernel impacts), after the 1-second timer ends, this cumulative value is recorded. Store the data in the register, and simultaneously clear the counter to begin the next cycle of statistics.
[0054] 5) Loss rate calculation module.
[0055] The loss rate calculation module is used to calculate the cleaning loss rate based on the effective number of grain impacts.
[0056] Before calculating the loss rate, obtain the following job parameters: Average weight of a single grain (unit: g): 0.025 g for rice, 0.045 g for wheat, and 0.2 g for soybean.
[0057] Header width of a combine harvester (unit: m) Crop yield per unit area (unit: g / m²) 2 ).
[0058] The distance traveled by the combine harvester during the statistical period (unit: m) is calculated using the GPS latitude and longitude coordinates of the combine harvester.
[0059] Step 1: Calculate the loss quality within the current statistical period : .
[0060] Step 2: Calculate the estimated harvest quality for the current statistical period. : .
[0061] Step 3: Calculate the cleaning loss rate : .
[0062] like Figure 1 As shown, the calculated cleaning loss rate The data is transmitted to the vehicle-mounted display terminal via the CAN bus. If... A green indicator shows normal operation; if It displays red and triggers a buzzer alarm.
[0063] Based on the above-mentioned combine harvester cleaning loss rate detection system, this application embodiment also provides a combine harvester cleaning loss rate detection method based on adaptive filtering, including the following steps.
[0064] S1: Acquire the preprocessed main signal and reference signal; the main signal includes the grain impact signal and the mechanical vibration noise signal, and the reference signal includes the mechanical vibration noise signal.
[0065] S2: Perform bandpass filtering on the preprocessed main signal and reference signal to obtain the denoised main signal and denoised reference signal.
[0066] S3: The LMS adaptive filtering algorithm is used to dynamically estimate and suppress the mechanical vibration noise signal in the denoised main signal using the denoised reference signal, thereby obtaining a pure grain impact signal. Specifically, the LMS adaptive filtering algorithm is used to dynamically estimate the mechanical vibration noise signal in the denoised main signal using the denoised reference signal, thereby obtaining an estimated value of the mechanical vibration noise signal; the estimated value of the mechanical vibration noise signal is then removed from the denoised main signal to obtain a pure grain impact signal.
[0067] S4: Set a dynamic threshold based on idling noise, and perform pulse capture on the pure grain impact signal based on the dynamic threshold to count the number of valid grain impacts. Specifically, this includes: in the idling state where the combine harvester is started but not operating, collecting the error signal between the idling noise and the main signal output by the LMS adaptive filtering algorithm, and calculating the average amplitude of the absolute value of the error signal; setting a dynamic threshold based on the average amplitude.
[0068] S5: Calculate the cleaning loss rate based on the effective number of grain impacts.
[0069] This application utilizes a differential piezoelectric sensor combined with an LMS adaptive filtering mechanism to dynamically suppress non-grain vibration interference under complex field conditions, effectively extract high signal-to-noise ratio grain impact signals, and achieve high-precision, real-time online detection of cleaning loss rate. This solves the technical problems of weak anti-interference ability and large detection deviation of traditional fixed filtering methods, and significantly improves the intelligence and precision of combine harvester performance evaluation.
[0070] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0071] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A combine harvester cleaning loss rate detection system based on adaptive filtering, characterized in that, include: The differential piezoelectric sensor installed at the tail of the combine harvester is used to collect a main signal including grain impact signal and mechanical vibration noise signal, and a reference signal including mechanical vibration noise signal. A signal preprocessing circuit, connected to the differential piezoelectric sensor, is used to amplify, filter, and boost the main signal and the reference signal to obtain a preprocessed main signal and a preprocessed reference signal. A microcontroller is connected to the signal preprocessing circuit to synchronously acquire the preprocessed main signal and the preprocessed reference signal, and sequentially perform bandpass filtering, adaptive noise suppression based on the LMS adaptive filtering algorithm, seed impact signal pulse capture under dynamic threshold, and cleaning loss rate calculation.
2. The combine harvester cleaning loss rate detection system based on adaptive filtering according to claim 1, characterized in that, The differential piezoelectric sensor includes an upper sensitive plate and a lower sensitive plate. The upper sensitive plate is used to collect a main signal including a grain impact signal and a mechanical vibration noise signal, and the lower sensitive plate is used to collect a reference signal including a mechanical vibration noise signal.
3. The combine harvester cleaning loss rate detection system based on adaptive filtering according to claim 1, characterized in that, The signal preprocessing circuit includes a charge amplifier circuit, an RC low-pass filter circuit, and a DC boost circuit connected in sequence. A charge amplifier circuit, connected to the differential piezoelectric sensor, is used to amplify the main signal and the reference signal; An RC low-pass filter circuit is connected to the output of the charge amplifier circuit and is used to perform preliminary filtering on the amplified main signal and reference signal. A DC boost circuit, connected to the output of the RC low-pass filter circuit, is used to boost the pre-filtered main signal and reference signal to the acquisition range of the microcontroller, so as to obtain the pre-processed main signal and pre-processed reference signal.
4. The combine harvester cleaning loss rate detection system based on adaptive filtering according to claim 1, characterized in that, The microcontroller includes: A dual-channel ADC synchronous acquisition module is connected to the signal preprocessing circuit and is used to synchronously acquire the preprocessed main signal and the preprocessed reference signal; The pre-filter module is used to perform bandpass filtering on the preprocessed main signal and the preprocessed reference signal to obtain the denoised main signal and the denoised reference signal. The LMS adaptive filtering module is used to dynamically estimate and suppress mechanical vibration noise signals in the denoised main signal using the denoised reference signal to obtain a pure grain impact signal. The effective grain impact count module is used to set a dynamic threshold based on idling noise, and to perform pulse capture on the pure grain impact signal based on the dynamic threshold to count the effective grain impact count. The loss rate calculation module is used to calculate the cleaning loss rate based on the effective number of grain impacts.
5. The combine harvester cleaning loss rate detection system based on adaptive filtering according to claim 4, characterized in that, The dynamic threshold is: ;in, For dynamic thresholds, As a preset safety factor, The average amplitude of the absolute value of the idling noise signal.
6. The combine harvester cleaning loss rate detection system based on adaptive filtering according to claim 4, characterized in that, The formula for calculating the cleaning loss rate is as follows: in, To determine the cleaning loss rate, The effective number of kernel impacts. The average weight of a single kernel. The travel distance of the combine harvester For the header width of the combine harvester, This refers to crop yield per unit area.
7. A method for detecting the cleaning loss rate of a combine harvester based on adaptive filtering, characterized in that, The method is applied to the combine harvester cleaning loss rate detection system based on adaptive filtering as described in any one of claims 1-6, and the method includes: Acquire preprocessed main signal and reference signal; the main signal includes grain impact signal and mechanical vibration noise signal, and the reference signal includes mechanical vibration noise signal. Bandpass filtering is performed on the preprocessed main signal and reference signal to obtain the denoised main signal and denoised reference signal; The LMS adaptive filtering algorithm is used to dynamically estimate and suppress the mechanical vibration noise signal in the denoised main signal using the denoised reference signal, so as to obtain a pure grain impact signal. A dynamic threshold is set based on the idling noise, and the pure grain impact signal is pulse captured based on the dynamic threshold to count the number of valid grain impacts. The cleaning loss rate is calculated based on the effective number of grain impacts.
8. The method for detecting the cleaning loss rate of a combine harvester based on adaptive filtering according to claim 7, characterized in that, The LMS adaptive filtering algorithm is used to dynamically estimate and suppress mechanical vibration noise in the denoised main signal using the denoised reference signal, resulting in a pure grain impact signal. Specifically, this includes: The LMS adaptive filtering algorithm is used to dynamically estimate the mechanical vibration noise signal in the denoised main signal using the denoised reference signal, and the estimated value of the mechanical vibration noise signal is obtained. The mechanical vibration noise signal estimate is removed from the denoised main signal to obtain a pure grain impact signal.
9. The method for detecting the cleaning loss rate of a combine harvester based on adaptive filtering according to claim 8, characterized in that, The formula for calculating the estimated value of the mechanical vibration noise signal is as follows: in, Let n be the estimated value of the mechanical vibration noise signal at time n. This represents the weight vector of the filter in the LMS adaptive filtering algorithm. Let n be the denoised reference signal. Let be the filter weight coefficients at time n. Let M be the reference signal at time n-1, and M be the order of the weight vector.
10. The method for detecting the cleaning loss rate of a combine harvester based on adaptive filtering according to claim 7, characterized in that, Setting a dynamic threshold based on idling noise specifically includes: When the combine harvester is running idle but not in operation, the error signal between the idle noise and the main signal output by the LMS adaptive filtering algorithm is collected, and the average amplitude of the absolute value of the error signal is calculated. A dynamic threshold is set based on the average amplitude.