A method for detecting the cleaning loss rate of an air suction harvester

Abstract

The present invention relates to the field of loss rate detection technology, and specifically to a method for detecting the cleaning loss rate of an air-suction harvester. The method comprises the following steps: intermittently sampling the material discharged from the tail of the cleaning screen through an air suction system, and separating the collected material. Photoelectric sensors and weight sensors are used to detect the amount of retained grains in real time to obtain grain amount data, and the sensors are calibrated regularly to ensure the detection accuracy. By processing the detection data, the cleaning loss rate per unit time is calculated using the loss rate calculation formula. The present invention not only significantly improves the accuracy of sensor calibration, ensures the comprehensive exploration of various parameter combinations, and avoids errors and instabilities caused by manual operations, but also can maintain high-precision detection capabilities under various working conditions. The collected lost grains are transmitted to the granary through an auger, so that the grains are returned to the granary, thereby improving resource utilization efficiency.

Description

Technical Field

[0001] The invention relates to the technical field of loss rate detection, and in particular to a method for detecting the cleaning loss rate of an air suction harvester. Background Art

[0002] The cleaning system plays a key role in agricultural production. It is mainly used to separate and collect crop grains to ensure the efficiency and accuracy of the harvesting process. To achieve this goal, the cleaning system is usually equipped with photoelectric sensors and weight sensors to detect and monitor the number and weight of grains. Accurate calibration of sensors is crucial to ensure the accuracy of data, thereby improving the efficiency and effectiveness of the entire cleaning process.

[0003] However, the inventors have discovered that this technical solution still has at least the following defects:

[0004] Sensor calibration relies on manual operation: The calibration of photoelectric sensors and weight sensors in traditional cleaning systems usually relies on manual operation. This method is susceptible to human errors, resulting in low calibration accuracy and affecting the detection accuracy of the system.

[0005] Unstable calibration process: Due to the lack of consistency in manual calibration, the sensor calibration process is often unstable. The results of each calibration may be different, which makes it difficult to guarantee the overall performance of the system and affects the continuity and reliability of the cleaning process.

[0006] Lack of automation and real-time performance: Existing technologies lack an automated calibration mechanism and cannot achieve real-time monitoring and dynamic adjustment. This results in the inability to correct sensor deviations in a timely manner during the cleaning process, increasing the risk of system operation and maintenance costs. Summary of the invention

[0007] Based on the above objectives, the present invention provides a method for detecting the cleaning loss rate of an air-suction harvester.

[0008] A method for detecting the cleaning loss rate of an air suction harvester comprises the following steps:

[0009] S1, collection of cleaning materials: intermittent sampling of the materials discharged from the tail of the cleaning screen is carried out through the air suction system, and all the cleaning materials within the sampling time are collected;

[0010] S2, dynamic sampling optimization: using dynamic sampling algorithm, dynamically adjust sampling time and frequency according to real-time data analysis during the cleaning process;

[0011] S3, material separation: using a collecting device to separate the collected materials, specifically including grains, straws and weeds, retaining the grains and discharging the remaining materials;

[0012] S4, grain quantity detection: using photoelectric sensors and weight sensors to detect the amount of retained grains in real time to obtain grain quantity data;

[0013] S5, automatic calibration: equipped with automatic calibration device to calibrate the photoelectric sensor and weight sensor regularly;

[0014] S6, data processing: processing the detected grain quantity data through a data processing unit, specifically including data cleaning, filtering and formatting;

[0015] S7, loss rate calculation: Based on the processed grain quantity data, the loss rate calculation formula is used to calculate the cleaning loss rate per unit time.

[0016] Optionally, S1 includes:

[0017] S11, start the air suction system: activate the air suction system installed at the tail of the cleaning screen, check the various parameters of the air suction system, and adjust the suction force to adapt to the current cleaning environment;

[0018] S12, setting sampling parameters: according to the current cleaning working conditions, setting the sampling time and sampling parameters, including sampling frequency and sampling interval;

[0019] S13, intermittent sampling: within the set time, the air suction system is used to intermittently sample the material discharged from the tail of the cleaning screen;

[0020] S14, material collection and storage: the air suction system sucks all the materials discharged from the tail of the cleaning screen within the sampling time and collects them into the material collection container. The material collection container is equipped with a sensor for monitoring the total amount of collected materials and sends a prompt signal when the set capacity is reached;

[0021] S15, material pretreatment: the collected materials are transferred to the material processing unit for basic processing, including material screening and material drying;

[0022] S16, close the air suction system: after the sampling time is over, the air suction system is automatically closed, and the sampling data of this sampling is recorded, including the sampling time and the total amount of collected materials.

[0023] Optionally, S2 includes:

[0024] S21, real-time data collection: sensors installed in the cleaning system monitor and collect various parameter data in the cleaning process in real time, including material flow, equipment vibration frequency and air flow speed, and detect operation parameters, including travel speed and operation cutting width;

[0025] S22, data transmission and storage: transmitting the real-time collected data to the central data processing unit via a wireless or wired network, and storing and preliminarily organizing the data;

[0026] S23, dynamic sampling algorithm analysis: the central data processing unit analyzes the real-time data, uses the dynamic sampling algorithm to evaluate the current cleaning conditions, calculates and recommends the optimized sampling time and sampling frequency;

[0027] S24, parameter adjustment: according to the analysis results of the dynamic sampling algorithm, the sampling time and frequency of the air suction system are automatically adjusted to adapt to the current cleaning conditions;

[0028] S25, Feedback and Verification: Record and feedback the results of each sampling, and adjust and optimize the algorithm by comparing the actual sampling data with the expected results;

[0029] S26, repeated optimization: after each sampling cycle, the dynamic sampling optimization module continuously executes the above steps according to the latest cleaning conditions and real-time data to dynamically optimize the sampling parameters.

[0030] Optionally, the S23 includes:

[0031] S231, data preprocessing: preprocessing the data collected in real time;

[0032] S232, Multilayer Perceptron Model Training: Use the preprocessed historical data to train a multilayer perceptron (MLP) model;

[0033] S233, Model Verification and Optimization: Verify model performance and optimize the model, including:

[0034] Verify model performance: Use cross-validation method to verify model performance;

[0035] Optimize the model: Optimize the model by adjusting hyperparameters such as learning rate, number of hidden layers, and number of neurons;

[0036] S234, real-time data input and prediction: the pre-processed real-time cleaned data is input into the trained multi-layer perceptron model, and the model makes predictions based on the input data, calculates and recommends the optimized sampling time and frequency y.

[0037] Optionally, S3 includes:

[0038] S31, material collection and transportation: the materials collected by the cleaning material collection module are transported to the material separation device through the conveying device;

[0039] S32, preliminary screening: preliminary screening of the materials transmitted to the material separation device by a vibrating screen, wherein the vibrating screen uses screens with different apertures to separate the residual materials from the grains, and the residual materials include straw and weeds;

[0040] S33, airflow separation: the preliminarily screened materials are subjected to airflow separation. The airflow separator adjusts the airflow speed to blow weeds and straw out of the separation device and separate the grains by sedimentation;

[0041] S34, gravity separation: the materials are further separated by a gravity separation device, and the grains are separated from the rest of the materials by utilizing the difference in the density of the materials;

[0042] S35, miscellaneous material discharge: the separated miscellaneous materials are discharged from the separation device through a discharge channel, the miscellaneous materials including remaining straw, small pieces of weeds and other impurities;

[0043] S36 collects the separated lost grains and transfers them to the granary through the auger.

[0044] Optionally, S4 includes:

[0045] S41, photoelectric sensor arrangement: photoelectric sensors are installed on both sides of the transparent detection channel through which the material flows, to detect the number and size of the grains passing through the channel;

[0046] S42, weight sensor arrangement: a high-precision weight sensor is installed at the end of the detection channel to detect the total weight of the grains after passing through the photoelectric sensor in real time;

[0047] S43, data collection: the photoelectric sensor collects the light intensity change data when the grains pass through the detection channel in real time, and converts the data into digital signals, records the number of grains passing through each time interval, collects the grain weight data in real time, and records the total weight in each time interval;

[0048] S44, data processing and correction: denoising and smoothing the data collected by the photoelectric sensor to eliminate the influence of ambient light and other interference factors;

[0049] S45, data fusion and analysis: the data from the photoelectric sensor and the weight sensor are fused to obtain the number and weight of grains per unit time.

[0050] Optionally, the S5 includes:

[0051] S51, initializing calibration parameters: initializing calibration parameter set A, including multiple sets of calibration parameters ;

[0052] S52, select calibration parameters: select the current calibration parameters from the calibration parameter set A using the multi-armed bandit algorithm , expressed as:

[0053] ;

[0054] in, is a parameter The estimated reward value of is a parameter The number of times selected, t is the current calibration round;

[0055] S53, perform calibration: use the selected calibration parameters Calibrate photoelectric sensors and weight sensors;

[0056] S54, collect calibration results: after the calibration is completed, collect the output data of the sensor, calculate the detection accuracy after calibration, and convert the calibration results into reward signals , update the multi-armed bandit algorithm;

[0057] S55, update calibration parameter estimation: update the estimated reward value of the selected calibration parameter based on the collected calibration results ;

[0058] S56, adjust calibration strategy: by balancing exploration and utilization, continuously adjust the calibration strategy and select a new calibration parameter combination for calibration;

[0059] S57, Feedback and Optimization: Record the results and adjustment strategies of each round of calibration into the central data processing unit, and continuously optimize the multi-armed bandit algorithm by comparing the actual calibration results with the expected results;

[0060] S58, regular calibration: the system periodically triggers the above calibration process according to the set time interval or sensor status change.

[0061] Optionally, the S53 includes:

[0062] S531, initializing calibration parameters: obtaining calibration parameters for the current calibration round from the selected calibration parameter set ;

[0063] S532, sensor initial state record: record the initial state data of the photoelectric sensor and the weight sensor before applying the calibration parameters, including the light intensity baseline value and the initial weight baseline value;

[0064] S533, apply photoelectric sensor calibration parameters: according to the calibration parameters , Photoelectric adjusts the settings of the photoelectric sensor;

[0065] S534, applying weight sensor calibration parameters, specifically including:

[0066] Adjust weight bias: Set a new weight bias ,in is the initial weight baseline value;

[0067] Calibrate the weighing platform: Check and adjust the weighing platform of the weight sensor;

[0068] S535, calibration parameter application confirmation: test the photoelectric sensor and weight sensor.

[0069] Optionally, S6 includes:

[0070] S61, data collection: obtaining grain quantity data from the photoelectric sensor and the weight sensor in real time, and transmitting the data to the data processing unit;

[0071] S62, data cleaning: cleaning the original data to remove outliers and noise data;

[0072] S63, data filtering: use filtering technology to smooth the data and remove high-frequency noise;

[0073] S64, data formatting: converting the processed data into a standard format, including:

[0074] Timestamp addition: add timestamp to each data record;

[0075] Unit conversion: unify the data units;

[0076] S65, data storage: storing the cleaned, filtered and formatted data in a database of a central data processing unit, the database using distributed storage technology;

[0077] S66, Data Verification and Calibration: Verify the processed data.

[0078] Optionally, the S7 includes:

[0079] S71, data preparation: obtain the processed grain quantity data from the data processing unit, including the weight of the lost grains detected per unit time and the data required for the actual weight of grain harvested per unit time;

[0080] S72, unit time data division: the processed grain quantity data is divided according to unit time, and the time window size is T;

[0081] S73, calculating the weight of lost grains per unit time: accumulating the weight of lost grains detected by the cleaning system per unit time .

[0082] S74, calculating the actual weight of grain harvested per unit time: calculating the actual weight of grain harvested per unit time , expressed as:

[0083] ;

[0084] in, is the actual cutting width, is the walking distance per unit time, is the yield per unit area;

[0085] S75, loss rate calculation: based on the processed grain quantity data, calculate the cleaning loss rate per unit time , expressed as:

[0086] ;

[0087] in, is the weight of grain loss detected per unit time, is the actual weight of grain harvested per unit time, is the cleaning loss rate.

[0088] Beneficial effects of the present invention:

[0089] The present invention improves calibration accuracy and stability: The present invention significantly improves the calibration accuracy of photoelectric sensors and weight sensors by introducing an automatic calibration device and an optimization algorithm. The multi-armed bandit algorithm is used for parameter selection and optimization to ensure that every parameter combination in the calibration process is fully explored and utilized, avoiding errors and instability caused by manual operation. In this way, the system can maintain high-precision detection capabilities under various working conditions, thereby improving the reliability of the cleaning process.

[0090] The present invention realizes real-time monitoring and dynamic adjustment: The automatic calibration system of the present invention has real-time monitoring and dynamic adjustment functions, and can automatically adjust the calibration parameters according to the real-time detection data of the sensor. Through regular calibration and feedback optimization, the system can continuously adapt to environmental changes and equipment wear and tear, and maintain the best working state. This innovation not only reduces the need for manual intervention, but also reduces maintenance costs and improves the overall efficiency of the system.

[0091] The present invention enhances the adaptability and reliability of the system: through the data fusion and analysis module, the present invention can comprehensively process the data of the photoelectric sensor and the weight sensor to obtain the number and weight of grains per unit time. Combined with the application of the automatic calibration system, the stability and consistency of the sensor in long-term operation are ensured. The high adaptability and reliability of the system enable it to operate efficiently under different cleaning conditions, significantly improving the working effect and service life of the cleaning system. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0093] Figure 1 It is a schematic diagram of a method for detecting the cleaning loss rate of an air suction harvester according to an embodiment of the present invention;

[0094] Figure 2 A schematic diagram of dynamic sampling optimization according to an embodiment of the present invention. DETAILED DESCRIPTION

[0095] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0096] It should be noted that the references to "one embodiment", "embodiment", "exemplary embodiments", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0097] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0098] like Figure 1-Figure 2 As shown, a method for detecting the cleaning loss rate of an air suction harvester comprises the following steps:

[0099] S1, collection of cleaning materials: intermittent sampling of materials discharged from the tail of the cleaning screen is carried out through the air suction system. The sampling time is 10 seconds, and all the cleaning materials within the sampling time are collected;

[0100] S2, dynamic sampling optimization: adopt dynamic sampling algorithm, dynamically adjust sampling time and frequency according to real-time data analysis during the cleaning process, optimize sampling effect, and ensure that representative samples can be obtained under different cleaning conditions;

[0101] S3, material separation: using a collecting device to separate the collected materials, specifically including grains, straws and weeds, retaining the grains and discharging the remaining materials;

[0102] S4, grain quantity detection: using photoelectric sensors and weight sensors to detect the amount of retained grains in real time to obtain grain quantity data;

[0103] S5, automatic calibration: equipped with automatic calibration device, regular calibration of photoelectric sensors and weight sensors to ensure detection accuracy. The system can automatically adjust the calibration parameters according to the detection results, reduce human errors, and improve the overall stability and accuracy of the system;

[0104] S6, data processing: processing the detected grain quantity data through a data processing unit, specifically including data cleaning, filtering and formatting, to ensure the accuracy and reliability of the data;

[0105] S7, loss rate calculation: Based on the processed grain quantity data, the loss rate calculation formula is used to calculate the cleaning loss rate per unit time.

[0106] The detection accuracy and stability of the cleaning system are improved through automatic calibration, dynamic sampling optimization, data fusion and processing. Real-time data analysis and adjustment ensure that representative samples are obtained under different cleaning conditions to reduce human errors. The system achieves efficient and accurate grain quantity detection and loss rate calculation, improving the reliability and efficiency of the cleaning process.

[0107] S1 includes:

[0108] S11, start the air suction system: activate the air suction system installed at the tail of the cleaning screen, check the various parameters of the air suction system, ensure that the system is in normal working condition, and adjust the suction force to adapt to the current cleaning environment;

[0109] S12, setting sampling parameters: according to the current cleaning conditions, the sampling time is set to 10 seconds, and the sampling parameters are set, including the sampling frequency and sampling interval, to ensure the representativeness and effectiveness of the sampling;

[0110] S13, intermittent sampling: within the set time, the air suction system is used to intermittently sample the material discharged from the tail of the cleaning screen to ensure that the material is evenly sucked into the air suction system in each sampling cycle;

[0111] S14, material collection and storage: the air suction system sucks all the materials discharged from the tail of the cleaning screen within the sampling time and collects them into the material collection container. The material collection container is equipped with a sensor for monitoring the total amount of collected materials and sends a prompt signal when the set capacity is reached to prevent overload;

[0112] S15, material pretreatment: The collected materials are transferred to the material handling unit for basic processing, including material screening and material drying, to ensure that the material state is suitable for subsequent material differentiation and detection steps.

[0113] S16, close the air suction system: after the sampling time is over, the air suction system is automatically closed, and the sampling data of this sampling is recorded, including the sampling time and the total amount of collected materials, for subsequent analysis and processing.

[0114] The accuracy and representativeness of sampling are improved through the systematic cleaning and material collection process. The efficient operation of each step is ensured by starting the air suction system, setting sampling parameters, intermittent sampling, material collection and storage, material pretreatment and system shutdown. Sensor monitoring and automated processes reduce human intervention and errors, improve the stability of the overall system and the reliability of data, and facilitate subsequent analysis and processing.

[0115] S2 includes:

[0116] S21, real-time data collection: sensors installed in the cleaning system monitor and collect various parameter data in the cleaning process in real time, including material flow, equipment vibration frequency and air flow speed, and detect operation parameters, including travel speed and operation cutting width;

[0117] S22, data transmission and storage: the real-time collected data is transmitted to the central data processing unit via wireless or wired network, and the data is stored and preliminarily sorted to ensure the integrity and availability of the data;

[0118] S23, dynamic sampling algorithm analysis: the central data processing unit analyzes the real-time data, uses the dynamic sampling algorithm to evaluate the current cleaning conditions, calculates and recommends the optimized sampling time and sampling frequency;

[0119] S24, parameter adjustment: according to the analysis results of the dynamic sampling algorithm, the sampling time and frequency of the air suction system are automatically adjusted to adapt to the current cleaning conditions and ensure the representativeness of the collected samples;

[0120] S25, Feedback and Verification: Record and feedback the results of each sampling, verify the effectiveness of the dynamic sampling algorithm by comparing the actual sampling data with the expected results, and adjust and optimize the algorithm;

[0121] S26, repeated optimization: after each sampling cycle, the dynamic sampling optimization module continuously executes the above steps according to the latest cleaning conditions and real-time data, and dynamically optimizes the sampling parameters to ensure that representative samples can be obtained under different cleaning conditions;

[0122] Through the dynamic sampling optimization process, real-time adjustment and optimization of the sampling of the cleaning system is achieved. Real-time data acquisition, data transmission and storage, dynamic sampling algorithm analysis, parameter adjustment, feedback and verification, and repeated optimization steps ensure the representativeness and accuracy of the sampling. Automated dynamic adjustment reduces human errors, improves the adaptability and efficiency of the system, ensures that high-quality samples can be obtained under different cleaning conditions, and provides a reliable basis for subsequent data processing and analysis.

[0123] S23 includes:

[0124] S231, data preprocessing: preprocessing the data collected in real time, specifically including:

[0125] Denoising: Apply filtering technology to remove noise and ensure the purity of data;

[0126] Standardization: Standardize the data to a uniform scale so that the model can better handle data of different dimensions;

[0127] Feature extraction: Extract key features, including mean, standard deviation, and trend, to reduce data dimensions and improve model processing efficiency;

[0128] S232, Multilayer Perceptron Model Training: Use the preprocessed historical data to train a multilayer perceptron (MLP) model, including:

[0129] Input layer: includes all preprocessed feature data x;

[0130] Hidden layer: Several hidden layers, each layer contains several neurons, and uses nonlinear activation function f to capture the nonlinear relationship of data;

[0131] Output layer: Generates optimized sampling time and frequency, expressed as:

[0132] ;

[0133] in, , is the weight matrix, , is the bias, f is the nonlinear activation function;

[0134] S233, Model Verification and Optimization: Verify model performance and optimize the model, including:

[0135] Verify model performance: Use cross-validation method to verify model performance, expressed as:

[0136] ;

[0137] Where k is the number of folds, is the mean square error of the i-th fold;

[0138] Optimize the model: Optimize the model by adjusting hyperparameters such as learning rate, number of hidden layers, and number of neurons to ensure good generalization and accuracy, expressed as:

[0139] ;

[0140] in, are model parameters, is the learning rate, is the loss function;

[0141] S234, real-time data input and prediction: The pre-processed real-time cleaned data is input into the trained multi-layer perceptron model. The model predicts based on the input data, calculates and recommends the optimized sampling time and frequency y, which is expressed as:

[0142] ;

[0143] Among them, x is the preprocessed input data, , is the weight matrix, , is the bias, and f is the nonlinear activation function.

[0144] Through data preprocessing and multi-layer perceptron model training, high efficiency and accuracy of dynamic sampling optimization are achieved. Data preprocessing includes denoising, standardization and feature extraction to ensure data purity and processing efficiency. Real-time data input and prediction realize dynamic adjustment of sampling time and frequency, improving the adaptability and sample representativeness of the cleaning system.

[0145] S3 includes:

[0146] S31, material collection and transportation: the materials collected by the cleaning material collection module are transported to the material separation device through the conveying device to ensure that the materials are evenly distributed at the entrance of the separation device;

[0147] S32, preliminary screening: preliminary screening of the material delivered to the material separation device by a vibrating screen, the vibrating screen using screens of different apertures to separate larger debris such as straw and large weeds from the grains;

[0148] S33, airflow separation: the materials after preliminary screening are subjected to airflow separation. The airflow separator adjusts the airflow speed to blow the lighter weeds and straw out of the separation device and separate the heavier grains by settling;

[0149] S34, gravity separation: the materials are further separated by a gravity separation device, and the density difference of the materials is used to separate the seeds with higher density from the impurities with lower density;

[0150] S35, miscellaneous material discharge: the separated miscellaneous materials are discharged from the separation device through a discharge channel, the miscellaneous materials including remaining straw, small pieces of weeds and other impurities;

[0151] S36 collects the separated lost grains and transfers them to the granary through the auger.

[0152] The separation efficiency of seeds and impurities is improved through the systematic material separation process. Including material collection and transportation, preliminary screening, air flow separation, gravity separation and residual material discharge, ensuring that the materials are fully processed in each step. The automated and refined processing flow improves the cleaning efficiency and seed purity, reduces manual intervention, and improves the stability and reliability of the system.

[0153] S4 includes:

[0154] S41, photoelectric sensor arrangement: photoelectric sensors are installed on both sides of the transparent detection channel through which the material flows, to detect the number and size of the grains passing through the channel;

[0155] S42, weight sensor arrangement: a high-precision weight sensor is installed at the end of the detection channel to detect the total weight of the grains after passing through the photoelectric sensor in real time;

[0156] S43, data collection: the photoelectric sensor collects the light intensity change data when the grains pass through the detection channel in real time, and converts the data into digital signals, records the number of grains passing through each time interval, collects the grain weight data in real time, and records the total weight in each time interval;

[0157] S44, data processing and correction: De-noise and smooth the data collected by the photoelectric sensor to eliminate the influence of ambient light and other interference factors, expressed as:

[0158] ;

[0159] in, is the number of grains at time t, is the light intensity change of the i-th seed, is the baseline light intensity;

[0160] The data collected by the weight sensor is denoised and calibrated to ensure the accuracy of the weight data, which is expressed as:

[0161] ;

[0162] in, is the net weight at time t, is the original measured weight, is the tare weight;

[0163] S45, data fusion and analysis: The data of the photoelectric sensor and the weight sensor are fused to obtain the number and weight of grains per unit time, expressed as:

[0164] ;

[0165] is the unit weight at time t, and the quantity and weight data are analyzed to determine the average weight and distribution of the grains.

[0166] Through the systematic arrangement and data processing of photoelectric sensors and weight sensors, high-precision grain quantity detection is achieved, which can determine the average weight and distribution of grains, provide accurate feedback on the cleaning process, and improve the overall performance and stability of the system.

[0167] S5 includes:

[0168] S51, initializing calibration parameters: initializing calibration parameter set A, including multiple sets of calibration parameters ;

[0169] S52, select calibration parameters: select the current calibration parameters from the calibration parameter set A using the multi-armed bandit algorithm , expressed as:

[0170] ;

[0171] in, is a parameter The estimated reward value of is a parameter The number of times selected, t is the current calibration round;

[0172] S53, perform calibration: use the selected calibration parameters Calibrate photoelectric sensors and weight sensors, ensuring that the calibration parameters are applied during sensor adjustment.

[0173] S54, collect calibration results: after the calibration is completed, collect the output data of the sensor, calculate the detection accuracy after calibration, and convert the calibration results into reward signals , update the multi-armed bandit algorithm;

[0174] S55, update calibration parameter estimation: update the estimated reward value of the selected calibration parameter based on the collected calibration results :

[0175] ;

[0176] in, is the kth selection parameter The reward value of

[0177] S56, adjust calibration strategy: by balancing exploration and utilization, continuously adjust the calibration strategy and select new calibration parameter combinations for calibration to gradually improve calibration accuracy and system stability;

[0178] S57, Feedback and Optimization: Record the results and adjustment strategies of each round of calibration into the central data processing unit, and continuously optimize the multi-armed bandit algorithm by comparing the actual calibration results with the expected results to ensure that the selection of calibration parameters gradually approaches the optimal value;

[0179] S58, regular calibration: the system regularly triggers the above calibration process according to the set time interval or sensor status change to ensure that the detection accuracy of the sensor remains in the best state for a long time;

[0180] The automatic calibration process improves the detection accuracy and system stability of the sensor. The multi-armed bandit algorithm is used to balance exploration and utilization, ensuring that the calibration parameters gradually approach the optimal value. Real-time optimization improves the adaptability and reliability of the system, ensuring the long-term stable operation of the sensor and providing accurate data support for the cleaning process.

[0181] S53 includes:

[0182] S531, initializing calibration parameters: obtaining calibration parameters for the current calibration round from the selected calibration parameter set .

[0183] S532, sensor initial state record: record the initial state data of the photoelectric sensor and the weight sensor before applying the calibration parameters, including the light intensity baseline value and the initial weight baseline value.

[0184] S533, apply photoelectric sensor calibration parameters: according to the calibration parameters , Photoelectric adjusts the settings of the photoelectric sensor, including:

[0185] Adjust light intensity value: set a new light intensity threshold ,in is the initial light intensity baseline value;

[0186] Calibrate the optical path alignment: adjust the optical path alignment of the photoelectric sensor to ensure the accuracy of the sensor receiving the signal;

[0187] S534, applying weight sensor calibration parameters, specifically including:

[0188] Adjust weight bias: Set a new weight bias ,in is the initial weight baseline value;

[0189] Calibrate the weighing platform: Check and adjust the weighing platform of the weight sensor to ensure the balance and stability of the weighing platform;

[0190] S535, calibration parameter application confirmation: Test the photoelectric sensor and weight sensor to confirm that the calibration parameters have been correctly applied and are effective, including:

[0191] Photoelectric sensor test: Use a standard light source to test and record the response data of the photoelectric sensor to ensure its detection accuracy under the new threshold;

[0192] Weight sensor test: Use a known standard weight to test and record the weight sensor response data to ensure its weighing accuracy under the new bias;

[0193] The detection accuracy and reliability of the sensor are improved through a systematic calibration process. This includes initializing calibration parameters, recording the initial state of the sensor, applying photoelectric and weight sensor calibration parameters, and confirming the application of calibration parameters. By adjusting the light intensity threshold and weight bias, calibrating the optical path alignment and weighing platform, the high-precision operation of the sensor is ensured. The application confirmation step ensures the validity and accuracy of the calibration parameters, reduces human errors, and improves the overall stability of the system and data reliability.

[0194] S6 includes:

[0195] S61, data collection: obtaining grain quantity data from the photoelectric sensor and the weight sensor in real time, and transmitting the data to the data processing unit;

[0196] S62, data cleaning: Clean the raw data to remove outliers and noise data, including:

[0197] Outlier detection and removal: by setting the threshold range , remove abnormal data points that are out of range;

[0198] Missing value filling: Use interpolation to fill in missing data, expressed as:

[0199] ;

[0200] S63, data filtering: use filtering technology to smooth the data and remove high-frequency noise. The specific steps include:

[0201] Moving average filtering: ;

[0202] in, is the filtered data, is the original data, N is the window size;

[0203] Kalman filter: ;

[0204] in, is the current estimate, is the Kalman gain, is the measurement value, H is the observation matrix;

[0205] S64, Data Formatting: Convert the processed data into a standard format for subsequent processing and analysis, including:

[0206] Timestamp addition: add timestamp to each data record;

[0207] Unit conversion: unify the data units;

[0208] S65, Data storage: The cleaned, filtered and formatted data is stored in the database of the central data processing unit to ensure data persistence and traceability. The database uses distributed storage technology to improve data access speed and reliability;

[0209] S66, Data Verification and Calibration: Verify the processed data to ensure its accuracy and consistency, including:

[0210] Cross-validation: Divide the data into training sets and test sets, and perform cross-validation to ensure that no new errors are introduced during the processing;

[0211] Consistency check: Compare the data changes before and after processing to ensure that no valid information is lost or tampered with during data processing;

[0212] The accuracy and reliability of grain quantity data are ensured through a systematic data processing process, including data collection, cleaning, filtering, formatting, storage, verification and calibration. Noise and abnormal data are removed through steps such as outlier detection, missing value filling, moving average filtering and Kalman filtering. Data formatting and distributed storage improve data access speed and traceability. Cross-validation and consistency checks ensure the accuracy and consistency of data processing, providing a high-quality data foundation for subsequent analysis.

[0213] S7 includes:

[0214] S71, data preparation: obtain the processed grain quantity data from the data processing unit, including the weight of the lost grains detected per unit time and the data required for the actual weight of grain harvested per unit time;

[0215] S72, unit time data division: the processed grain quantity data is divided according to unit time, and the time window size is T;

[0216] S73, calculating the weight of lost grains per unit time: accumulating the weight of lost grains detected by the cleaning system per unit time ;

[0217] S74, calculating the actual weight of grain harvested per unit time: calculating the actual weight of grain harvested per unit time , expressed as:

[0218] ;

[0219] in, is the actual cutting width, is the walking distance per unit time, is the yield per unit area;

[0220] S75, loss rate calculation: based on the processed grain quantity data, calculate the cleaning loss rate per unit time , expressed as:

[0221] ;

[0222] in, is the weight of grain loss detected per unit time, is the actual weight of grain harvested per unit time, is the cleaning loss rate.

[0223] Through the systematic cleaning loss rate calculation process, the accuracy and efficiency of data processing are improved. The automated calculation process reduces human errors and provides real-time and accurate loss rate data, which facilitates timely adjustment of cleaning parameters, optimizes harvesting effects, and improves overall operational efficiency.

[0224] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0225] 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 method for detecting the cleaning loss rate of an air suction harvester, characterized in that: The following steps are involved: S1, collection of cleaning materials: intermittent sampling of the materials discharged from the tail of the cleaning screen is carried out through the air suction system, and all the cleaning materials within the sampling time are collected; S2, dynamic sampling optimization: using dynamic sampling algorithm, dynamically adjust sampling time and frequency according to real-time data analysis during the cleaning process; S3, material separation: using the collecting device to separate the collected materials, specifically including grains, straw and weeds, retaining the grains, discharging the remaining materials, and collecting the lost grains, which are transferred to the granary through the auger; S4, grain quantity detection: using photoelectric sensors and weight sensors to detect the amount of retained grains in real time to obtain grain quantity data; S5, automatic calibration: equipped with automatic calibration device to calibrate the photoelectric sensor and weight sensor regularly, including: S51, initializing calibration parameters: initializing calibration parameter set A, including multiple sets of calibration parameters ; S52, select calibration parameters: select the current calibration parameters from the calibration parameter set A using the multi-armed bandit algorithm , expressed as: ； in, is a parameter The estimated reward value of is a parameter The number of times selected, t is the current calibration round; S53, perform calibration: use the selected calibration parameters Calibrate photoelectric sensors and weight sensors; S54, collect calibration results: after the calibration is completed, collect the output data of the sensor, calculate the detection accuracy after calibration, and convert the calibration results into reward signals , update the multi-armed bandit algorithm; S55, update calibration parameter estimation: update the estimated reward value of the selected calibration parameter based on the collected calibration results ; S56, adjust calibration strategy: by balancing exploration and utilization, continuously adjust the calibration strategy and select a new calibration parameter combination for calibration; S57, Feedback and Optimization: Record the results and adjustment strategies of each round of calibration into the central data processing unit, and continuously optimize the multi-armed bandit algorithm by comparing the actual calibration results with the expected results; S58, periodic calibration: the system triggers the calibration process periodically according to the set time interval or sensor status change; S6, data processing: processing the detected grain quantity data through a data processing unit, specifically including data cleaning, filtering and formatting; S7, loss rate calculation: Based on the processed grain quantity data, the loss rate calculation formula is used to calculate the cleaning loss rate per unit time.

2. A method for detecting the cleaning loss rate of an air suction harvester according to claim 1, characterized in that: The S1 includes: S11, start the air suction system: activate the air suction system installed at the tail of the cleaning screen, check the various parameters of the air suction system, and adjust the suction force to adapt to the current cleaning environment; S12, setting sampling parameters: according to the current cleaning working conditions, setting the sampling time and sampling parameters, including sampling frequency and sampling interval; S13, intermittent sampling: within the set time, the air suction system is used to intermittently sample the material discharged from the tail of the cleaning screen; S14, material collection and storage: the air suction system sucks all the materials discharged from the tail of the cleaning screen within the sampling time and collects them into the material collection container. The material collection container is equipped with a sensor for monitoring the total amount of collected materials and sends a prompt signal when the set capacity is reached; S15, material pretreatment: the collected materials are transferred to the material processing unit for basic processing, including material screening and material drying; S16, close the air suction system: after the sampling time is over, the air suction system is automatically closed, and the sampling data of this sampling is recorded, including the sampling time and the total amount of collected materials.

3. The method for detecting the cleaning loss rate of an air suction harvester according to claim 1, characterized in that: The S2 includes: S21, real-time data collection: sensors installed in the cleaning system monitor and collect various parameter data in the cleaning process in real time, including material flow, equipment vibration frequency and air flow speed, and detect operation parameters, including travel speed and operation cutting width; S22, data transmission and storage: transmitting the real-time collected data to the central data processing unit via a wireless or wired network, and storing and preliminarily organizing the data; S23, dynamic sampling algorithm analysis: the central data processing unit analyzes the real-time data, uses the dynamic sampling algorithm to evaluate the current cleaning conditions, calculates and recommends the optimized sampling time and sampling frequency; S24, parameter adjustment: according to the analysis results of the dynamic sampling algorithm, the sampling time and frequency of the air suction system are automatically adjusted to adapt to the current cleaning conditions; S25, Feedback and Verification: Record and feedback the results of each sampling, and adjust and optimize the algorithm by comparing the actual sampling data with the expected results; S26, repeated optimization: after each sampling cycle, the dynamic sampling optimization module continuously executes the above steps according to the latest cleaning conditions and real-time data to dynamically optimize the sampling parameters.

4. A method for detecting the cleaning loss rate of an air suction harvester according to claim 3, characterized in that: The S23 includes: S231, data preprocessing: preprocessing the data collected in real time; S232, Multilayer Perceptron Model Training: Use the preprocessed historical data to train the multilayer perceptron model; S233, Model Verification and Optimization: Verify model performance and optimize the model, including: Verify model performance: Use cross-validation method to verify model performance; Optimize the model: Optimize the model by adjusting hyperparameters; S234, real-time data input and prediction: the pre-processed real-time cleaned data is input into the trained multi-layer perceptron model, and the model makes predictions based on the input data, calculates and recommends the optimized sampling time and frequency y.

5. The method for detecting the cleaning loss rate of an air suction harvester according to claim 1, characterized in that: The S3 includes: S31, material collection and transportation: the materials collected by the cleaning material collection module are transported to the material separation device through the conveying device; S32, preliminary screening: preliminary screening of the material delivered to the material separation device by a vibrating screen, the vibrating screen using screens of different apertures to separate larger debris such as straw and large weeds from the grains; S33, airflow separation: the materials after preliminary screening are subjected to airflow separation. The airflow separator adjusts the airflow speed to blow the lighter weeds and straw out of the separation device and separate the heavier grains by settling; S34, gravity separation: the materials are further separated by a gravity separation device, and the density difference of the materials is used to separate the seeds with higher density from the impurities with lower density; S35, miscellaneous material discharge: the separated miscellaneous materials are discharged from the separation device through a discharge channel, and the miscellaneous materials include remaining straw, small pieces of weeds and other impurities; S36 collects the separated lost grains and transfers them to the granary through the auger.

6. The method for detecting the cleaning loss rate of an air suction harvester according to claim 1, characterized in that: The S4 includes: S41, photoelectric sensor arrangement: photoelectric sensors are installed on both sides of the transparent detection channel through which the material flows, to detect the number and size of the grains passing through the channel; S42, weight sensor arrangement: a high-precision weight sensor is installed at the end of the detection channel to detect the total weight of the grains after passing through the photoelectric sensor in real time; S43, data collection: the photoelectric sensor collects the light intensity change data when the grains pass through the detection channel in real time, and converts the data into digital signals, records the number of grains passing through each time interval, collects the grain weight data in real time, and records the total weight in each time interval; S44, data processing and correction: denoising and smoothing the data collected by the photoelectric sensor to eliminate the influence of ambient light and other interference factors; S45, data fusion and analysis: the data from the photoelectric sensor and the weight sensor are fused to obtain the number and weight of grains per unit time.

7. The method for detecting the cleaning loss rate of an air suction harvester according to claim 1, characterized in that: The S53 includes: S531, initializing calibration parameters: obtaining calibration parameters for the current calibration round from the selected calibration parameter set ; S532, sensor initial state record: record the initial state data of the photoelectric sensor and the weight sensor before applying the calibration parameters, including the light intensity baseline value and the initial weight baseline value; S533, apply photoelectric sensor calibration parameters: according to the calibration parameters , Photoelectric adjusts the settings of the photoelectric sensor; S534, applying weight sensor calibration parameters, specifically including: Adjust weight bias: Set a new weight bias ,in is the initial weight baseline value; Calibrate the weighing platform: Check and adjust the weighing platform of the weight sensor; S535, calibration parameter application confirmation: test the photoelectric sensor and weight sensor.

8. The method for detecting the cleaning loss rate of an air suction harvester according to claim 1, characterized in that: The S6 includes: S61, data collection: obtaining grain quantity data from the photoelectric sensor and the weight sensor in real time, and transmitting the data to the data processing unit; S62, data cleaning: cleaning the original data to remove outliers and noise data; S63, data filtering: use filtering technology to smooth the data and remove high-frequency noise; S64, Data Formatting: Convert the processed data into a standard format for subsequent processing and analysis, including: Timestamp addition: add timestamp to each data record; Unit conversion: unify the data units; S65, data storage: storing the cleaned, filtered and formatted data in a database of a central data processing unit, the database using distributed storage technology; S66, Data Verification and Calibration: Verify the processed data.

9. The method for detecting the cleaning loss rate of an air suction harvester according to claim 1, characterized in that: The S7 includes: S71, data preparation: obtain the processed grain quantity data from the data processing unit, including the weight of the lost grains detected per unit time and the data required for the actual weight of grain harvested per unit time; S72, unit time data division: the processed grain quantity data is divided according to unit time, and the time window size is T; S73, calculating the weight of lost grains per unit time: accumulating the weight of lost grains detected by the cleaning system per unit time ; S74, calculating the actual weight of grain harvested per unit time: calculating the actual weight of grain harvested per unit time , expressed as: ； in, is the actual cutting width, is the walking distance per unit time, is the yield per unit area; S75, loss rate calculation: based on the processed grain quantity data, calculate the cleaning loss rate per unit time , expressed as: ； in, is the weight of grain loss detected per unit time, is the actual weight of grain harvested per unit time, is the cleaning loss rate.

Images