Multi-channel seed counting monitoring system based on data fusion

Abstract

The invention discloses a multi-channel seed counting monitoring system based on data fusion, which is applied to the technical field of automatic seed counting of seed metering devices, and comprises a seed flow dispersion structure, a photoelectric sensor and a signal processing module, the signal processing module is used for processing output signals acquired by the photoelectric sensor from each seed monitoring channel, the method specifically comprises the following steps: determining a seed passing event based on digital signal level change of each seed monitoring channel, and updating a counter in combination with an event triggering interval threshold value to obtain a total seed count; in each time step length, performing state prediction by using an unscented Kalman filtering algorithm, generating Sigma points, propagating the Sigma points by using a system model, calculating a mean value and a covariance of a prediction state, and updating state estimation by combining the propagated Sigma points converted by an observation model; and adjusting the count of the seed monitoring channel and the total seed count based on the updated state estimation, and updating the counter. The seed counting accuracy is improved.

Description

Technical Field

[0001] The invention relates to the technical field of automatic seed counting of a seed meter, and in particular to a multi-channel seed counting monitoring system based on data fusion. Background Art

[0002] In modern agricultural production, precision seeding is essential to improve crop yield and quality. Seed counting is a key link in the precision seeding system, and its accuracy directly affects the uniformity and effect of seeding. Traditional seed counting methods mainly rely on mechanical counters, which trigger the counting mechanism through physical contact with seeds. However, mechanical counters have some limitations, such as being susceptible to seed size and shape, and possible jamming and wear problems during high-speed seeding.

[0003] To overcome these limitations, modern seeding machinery increasingly uses sensor technology to achieve seed counting. Common sensor types include photoelectric sensors, capacitive sensors, and ultrasonic sensors. Photoelectric sensors use the interruption of a light beam to detect the passage of seeds. They have the advantages of fast response speed, high resolution, and no influence from seed materials, making them very suitable for high-speed counting applications. Capacitive sensors detect seeds by measuring the change in capacitance caused by seeds. They are sensitive to the moisture content and conductivity of seeds and may need to be calibrated to adapt to different types of seeds. Ultrasonic sensors detect seeds by emitting ultrasonic waves and receiving echoes reflected by seeds. They can penetrate materials to a certain extent, but may be affected by environmental noise and the location of sensor installation. Although these sensor technologies are improvements over mechanical counters, they still face challenges in practical applications. For example, photoelectric sensors may not be able to accurately distinguish when multiple seeds pass through at the same time, capacitive sensors may be too sensitive to the physical characteristics of seeds, and ultrasonic sensors may produce false alarms due to environmental factors.

[0004] Therefore, how to provide a multi-channel seed counting monitoring system based on data fusion that can overcome the influence of seed physical properties, sowing speed and environmental factors in existing sowing counting technology and effectively improve seed counting accuracy and adaptability is an urgent problem that technicians in this field need to solve. Summary of the invention

[0005] In view of this, the present invention proposes a multi-channel seed counting monitoring system based on data fusion.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A multi-channel seed counting monitoring system based on data fusion, comprising: a seed flow dispersion structure, a photoelectric sensor, and a signal processing module;

[0008] Seed flow dispersion structure, including: dispersion plate, partition plate, seed monitoring channel;

[0009] The seed monitoring channel is located at the exit of the seed flow dispersion structure; the partition plate is arranged inside the seed flow dispersion structure, and is used to separate the seeds into two seed flows in advance; the dispersion plate is arranged inside the seed flow dispersion structure, and is used to disperse the seeds evenly into any seed monitoring channel, and a photoelectric sensor is installed in each seed monitoring channel;

[0010] The dispersion plate includes a first dispersion plate and a second dispersion plate, wherein the first dispersion plate and the second dispersion plate are respectively arranged in two spaces divided by the partition plate, so that the seeds pass through the first dispersion plate and the second dispersion plate in sequence; the first dispersion plate and the second dispersion plate are both in arc shape;

[0011] The signal processing module is used to process the output signals collected by the photoelectric sensor from each seed monitoring channel to count the seeds, specifically:

[0012] Step 1: Obtain the output signal of the photoelectric sensor of each seed monitoring channel and perform data preprocessing;

[0013] Step 2: Based on the digital signal level change of each seed monitoring channel, the seed passing event is determined, and combined with the event trigger interval threshold, the counter is updated to obtain the total seed count;

[0014] Step 3: In each time step, the unscented Kalman filter algorithm is used to predict the state, generate Sigma points, and propagate the Sigma points using the system model to calculate the mean and covariance of the predicted state, and update the state estimate in combination with the propagated Sigma points converted by the observation model;

[0015] Step 4: Based on the updated state estimate, adjust the count of each seed monitoring channel and the total seed count, and update the counter.

[0016] Optionally, in step 1, data preprocessing includes: amplification, filtering and analog-to-digital conversion.

[0017] Optionally, in step 2, the seed passing event is determined based on the digital signal level change of each seed monitoring channel as follows:

[0018]

[0019] Among them, event i (n) is the seed passing event; i is the seed monitoring channel; n is the sampling index; D i (n) is the digital signal level change; T c Trigger threshold for the event.

[0020] Optionally, in step 2, based on the seed passing event, combined with the event triggering interval threshold, the counter is updated to obtain the total seed count as follows:

[0021] Δt i (n) = t event,i (n)-t event，i (n-1);

[0022]

[0023] Among them, Δt i (n) is the time interval between consecutive events; i is the seed monitoring channel; n is the sampling index; t event,i (n) and t event，i (n-1) is the timestamp of the consecutive events; C i (n) is the counter update; T s is the event trigger interval threshold; C all (n) is the total seed count; L is the number of seed monitoring channels.

[0024] Optionally, in step 3, the state prediction is performed using the unscented Kalman filter algorithm in each time step, as follows:

[0025] x k|k-1 =f(x k ,u k )+w k ;

[0026]

[0027] Among them, x k|k-1 is the state prediction of the given k-1 time step information; f is the system dynamic equation; x k The state prediction at time step k; u k is the control input; w k is the process noise of time step k; P k|k-1 is the covariance matrix of the state prediction given k-1 time step information at time step k; F k is the state transfer matrix; P k-1 is the state covariance matrix of time step k-1; T is the transpose; Q k is the process noise covariance at time step k.

[0028] Optionally, in step 3, Sigma points are generated based on state prediction as follows:

[0029]

[0030] in, is the Sigma point; x k|k-1is the state prediction of the given k-1 time step information at time step k; a is the index of the Sigma point; P k|k-1 is the covariance matrix of the state prediction given the k-1 time information of the time step k; k is the time step; λ is the weight parameter; I is the identity matrix.

[0031] Optionally, in step 3, the system model is used to propagate Sigma points and calculate the mean and covariance of the predicted state as follows:

[0032]

[0033] in, is the Sigma point after propagation through the system model; a is the index of the Sigma point; k is the time step; f is the system dynamic equation; is the Sigma point; is the mean of the predicted state; b is the dimension of the state; is the mean weight; P k+1|k is the covariance of the predicted state; is the covariance weight; T is the transpose.

[0034] Optionally, in step 3, the state estimate is updated based on the mean and covariance of the predicted state in combination with the propagated Sigma points transformed by the observation model, as follows:

[0035]

[0036] Among them, y (a) is the observed residual; a is the index of the Sigma point; is the propagation Sigma point after transformation by the observation model, h is the observation equation; is the propagation Sigma point; k is the time step; is the predicted observed mean; S is the observed residual covariance matrix; H is the observation matrix; R is the observed noise covariance matrix; P k+1|k The covariance of the predicted state; T is the transpose; K is the Kalman gain; is the mean of the predicted state; I is the identity matrix.

[0037] Optionally, in step 4, based on the updated state estimate, the count of each seed monitoring channel and the total seed count are adjusted as follows:

[0038]

[0039] Among them, e k is the prediction error; is the updated state estimate after time step k; C i，kis the current count of seed monitoring channel i at time step k; ΔC i，k is the adjustment amount; Q k To adjust the gain; C i,k+1 is the count of seed monitoring channel i after updating at time step k+1; C all，k+1 Total seed count after time step k+1 update; L is the number of seed monitoring channels.

[0040] It can be seen from the above technical solutions that, compared with the prior art, the present invention proposes a multi-channel seed counting monitoring system based on data fusion. By using a partition plate to separate the seed dispersion and diversion structure into two spaces, the seeds are divided into two seed flows, and the first dispersion plate and the second dispersion plate are set in the two spaces divided by the partition plate to further assist the seed diversion, so that the continuous and dense seed flow is dispersed into intermittent and dispersed monomer seeds, and it is ensured that they are evenly distributed in each seed monitoring channel, which effectively reduces the monitoring deviation caused by seed overlap and improves the seed counting accuracy, that is, effectively reduces the phenomenon that the photoelectric sensor may not be able to accurately distinguish when multiple seeds pass through at the same time. Further, the present invention adopts a seed counting method based on data fusion, and uses an unscented Kalman filter algorithm to fuse data from different photoelectric sensors, predict and update the state estimation of seed counting, which can further reduce the error that may be introduced by a single sensor. The unscented Kalman filter algorithm can handle nonlinear systems and is suitable for complex dynamics in the seed counting process. In the face of sensor noise and external interference, the unscented Kalman filter algorithm provides better robustness and ensures the stability and reliability of counting. By calculating the time intervals between consecutive events, it is possible to effectively distinguish between the situations where a single seed falls and multiple seeds fall at the same time, thereby further improving the accuracy of seed counting. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in 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 described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0042] Figure 1 It is a schematic diagram of the system structure of the present invention.

[0043] Figure 2 It is a schematic diagram of the signal processing flow of the signal processing module of the present invention.

[0044] Among them, there are a seed metering device-1, a seed dropping inlet-2, a seed dispersion and diversion structure-3, a seed inlet-31, a first dispersion plate-32, a second dispersion plate-33, a partition plate-34, a seed monitoring channel-35, a photoelectric sensor receiving end-36, a photoelectric sensor transmitting end-37, a signal processing module-38, a monitoring light-4, and a seed guide tube-5. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Embodiment 1:

[0047] Embodiment 1 of the present invention discloses a multi-channel seed counting monitoring system based on data fusion, such as Figure 1 As shown, including:

[0048] A seed flow dispersion structure 3, located at the discharge port of the seed metering device 1, is used to disperse the continuous seed flow that passes through the seed drop inlet 2 and the seed inlet 31 in sequence into a discontinuous seed flow, a photoelectric sensor, including a photoelectric sensor receiving end 36 and a photoelectric sensor transmitting end 37 that transmits a monitoring light 4 to the photoelectric sensor receiving end 36, and is used to monitor whether the seeds fall from the seed monitoring channel 35 (the embodiment of the present invention is provided with 9 seed monitoring channels), and a control module 38 to realize data collection, processing and communication;

[0049] The seed flow dispersion structure 3 includes: a dispersion plate, a partition plate 34, and a seed monitoring channel 35;

[0050] The seed monitoring channel 35 is located at the exit of the seed flow dispersion structure 3; the partition plate 34 is arranged inside the seed flow dispersion structure 3, and is used to separate the seeds into two seed flows in advance; the dispersion plate is arranged inside the seed flow dispersion structure 3, and is used to make the seeds evenly dispersed into any seed monitoring channel 35 and then enter the seed guide tube 5, and each seed monitoring channel 35 is equipped with a photoelectric sensor;

[0051] The dispersion plate includes a first dispersion plate 32 and a second dispersion plate 33. The first dispersion plate 32 and the second dispersion plate 33 are respectively provided in the two spaces divided by the partition plate 34, so that the seeds pass through the first dispersion plate 32 and the second dispersion plate 33 in sequence; the first dispersion plate 32 and the second dispersion plate 33 are both in arc shape;

[0052] The signal processing module 38 is used to process the output signals collected by the photoelectric sensor from each seed monitoring channel 35 to count the seeds, specifically:

[0053] Step 1: Obtain the output signal of the photoelectric sensor of each seed monitoring channel and perform data preprocessing.

[0054] Set the initial state, including: seed counter C of each seed monitoring channel i =0,P 0 , x 0 ; Where i is the seed monitoring channel; P 0 is the initial covariance matrix; x 0 is the initial state vector.

[0055] Output signal S of photoelectric sensor i (t) Perform data preprocessing, including amplification, filtering, and analog-to-digital conversion. The analog-to-digital conversion is as follows:

[0056] D i (n) = ADC[A i (t n )];

[0057] Among them, D i (n) is the digital signal level change; A i It is the analog signal of the photoelectric sensor output signal; n is the time of the nth sampling; n is the sampling index.

[0058] Step 2: Based on the digital signal level change of each seed monitoring channel, the seed passing event is determined, and combined with the event trigger interval threshold, the counter is updated to obtain the total seed count.

[0059] Seed passing events are determined based on the change in digital signal level of each seed monitoring channel as follows:

[0060]

[0061] Among them, event i (n) is the seed passing event; i is the seed monitoring channel; n is the sampling index; D i (n) is the digital signal level change; T c Trigger threshold for the event.

[0062] Based on the seed passing event, combined with the event trigger interval threshold, the counter is updated to obtain the total seed count as follows:

[0063] Δt i (n) = t event,i (n)-t e v ent，i (n-1);

[0064]

[0065] Among them, Δt i (n) is the time interval between consecutive events; i is the seed monitoring channel; n is the sampling index; t event,i (n) and t event，i (n-1) is the timestamp of the consecutive events; C i (n) is the counter update; T s is the event trigger interval threshold, which is used to distinguish between a single seed falling and multiple seeds falling without intervals; C all (n) is the total seed count; L is the number of seed monitoring channels.

[0066] Step 3: In each time step, the unscented Kalman filter (UKF) algorithm is used to predict the state, generate Sigma points, and propagate the Sigma points using the system model. The mean and covariance of the predicted state are calculated, and the state estimate is updated in combination with the propagated Sigma points after conversion by the observation model.

[0067] In each time step, the unscented Kalman filter algorithm is used to predict the state as follows:

[0068] x k|k - 1 =f(x k ,u k )+W k ;

[0069]

[0070] Among them, x k|k-1 is the state prediction of the given k-1 time step information; f is the system dynamic equation; x k The state prediction at time step k; u k is the control input; W k is the process noise of time step k; P k|k-1 is the covariance matrix of the state prediction given k-1 time step information at time step k; F k is the state transfer matrix; P k-1 is the state covariance matrix of time step k-1; T is the transpose; Q k is the process noise covariance at time step k.

[0071] Based on the state prediction, Sigma points are generated as follows:

[0072]

[0073] in, is the Sigma point; x k|k-1 is the state prediction of the given k-1 time step information at time step k; a is the index of the Sigma point; Pk|k-1 is the covariance matrix of the state prediction given the k-1 time information of the time step k; k is the time step; λ is the weight parameter; I is the identity matrix.

[0074] Using the system model to propagate the Sigma point, the mean and covariance of the predicted state are calculated as follows:

[0075]

[0076] in, is the Sigma point after propagation through the system model; a is the index of the Sigma point; k is the time step; f is the system dynamic equation; is the Sigma point; is the mean of the predicted state; b is the dimension of the state; is the mean weight; P k+1|k is the covariance of the predicted state; is the covariance weight; T is the transpose.

[0077] Based on the mean and covariance of the predicted state, the state estimate is updated in combination with the propagated Sigma point after transformation by the observation model, as follows:

[0078]

[0079] Among them, y (a) is the observed residual; a is the index of the Sigma point; is the propagation Sigma point after transformation by the observation model, h is the observation equation; is the propagation Sigma point; k is the time step; is the predicted observed mean; S is the observed residual covariance matrix; H is the observation matrix, which describes the first-order partial derivative of the observation vector with respect to the state vector, and its elements are defined as Among them, H ej is the element in the e-th row and j-th column of the observation matrix, h e is the e-th element in the observation vector h, x j is the jth element in the state vector x, is the partial derivative; R is the observation noise covariance matrix; P k+1|k The covariance of the predicted state; T is the transpose; K is the Kalman gain; is the mean of the predicted state; I is the identity matrix.

[0080] Step 4: Based on the updated state estimate, adjust the count of each seed monitoring channel and the total seed count, and update the counter.

[0081] Based on the updated state estimate, adjust the count of each seed monitoring channel and the total seed count as follows:

[0082]

[0083] Among them, e k is the prediction error; is the updated state estimate after time step k; C i，k is the current count of seed monitoring channel i at time step k; ΔC i，k is the adjustment amount; Q k To adjust the gain; C i，k+1 is the count of seed monitoring channel i after updating at time step k+1; C all，k+1 Total seed count after time step k+1 update; L is the number of seed monitoring channels.

[0084] The embodiment of the present invention discloses a multi-channel seed counting monitoring system based on data fusion. By using a partition plate to separate the seed dispersion and diversion structure into two spaces, the seeds are divided into two seed flows, and a first dispersion plate and a second dispersion plate are set in the two spaces divided by the partition plate to further assist the seed diversion, so that the continuous and dense seed flow is dispersed into intermittent and dispersed monomer seeds, and it is ensured that they are evenly distributed in each seed monitoring channel, effectively reducing the monitoring deviation caused by seed overlap, and improving the seed counting accuracy, that is, effectively reducing the phenomenon that the photoelectric sensor may not be able to accurately distinguish when multiple seeds pass through at the same time. Further, the present invention adopts a seed counting method based on data fusion, and uses an unscented Kalman filter algorithm to fuse data from different photoelectric sensors, predict and update the state estimation of seed counting, which can further reduce the error that may be introduced by a single sensor. The unscented Kalman filter algorithm can handle nonlinear systems and is suitable for complex dynamics in the seed counting process. In the face of sensor noise and external interference, the unscented Kalman filter algorithm provides better robustness to ensure the stability and reliability of counting. By calculating the time intervals between consecutive events, it is possible to effectively distinguish between the situations where a single seed falls and multiple seeds fall at the same time, thereby further improving the accuracy of seed counting.

[0085] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0086] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-channel seed counting monitoring system based on data fusion, characterized in that: include: Seed flow dispersion structure, photoelectric sensor, signal processing module; The seed flow dispersion structure includes: a dispersion plate, a partition plate, and a seed monitoring channel; The seed monitoring channel is located at the outlet of the seed flow dispersion structure; the partition plate is arranged inside the seed flow dispersion structure, and is used to separate the seeds into two seed flows in advance; the dispersion plate is arranged inside the seed flow dispersion structure, and is used to make the seeds evenly dispersed into any of the seed monitoring channels, and each of the seed monitoring channels is equipped with the photoelectric sensor; The dispersion plate comprises a first dispersion plate and a second dispersion plate, and the first dispersion plate and the second dispersion plate are respectively arranged in the two spaces divided by the partition plate, so that the seeds pass through the first dispersion plate and the second dispersion plate in sequence; the first dispersion plate and the second dispersion plate are both in arc shape; The signal processing module is used to process the output signals collected by the photoelectric sensor from each of the seed monitoring channels to perform seed counting, specifically: Step 1: Obtain the output signal of the photoelectric sensor of each seed monitoring channel and perform data preprocessing; Step 2: Based on the digital signal level change of each seed monitoring channel, determine the seed passing event, and update the counter in combination with the event trigger interval threshold to obtain the total seed count; Step 3: In each time step, the unscented Kalman filter algorithm is used to predict the state, generate a set of selected sample Sigma points for capturing and approximating the state distribution of the nonlinear system, and propagate the Sigma points using the system model, calculate the mean and covariance of the predicted state, and update the state estimate in combination with the propagated Sigma points after transformation by the observation model; Step 4: Based on the updated state estimate, adjust the count of each seed monitoring channel and the total seed count, and update the counter.

2. A multi-channel seed counting monitoring system based on data fusion according to claim 1, characterized in that: In step 1, the data preprocessing includes: amplification, filtering and analog-to-digital conversion.

3. The multi-channel seed counting monitoring system based on data fusion according to claim 1 is characterized in that: In step 2, a seed passing event is determined based on the digital signal level change of each of the seed monitoring channels, as follows: Among them, event i (n) is the seed passing event; i is the seed monitoring channel; n is the sampling index; D i (n) is the digital signal level change; T c Trigger threshold for the event.

4. The multi-channel seed counting monitoring system based on data fusion according to claim 1 is characterized in that: In step 2, based on the seed passing event and the event triggering interval threshold, the counter is updated to obtain the total seed count as follows: Δt i (n)=t event,i (n)-t event,i (n-1); Among them, Δt i (n) is the time interval between consecutive events; i is the seed monitoring channel; n is the sampling index; t event,i (n) and t event,i (n-1) is the timestamp of the consecutive events; C i (n) is the counter update; T s is the event trigger interval threshold; C all (n) is the total seed count; L is the number of seed monitoring channels.

5. The multi-channel seed counting monitoring system based on data fusion according to claim 1 is characterized in that: In step 3, the unscented Kalman filter algorithm is used to predict the state in each time step as follows: x k|k-1 =f(x k ,u k )+w k ; Among them, x k|k-1 is the state prediction of the given k-1 time step information; f is the system dynamic equation; x k The state prediction at time step k; u k is the control input; w k is the process noise of time step k; P k|k-1 is the covariance matrix of the state prediction given k-1 time step information at time step k; F k is the state transfer matrix; P k-1 is the state covariance matrix of time step k-1; T is the transpose; Q k is the process noise covariance at time step k.

6. The multi-channel seed counting monitoring system based on data fusion according to claim 1 is characterized in that: In step 3, based on the state prediction, Sigma points are generated as follows: in, is the Sigma point; x k|k-1 is the state prediction of the given k-1 time step information at time step k; a is the index of the Sigma point; P k|k-1 is the covariance matrix of the state prediction given the k-1 time information of the time step k; k is the time step; λ is the weight parameter; I is the identity matrix.

7. The multi-channel seed counting monitoring system based on data fusion according to claim 1 is characterized in that: In step 3, the Sigma point is propagated using the system model to calculate the mean and covariance of the predicted state as follows: in, is the Sigma point after propagation through the system model; a is the index of the Sigma point; k is the time step; f is the system dynamic equation; is the Sigma point; is the mean of the predicted state; b is the dimension of the state; is the mean weight; P k+1|k is the covariance of the predicted state; is the covariance weight; T is the transpose.

8. The multi-channel seed counting monitoring system based on data fusion according to claim 1 is characterized in that: In step 3, the state estimate is updated based on the mean and covariance of the predicted state and the propagated Sigma point transformed by the observation model as follows: Among them, y (a) is the observed residual; a is the index of the Sigma point; is the propagation Sigma point after transformation by the observation model, h is the observation equation; is the propagation Sigma point; k is the time step; is the predicted observed mean; S is the observed residual covariance matrix; H is the observation matrix; R is the observed noise covariance matrix; P k+1|k The covariance of the predicted state; R is the transpose; K is the Kalman gain; is the mean of the predicted state; I is the identity matrix.

9. The multi-channel seed counting monitoring system based on data fusion according to claim 1 is characterized in that: In step 4, based on the updated state estimate, the count of each seed monitoring channel and the total seed count are adjusted as follows: Among them, e k is the prediction error; is the updated state estimate after time step k; C i,k is the current count of seed monitoring channel i at time step k; ΔC i,k is the adjustment amount; Q k To adjust the gain; C i,k+1 is the count of seed monitoring channel i after updating at time step k+1; C all，k+1 Total seed count after time step k+1 update; L is the number of seed monitoring channels.

Images