A multi-sensor-based air-suction-disc-type seed-metering-unit seed-layer-thickness detection system and method

Abstract

The application provides a kind of based on multi-sensor's air suction disc type seed metering device seed layer thickness detection system and method, comprising the following steps: the distance of acquisition point to vibrating seed tray is obtained by at least one distance sensor;The data collected by at least one distance sensor is filtered;The filtered data are fused to obtain the seed layer thickness;The seed layer thickness is compared with the set seed layer thickness to determine the seed layer difference;According to the seed layer difference, the seed adding amount of the seed adding device is adjusted by RBF-PID control algorithm.The application uses RBF-PID closed-loop control, improves the accuracy of seed layer thickness detection, and optimizes the seed metering effect of seed metering device.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent agricultural equipment or the technical field of seeding quantity adjustment, and particularly relates to a kind of air suction disc type seed metering device seed layer thickness detection system and method based on multi-sensor. BACKGROUND

[0002] China's rice planting area is about 30 million mu, accounting for about 30% of the total grain planting area in China, and about 20% of the world's rice planting area, and whether rice can obtain high yield will involve the overall livelihood problem. Super rice is the first choice of the majority of grain growers with its super high yield, and with the continuous popularization of rice mechanization and intelligent planting, super rice factory seedling becomes one of the main ways of rice mechanized planting.

[0003] In the process of seedling transplanting, super rice requires 1-2 plants per hole, and the current mechanical seed metering device cannot meet the requirements of seed metering accuracy, and is easy to damage rice seeds. The air suction vibrating disc type seed metering device makes the rice seeds in a controllable "boiling" state through the periodic vibration of the seed disc, and the rice seeds are uniformly distributed in the seed disc space. The suction disc relies on the negative pressure provided by the fan to generate suction force on the moving rice seeds, thereby realizing efficient and reliable seed metering of the seed metering device.

[0004] The seed metering qualified rate of the seed metering device during seed metering depends largely on the seed suction effect of the suction disc, which is mainly affected by factors such as seed disc vibration frequency, amplitude, seed layer thickness, etc. In order to ensure the seed suction effect and maintain the seed disc seed layer thickness within a certain range, the seed disc seed layer needs to be monitored in real time. When the air suction vibrating disc type seed metering device is working, the seed disc is in a periodic equal-amplitude vibration process with a vibration amplitude of 1-6mm and a frequency of 1-15Hz, and it is difficult to monitor the seed layer thickness of the rice seeds in the seed disc in real time. The existing seed layer thickness acquisition under this dynamic operation condition cannot achieve good detection effect. SUMMARY

[0005] In view of the deficiencies in the prior art, the present application provides a kind of air suction disc type seed metering device seed layer thickness detection system and method based on multi-sensor, which obtains ranging data in real time by laser ranging sensor and ultrasonic ranging sensor, and carries out data fusion after filtering the measurement data of the two, reduces the instability caused by the technology itself of laser ranging and ultrasonic ranging, and obtains the seed layer thickness after data processing and data conversion. RBF-PID closed loop control is used to improve the accuracy of seed layer thickness detection and optimize the seed metering effect of the seed metering device.

[0006] The present application achieves the above technical purpose by the following technical means.

[0007] A kind of air suction disc type seed metering device seed layer thickness detection method based on multi-sensor, comprising the following steps:

[0008] The distance from collection point to vibrating seed tray is obtained by at least one distance sensor;

[0009] The data collected by at least one distance sensor is filtered;

[0010] The filtered data is fused to obtain seed layer thickness:

[0011] The seed layer thickness is compared with the set seed layer thickness to determine the seed layer difference;

[0012] According to the seed layer difference, the seed adding amount of the seed adding device is adjusted by RBF-PID control algorithm.

[0013] Further, the data collected by at least one distance sensor is filtered, specifically: the collected data is filtered by amplitude limiting filtering or sliding average filtering.

[0014] Further, the distance from collection point to vibrating seed tray is obtained by N first distance sensors and N second distance sensors;The fused filtered data is specifically:

[0015] The distance values a0, a1…a N-1 collected by N first distance sensors are determined, and the average value A=(a0+a1+…+a N-1 ) / N is determined;

[0016] The distance values b0, b1…b N-1 collected by N second distance sensors are determined, and the average value B=(b0+b1+…+b N-1 ) / N is determined;

[0017] When the initial non-seed is set, the distance values measured by first distance sensor and second distance sensor are all S0, then the seed layer thickness S A (k) at k time determined by first distance sensor is S0-A, and the seed layer thickness S B (k) at k time determined by second distance sensor is S0-B;

[0018] ΔS A (k)=S A (k)-S A (k-1), ΔS B (k)=S B (k)-S B (k-1),

[0019] Wherein S A (k-1) is the seed layer thickness at k-1 time determined by first distance sensor;S B(k-1) is the seed layer thickness at k-1 time determined by the second distance sensor;

[0020] If ΔS A (k) > set threshold, then make S A (k) = S A (k-1), enter the next adoption cycle; if ΔS B (k) > set threshold, then make S B (k) = S B (k-1), enter the next adoption cycle;

[0021] If ΔS A (k) < set threshold and ΔS B (k) < set threshold, calculate the seed layer thickness f m (k) = (S A (k) + S B (k)) / 2 at k time.

[0022] Further, according to the seed layer difference, the amount of seed adding device is adjusted by RBF-PID control algorithm, specifically:

[0023] Determine the seed layer difference e(k) at k time: e(k) = r(k) - f m (k), wherein r(k) is the given value of the seed layer thickness of the seed disc;

[0024] Take e(k)-e(k-1), e(k), e(k)-2e(k-1)+e(k-2) as the input of PID,

[0025] Wherein: e(k) represents the seed layer difference at k time, e(k-1) represents the seed layer difference at k-1 time, e(k-2) represents the seed layer difference at k-2 time;

[0026] The output of the PID is Δu(k), which can be represented as:

[0027] Δu(k) = K p [e(k)-e(k-1)] + K i e(k) + K d [e(k)-2e(k-1)+e(k-2)];

[0028] Adjust the K p , K i and K d parameters in PID through RBF network;

[0029] The seed adding device determines the amount of seed at k time according to the amount of seed at k-1 time and the PID output Δu(k), specifically: u(k) = u(k-1) + Δu(k),

[0030] wherein u(k) represents the controller output at time k; and u(k-1) represents the controller output at time k-1.

[0031] Further, the K p , K i and K d parameters in the PID are adjusted by the RBF network, specifically:

[0032] The filtered data is input into the RBF network, and the x vector of the RBF network Gaussian function is acted upon;

[0033] The RBF network Gaussian function is:

[0034] wherein h i (x) represents the Gaussian function value of the i th neuron in the hidden layer when the input is the x vector, c i represents the center of the Gaussian function of the i th neuron in the hidden layer; σ i represents the width of the Gaussian function of the i th neuron in the hidden layer, and m is the number of neurons in the hidden layer.

[0035] The identification output of the RBF network is: ω i is the output weight of the network; and y m (k) is the identification output of the network at time k.

[0036] The square of the error is used as the setting performance index of the RBF neural network:

[0037]

[0038] The parameter increment of the PID controller is adjusted by the gradient descent method:

[0039]

[0040]

[0041]

[0042] wherein:

[0043] e(k) represents the error at time k;

[0044] e(k-1) represents the error at time k-1;

[0045] e(k-2) represents the error at time k-2;

[0046] ΔK p (k), ΔK i (k), and ΔK d(k) represents the increment of the PID parameters at time k;

[0047] y m (k) represents the output of the RBF network at time k;

[0048] η p For K p - The proportion of the learning rate;

[0049] η i For K i - The learning rate of the integral;

[0050] η d For K d - The learning rate of the derivative;

[0051] E(k) represents the performance index at time k;

[0052] Adjusting the PID via an RBF network can be expressed as:

[0053]

[0054]

[0055]

[0056] In the above formula:

[0057] K p (k) represents the scaling parameter at time k;

[0058] K p (k-1) represents the scaling parameter at time k-1;

[0059] K p (k-2) represents the scaling parameter at time k-2;

[0060] K i (k) represents the integral parameter at time k;

[0061] K i (k-1) represents the integral parameter at time k-1;

[0062] K i (k-2) represents the integral parameter at time k-2;

[0063] K d (k) represents the differential parameter at time k;

[0064] K d (k-1) represents the differential parameter at time k-1;

[0065] K d(k-2) represents a differential parameter at k-2 time;

[0066] α c is an introduced PID controller parameter momentum factor;

[0067] ΔK p (k), ΔK i (k), ΔK d (k) is a PID parameter increment at k time.

[0068] Further, K p The learning rate η of proportion p The value is 0.055; K i The learning rate η of integration i The value is 0.09; K d The learning rate η of differential d The value is 0.06; the parameter momentum factor α c The value is 0.75.

[0069] A detection system of a seed layer thickness detection method of an air suction disc type seed metering device based on multiple sensors, comprising a distance information acquisition module, a filtering algorithm module, an information fusion module, a given value comparison module and an RBF-PID control module;

[0070] The distance information acquisition module is used for acquiring the distance from the distance acquisition point to the vibrating seed disc; the distance information acquisition module inputs the acquired information into the filtering algorithm module, which is used for filtering the data acquired by the distance information acquisition module; the filtering algorithm module inputs the filtered ranging data into the information fusion module, which is used for fusing the filtered ranging data, so as to determine the seed layer thickness; the given value comparison module determines the seed layer difference value according to the determined seed layer thickness and the set seed layer thickness.

[0071] The given value comparison module inputs the seed layer difference value into the RBF-PID control module, and the filtering algorithm module inputs the filtered ranging data into the RBF-PID control module; the RBF-PID control module outputs the seed adding amount, which is used for controlling the seed adding device.

[0072] Further, the distance information acquisition module comprises laser ranging sensors and ultrasonic ranging sensors; a plurality of laser ranging sensors and a plurality of ultrasonic ranging sensors are installed above the vibrating seed disc.

[0073] Further, the time interval adopted by adjacent laser ranging sensors or adjacent ultrasonic ranging sensors is an integer multiple of the seed disc vibration period.

[0074] The beneficial effects of the present application are:

[0075] The multi-sensor-based air-suction-disc-type seed-layer-thickness detection system and method provided by the application obtains ranging data in real time through laser ranging sensors and ultrasonic ranging sensors, filters the two sets of measurement data, and then performs data fusion, thereby reducing the instability caused by the technical characteristics of laser ranging and ultrasonic ranging, and performing data conversion after data processing to obtain the seed layer thickness; the use of RBF-PID closed-loop control improves the seed layer thickness detection accuracy and optimizes the seed-metering effect of the seed meter. BRIEF DESCRIPTION OF DRAWINGS

[0076] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. The drawings in the following description are some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0077] Figure 1 The flow chart of the multi-sensor-based air-suction-disc-type seed-metering device seed-layer-thickness detection method provided by the application.

[0078] Figure 2 The control principle diagram of the multi-sensor-based air-suction-disc-type seed-metering device seed-layer-thickness detection method provided by the application.

[0079] Figure 3 The sensor installation position diagram provided by the application.

[0080] Figure 4 The periodic sampling waveform diagram provided by the application.

[0081] Figure 5 The work flow chart of the multi-sensor-based air-suction-disc-type seed-metering device seed-layer-thickness detection method provided by the application.

[0082] In the drawings:

[0083] 101-first laser ranging sensor; 102-second laser ranging sensor; 103-third laser ranging sensor; 104-first ultrasonic ranging sensor; 105-second ultrasonic ranging sensor; 106-third ultrasonic ranging sensor; 107-rice seed. DETAILED DESCRIPTION

[0084] The application will be further described below in combination with the drawings and specific embodiments, but the protection scope of the application is not limited thereto.

[0085] Embodiments of the present application are described below in detail with reference to the accompanying drawings, examples of which are shown in the drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by reference to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0086] In the description of the present application, it should be understood that the orientation or positional relationship indicated by the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "axial", "radial", "vertical", "horizontal", "inner", "outer" and the like is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application. In addition, the terms "first", "second" are only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined as "first", "second" can be explicitly or implicitly included one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0087] In the present application, unless otherwise specifically defined and limited, the terms "mounting", "connection", "connection", "fixing" and the like should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or it can be integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0088] As shown in Figure 1 , Figure 2 and Figure 5 , the seed layer thickness detection method of the air suction disc type seed metering device based on multiple sensors of the present application comprises the following steps:

[0089] S01: acquiring the distance from the collection point to the vibrating seed disc through the laser ranging sensor and the ultrasonic ranging sensor;

[0090] S02: filtering the data collected by the laser ranging sensor and the ultrasonic ranging sensor, i.e. filtering the collected data through amplitude limiting filtering or sliding average filtering to remove the singular data values caused by vibration adjustment and other interference during the operation of the seed metering device, and acquiring stable and reliable effective distance data;

[0091] S03: fusing the filtered data to obtain the seed layer thickness, specifically:

[0092] through N laser ranging sensors and N ultrasonic ranging sensors;

[0093] N laser ranging sensors collect distance values a0, a1…a N-1 , determine the average value A=(a0+a1+…+a N-1 ) / N;

[0094] N ultrasonic ranging sensors collect distance values b0, b1…b N-1 , determine the average value B=(b0+b1+…+b N-1 ) / N;

[0095] Set the distance values measured by the laser ranging sensor and the ultrasonic ranging sensor when there is no initial seed as S0, then determine the seed layer thickness S A (k) = S0-A through the laser ranging sensor, and determine the seed layer thickness S B (k) = S0-B through the ultrasonic ranging sensor;

[0096] ΔS A (k) = S A (k) - S A (k-1), ΔS B (k) = S B (k) - S B (k-1),

[0097] wherein S A (k-1) is the seed layer thickness determined by the laser ranging sensor at k-1 time; S B (k-1) is the seed layer thickness determined by the ultrasonic ranging sensor at k-1 time;

[0098] If ΔS A (k) > set threshold value, then make S A (k) = S A (k-1), enter the next adoption cycle; if ΔS B (k) > set threshold value, then make S B (k) = S B (k-1), enter the next adoption cycle;

[0099] If ΔS A (k) < set threshold value and ΔS B (k) < set threshold value, calculate the seed layer thickness f m (k) = (S A (k) + S B (k)) / 2 at k time. The set threshold value is generally 1 cm.

[0100] S04: compare the seed layer thickness at time k with the set seed layer thickness to determine the seed layer difference value:

[0101] Determine the seed layer difference value e(k) at time k: e(k) = r(k) - f m (k), where r(k) is a given value of the seed disc seed layer thickness;

[0102] S05: adjust the seeding amount of the seeding device according to the seed layer difference value through the RBF-PID control algorithm, specifically:

[0103] S5.1: determine the seed layer difference value e(k) at time k: e(k) = r(k) - f m (k), where r(k) is a given value of the seed disc seed layer thickness;

[0104] S5.2: take e(k) - e(k-1), e(k), and e(k) - 2e(k-1) + e(k-2) as the inputs of the PID,

[0105] where: e(k) represents the seed layer difference value at time k, e(k-1) represents the seed layer difference value at time k-1, and e(k-2) represents the seed layer difference value at time k-2; the output of the PID is Δu(k), which can be represented as:

[0106] Δu(k) = K p [e(k) - e(k-1)] + K i e(k) + K d [e(k) - 2e(k-1) + e(k-2)];

[0107] S5.3: adjust the K p , K i , and K d parameters in the PID through the RBF network, specifically:

[0108] S5.3.1: input the filtered data into the RBF network, and take the x vector of the RBF network Gaussian function;

[0109] S5.3.2: the RBF network Gaussian function is: where h i (x) represents the Gaussian function value of the i-th neuron in the hidden layer when the input is the x vector, c i represents the center of the Gaussian function of the i-th neuron in the hidden layer; σ i represents the width of the Gaussian function of the i-th neuron in the hidden layer, and m is the number of neurons in the hidden layer;

[0110] S5.3.2: the identification output of the RBF network is ω i is the output weight of the network; ym (k) is the network identification output at time k;

[0111] S5.3.3: The square of the error is used as the setting performance index of the RBF neural network:

[0112] S5.3.4: The parameter increment of the PID controller is adjusted by the gradient descent method:

[0113]

[0114]

[0115]

[0116] wherein,

[0117] e(k) represents the seed layer difference value at time k;

[0118] e(k-1) represents the seed layer difference value at time k-1;

[0119] e(k-2) represents the seed layer difference value at time k-2;

[0120] ΔK p (k), ΔK i (k), ΔK d (k) is the PID parameter increment at time k;

[0121] y m (k) is the output of the RBF network at time k;

[0122] η p is the learning rate of K p -proportion;

[0123] η i is the learning rate of K i -integration;

[0124] η d is the learning rate of K d -differentiation;

[0125] E(k) represents the performance index at time k;

[0126] S5.3.4: The adjustment of the PID by the RBF network can be represented as:

[0127]

[0128]

[0129]

[0130] In the above formula, in the above formula:

[0131] K p (k) represents the proportional parameter at time k;

[0132] K p (k-1) represents the proportional parameter at time k-1;

[0133] K p (k-2) represents the proportional parameter at time k-2;

[0134] K i (k) represents the integral parameter at time k;

[0135] K i (k-1) represents the integral parameter at time k-1;

[0136] K i (k-2) represents the integral parameter at time k-2;

[0137] K d (k) represents the differential parameter at time k;

[0138] K d (k-1) represents the differential parameter at time k-1;

[0139] K d (k-2) represents the differential parameter at time k-2;

[0140] α c is a momentum factor of the introduced PID controller parameter;

[0141] ΔK p (k), ΔK i (k), ΔK d (k) is the PID parameter increment at time k.

[0142] K p - learning rate η of the proportional p takes a value of 0.055; K i - learning rate η of the integral i takes a value of 0.09; K d - learning rate η of the differential d takes a value of 0.06; the parameter momentum factor α c takes a value of 0.75.

[0143] S5.4: The dosing device determines the dosing amount at time k according to the dosing amount at time k-1 and the PID output Δu(k), specifically: u(k) = u(k-1) + Δu(k), wherein u(k) represents the controller output at time k; and u(k-1) represents the controller output at time k-1.

[0144] The application obtains ranging data in real time through a laser ranging sensor and an ultrasonic ranging sensor, and fuses the two measurement data after filtering, thereby reducing the instability caused by the technology of laser ranging and ultrasonic ranging itself, and the processor converts the data after processing to obtain the seed layer thickness of the air suction vibrating disc type seed metering device, and the seed adding device adjusts the seed layer thickness in combination with the set seed layer thickness range, constructs the RBF-PID closed loop control of the seed disc seed adding, and improves the seed layer thickness detection accuracy and optimizes the seed metering effect of the seed metering device.

[0145] The detection system of the seed layer thickness detection method of the air suction disc type seed metering device based on multiple sensors comprises a distance information acquisition module, a filtering algorithm module, an information fusion module, a given value comparison module and an RBF-PID control module.

[0146] The distance information acquisition module is used for acquiring the distance from the distance acquisition point to the vibrating seed disc, the distance information acquisition module inputs the acquired information into the filtering algorithm module, which is used for filtering the data acquired by the distance information acquisition module, the filtering algorithm module inputs the filtered ranging data into the information fusion module, which is used for fusing the filtered ranging data to determine the seed layer thickness, and the given value comparison module determines the seed layer difference value according to the determined seed layer thickness and the set seed layer thickness.

[0147] The given value comparison module inputs the seed layer difference value into the RBF-PID control module, the filtering algorithm module inputs the filtered ranging data into the RBF-PID control module, and the RBF-PID control module outputs the seed adding amount, which is used for controlling the seed adding device.

[0148] As shown in Figure 3 The distance information acquisition module comprises a laser ranging sensor and an ultrasonic ranging sensor, and a plurality of laser ranging sensors and a plurality of ultrasonic ranging sensors are installed above the vibrating seed disc. The laser ranging sensor and the ultrasonic ranging sensor are uniformly arranged on the sensor support, the uniformly distributed laser ranging sensor and the ultrasonic ranging sensor respectively measure the seed layer thickness at the position, and the average value of the filtered data is taken respectively, thereby reducing the error caused by the local uneven distribution in the seed disc vibration process. In the embodiment, the sensor support is provided with two rows, the first row of sensor supports are equidistantly installed with the first laser ranging sensor 101, the second laser ranging sensor 102 and the third laser ranging sensor 103, the second row of sensor supports are equidistantly installed with the first ultrasonic ranging sensor 101, the second ultrasonic ranging sensor 102 and the third ultrasonic ranging sensor 103, and the laser ranging sensor and the ultrasonic ranging sensor are arranged in a staggered manner. Figure 4As shown, the time interval adopted by the adjacent laser ranging sensor or the adjacent ultrasonic ranging sensor is an integer multiple of the disc vibration period. In the figure, m1 and m2 are the continuous detection points of the sensor, and the time interval of m1 and m2 is set to be an integer multiple of the disc vibration period, which can solve the problem that the ranging accuracy and conversion time of the existing laser ranging sensor and ultrasonic sensor cannot meet the sampling requirements of the scheme, improve the stability of obtaining sampling data, and improve the accuracy of thickness detection.

[0149] As an embodiment of the present application, when the multi-sensor-based air suction disc type seed metering device seed layer thickness detection system is working, the system is initialized after power-on, it is judged whether the set seed layer thickness needs to be adjusted, if the key is pressed, the set seed layer thickness is increased or decreased, and then the laser ranging sensor and the ultrasonic ranging sensor respectively obtain the distance value between the sensor and the seed disc in real time; if the key is not pressed, the ranging values of the laser ranging sensor and the ultrasonic ranging sensor are directly obtained. The ranging values of the laser ranging sensor and the ultrasonic ranging sensor obtained by the sensor are respectively filtered to remove periodic vibration and accidental interference, the laser ranging value and the ultrasonic ranging value are data fused, and the average thickness is displayed. At the same time, the ranging values of the three laser ranging sensors and the three ultrasonic ranging sensors are compared respectively to judge the uniformity of the seed layer thickness. If the uniformity is poor, a delay of 2s is performed, and then the judgment is performed again. If the uniformity is still very poor, an audible and visual alarm is performed; if the uniformity is good, it is judged whether the thickness exceeds the corresponding limit value. If the thickness value exceeds the set threshold value, the seed adding device performs corresponding action, that is, if the thickness value is too large, the seed adding is stopped, and if the thickness value is too small, the seed adding is started; if the thickness value is within the set threshold value, it is judged again whether the seed layer thickness range needs to be adjusted.

[0150] It should be understood that although the present specification is described in terms of various embodiments, not every embodiment contains only one independent technical solution, and the description of the specification is only for the sake of clarity, and those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be combined appropriately to form other embodiments that those skilled in the art can understand.

[0151] The series of detailed descriptions listed above are only specific descriptions of feasible embodiments of the present application, and are not used to limit the protection scope of the present application. Any equivalent embodiments or changes made without departing from the spirit of the present application should be included in the protection scope of the present application.

Claims

1. A multi-sensor-based air-sucker-type seed-metering device seed layer thickness detection method, characterized by, The method comprises the following steps: acquiring the distance from the collection point to the vibrating seed tray by at least one distance sensor; filtering the data collected by the at least one distance sensor; fusing the filtered data to obtain the seed layer thickness: comparing the seed layer thickness with the set seed layer thickness to determine the seed layer difference; adjusting the seed adding amount of the seed adding device according to the seed layer difference through the RBF-PID control algorithm, specifically: determining a seedbed difference value at time k : wherein is a given value of the seedbed thickness; is the seedbed thickness at time k; Will , As , wherein: represents the seed layer difference value at time k, represents the seed layer difference value at time k-1, represents the seed layer difference value at time k-2; the is , is represented as: ； Adjusting the parameters of PID by RBF network , and , specifically: inputting the filtered data into the RBF network as the x vector of the Gaussian function of the RBF network; The Gaussian function of the RBF network is: wherein, represents the Gaussian function value of the i-th neuron in the hidden layer when the input is the x vector, represents the center of the Gaussian function of the i-th neuron in the hidden layer; represents the width of the Gaussian function of the i-th neuron in the hidden layer, and m is the number of neurons in the hidden layer. The RBF network identification output is , is the output weight of the network; is the network identification output at time k The square of error is used as the setting performance index of RBF neural network: ; adjusting the parameter increment of the PID controller through the gradient descent method: ， ， ， wherein, denotes the seed layer difference value at time k, denotes the seed layer difference value at time k-1, denotes the seed layer difference value at time k-2; is the PID parameter increment at time k; is the output of the RBF network at time k; is the proportional learning rate; is the integral learning rate; is the derivative learning rate; denotes the performance indicator at time k;​​ adjusting the PID through the RBF network, which is expressed as: ， ， ， In the above formulae, denotes the proportional parameter at time k, denotes the proportional parameter at time k-1, denotes the proportional parameter at time k-2; denotes the integral parameter at time k, denotes the integral parameter at time k-1, denotes the integral parameter at time k-2; denotes the derivative parameter at time k, denotes the derivative parameter at time k-1, denotes the derivative parameter at time k-2; is a momentum factor for the introduced PID controller parameters; , , is a PID parameter increment at time k; The adding device determines the adding amount at the kth moment according to the adding amount at the (k-1)th moment and , and specifically: , wherein represents the controller output at the kth moment; represents the controller output at the (k-1)th moment.​ 2. The multi-sensor-based air-sucker-pad-type seed-metering-unit seed layer thickness detection method according to claim 1, characterized in that, filtering the data collected by the at least one distance sensor, specifically:

3. The multi-sensor-based air-sucker-pad-type seed-metering-unit seed layer thickness detection method according to claim 1, characterized in that, acquiring the distance from the collection point to the vibrating seed tray by N first distance sensors and N second distance sensors; the fusion of the filtered data is specifically: N first distance sensors collect distance values a0, a1...a N-1 , determine the average value A = (a0+a1+...+a N-1 ) / N; N second distance sensors collect distance values b0, b1…b N-1 , determine the average value B = (b0+b1+…+b N-1 ) / N; Set the initial distance value measured by the first distance sensor and the second distance sensor without seeding as S0, then the seed layer thickness S at time k is determined by the first distance sensor A (k) = S0-A, the seed layer thickness S at time k is determined by the second distance sensor B (k) = S0-B; , where S A (k-1) is the seed layer thickness determined by the first distance sensor at time k-1; S B (k-1) is the seed layer thickness determined by the second distance sensor at time k-1; If > the threshold is set, then the next adoption cycle is entered; if > the threshold is not set, then the next adoption cycle is entered; If <Set threshold and <Set threshold, calculate seed layer thickness at time k (S A (k)+S B (k)) / 2.

4. The multi-sensor-based air-sucker-pad-type seed-metering-unit seed layer thickness detection method according to claim 1, characterized by, - learning rate for the proportion with a value of 0.055; - learning rate for the integral with a value of 0.09; - learning rate for the derivative with a value of 0.06; the parameter momentum factor with a value of 0.

75.

5. A detection system for detecting the seed layer thickness of a multi-sensor-based air-sucker-type seed meter according to any one of claims 1 to 4, characterized in that, The method comprises a distance information collection module, a filtering algorithm module, an information fusion module, a given value comparison module, and an RBF-PID control module. The distance information collection module is used to acquire the distance from the collection point to the vibrating seed tray; the distance information collection module inputs the collected information into the filtering algorithm module, which is used to filter the data collected by the distance information collection module. The filtering algorithm module inputs the filtered ranging data into the information fusion module, which is used to fuse the filtered ranging data to determine the seed layer thickness; the given value comparison module determines the seed layer difference according to the determined seed layer thickness and the set seed layer thickness. The given value comparison module inputs the seed layer difference into the RBF-PID control module, the filtering algorithm module inputs the filtered ranging data into the RBF-PID control module, and the RBF-PID control module outputs the seed adding amount, which is used to control the seed adding device.

6. The detection system of the multi-sensor-based air-sucker-pad-type seed layer thickness detection method according to claim 5, wherein The distance information collection module comprises laser ranging sensors and ultrasonic ranging sensors; a plurality of laser ranging sensors and a plurality of ultrasonic ranging sensors are installed above the vibrating seed tray.

7. The detection system of the multi-sensor-based air-sucker-pad-type seed layer thickness detection method according to claim 6, characterized in that, The time interval adopted by adjacent laser ranging sensors or adjacent ultrasonic ranging sensors is an integer multiple of the seed tray vibration period.

Images