Agricultural material cleaning test bed and cleaning test method

Abstract

The invention discloses an agricultural material cleaning test bed and a cleaning test method. The agricultural material cleaning test bed comprises a rack with a cleaning chamber, and a cleaning sieve, a main fan, a front blowing fan, a feeding device, a loss rate sensor and a controller are mounted on the rack; the rack is further provided with a sensor set used for detecting the air speed and the air volume. The feeding device comprises a feeding box and a feeding quantity adjusting device which are arranged up and down, the feeding box is through up and down, the interior of the feeding box is divided into a plurality of feeding grids which are arranged in a square array, and the feeding quantity adjusting device can adjust the feeding quantity or feeding flow of materials fed to the head end of the cleaning sieve by the feeding grids. The feeding device is redesigned, so that accurate adjustment of materials fed to the screen surface is realized. According to the test method, the feeding position, the feeding amount and the flow of the materials can be adjusted according to different feeding schemes, various feeding schemes easy to clean can be simulated, the feeding effect of material accumulation can also be simulated, and various cleaning schemes are formed by combining operation parameters and the rotating speed of a fan.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural material cleaning tests, and particularly to an agricultural material cleaning test bench and a cleaning test method. Background Art

[0002] In modern agricultural production, the post-harvest processing of crops, especially the cleaning link of materials, is crucial for improving the quality and market value of agricultural products. Traditional agricultural material cleaning equipment usually includes basic components such as screening devices and blowers, which are mainly used to remove light impurities such as dust, grass clippings, and small particulate debris in grains. However, with the expansion of agricultural production scale and the progress of technology, the requirements for cleaning efficiency and quality are getting higher and higher. Therefore, it becomes more and more important to test the cleaning effect of the cleaning device and adjust parameters.

[0003] In the prior art, Patent CN 215844131 U discloses a seed cleaning test bench, which has a seed feeding mechanism, a vibrating screen, and a blowing mechanism. Through cleaning tests, parameters suitable for seed screening can be obtained for reference during the debugging of actual cleaning machines; in the test bench of Patent CN 116140200 A, a cross-flow large blower and a cross-flow small blower are set, and a wind speed sensor is set to monitor the wind speed to determine the optimal parameter combination of the grain cleaning device. The above prior art has the following disadvantages:

[0004] (1) The material feeding scheme is not flexible enough, and the feeding method is relatively single. It is difficult to flexibly adjust the feeding position, distribution amount, and flow rate of materials according to different material characteristics or cleaning requirements. This limits the adaptability and cleaning effect of the equipment in different application scenarios.

[0005] (2) The operating parameters of traditional equipment (such as vibration frequency, blower speed, etc.) are mostly adjusted relying on manual experience, lacking scientific basis and automation degree, and it is difficult to achieve precise control. In addition, existing equipment often cannot monitor and feedback the changes of key parameters in the cleaning process in real time, resulting in lag in optimization and adjustment, affecting the final cleaning quality and efficiency. Summary of the Invention

[0006] Object of the Invention: In order to overcome the deficiencies in the prior art, the present invention provides an agricultural material cleaning test bench and a cleaning test method with a flexible feeding method that can simulate various feeding methods to simulate various situations encountered in the cleaning process, and the adjustment of the operating mode is more intelligent.

[0007] Technical solution: To achieve the above-mentioned purpose, the agricultural material cleaning test bench of the present invention comprises a frame with a cleaning chamber, on which a cleaning screen, a main fan, a front blowing fan, a feeding device, a loss rate sensor and a controller are installed; the cleaning screen comprises a shaking plate, an upper fish scale screen, a finger screen, a lower woven screen and a vibrating screen drive motor; a sensor group for detecting wind speed and air volume is also installed on the frame; the controller can adjust the speed of the main fan and the front blowing fan, as well as the operating parameters of the cleaning screen, and can obtain data collected by each sensor;

[0008] The feeding device includes a feeding box arranged in an upper and lower manner and a feeding amount adjusting device. The feeding box is connected from top to bottom and has a plurality of feeding grids arranged in a square array inside. The feeding amount adjusting device can adjust the amount or flow rate of materials released by the feeding grid to the head end of the cleaning screen.

[0009] Furthermore, the feed amount adjustment device includes an outer frame, and also includes a baffle plate arranged under each row of the feeding grids, and a rotating shaft rotating relative to the outer frame is fixed on the baffle plate; an adjustment handle is fixed on each of the rotating shafts, and all the adjustment handles are connected to the same connecting rod; the feed amount adjustment device also includes an adjustment mechanism for changing the position of the connecting rod; the adjustment mechanism includes a connecting arm connected to the connecting rod, and also includes a handle rotatably installed relative to the outer frame, a bent arm is rotatably installed on the handle, a first spring is connected between the bent arm and the connecting arm, and a second spring is connected between the connecting rod and the outer frame.

[0010] Furthermore, the sensor group includes a plurality of first wind speed and air volume sensors placed below the cleaning screen, and a plurality of second wind speed and air volume sensors placed above the cleaning screen; all of the first wind speed and air volume sensors are arranged in a linear array in the front-to-back direction; and all of the second wind speed and air volume sensors are arranged in a square array in the front-to-back and left-to-right directions.

[0011] Furthermore, a material breaking device is installed above the cleaning screen.

[0012] Furthermore, the material scattering device includes a drum, on which a plurality of shifting assemblies arranged in a circular array are fixed, the shifting assemblies include a plurality of material shifting rods dispersedly arranged in the axial direction of the drum, and the drum is driven to operate by a material scattering motor.

[0013] Furthermore, a collection box for collecting grains is installed below the frame.

[0014] An agricultural material cleaning test method, the method comprising the following steps S101-S105:

[0015] Step S101: Initialize the cleaning scheme and set the vibration frequency of the cleaning screen through the controller , the speed of the main fan and the speed of the front blower fan , and set feeding parameters based on the feeding amount adjustment device of the feeding device, the feeding parameters include the material placement position, the feeding amount ratio And delivery traffic ;

[0016] Step S102: Start the cleaning test bench, feed materials to the front end of the cleaning screen through the feeding device, collect wind speed and volume data above and below the cleaning screen in real time through the sensor group, and monitor the cleaning loss rate through the loss rate sensor and impurity content ;

[0017] Step S103: judging the material distribution uniformity and air flow penetration efficiency according to the wind speed and air volume data; when detecting that the material is accumulated on the screen surface or the air flow resistance is abnormal, starting the material breaking device and adjusting the rotation speed of the drum;

[0018] Step S104: Based on the preset multi-objective optimization algorithm, combined with the SVR proxy model and the MOPSO algorithm, the speed of the main fan is optimized. , the speed of the front blower fan , vibration screen frequency Dynamically optimize the feeding parameters, generate the Pareto optimal solution set, and select the operating parameter combination with the best comprehensive evaluation value;

[0019] Step S105: According to the optimized parameter combination, adjust each parameter in real time, and execute S102 to S104 in a loop until the cleaning loss rate and the impurity content are stabilized within the target range.

[0020] Furthermore, the execution process of the multi-objective optimization algorithm in step S104 includes steps S201 to S203:

[0021] Step S201: Constructing the cleaning loss rate and impurity content The dual-objective optimization model is mathematically expressed as: ;

[0022] Step S202: construct an SVR agent model, whose optimization objective function is:

[0023]

[0024] in: is the penalty factor, and is the slack variable, is the weight vector, is the bias term; is the total number of trials; here, each parameter is a local variable;

[0025] Step S203: Use the MOPSO algorithm to globally optimize the cleaning parameters The fitness function is: ; is the fitness, is the loss rate weight, is the impurity rate weight;

[0026] Update the particle velocity using the linearly decreasing inertia weight strategy:

[0027]

[0028] where, is the current iteration number, is the total number of iterations, and are the maximum and minimum values of the weight respectively.

[0029] Preferably, the method further includes the following steps S301 - S302:

[0030] Step S301, adjust the feeding ratio and the feeding flow rate through the feeding amount adjusting device of the feeding device to simulate the agglomerated piled materials; and collect in real time the data set composed of the data of all the second wind speed and air volume sensors , is the data of the second wind speed and air volume sensor in the th row and the th column, are the total number of rows and the total number of columns of the second wind speed and air volume sensors respectively; calculate the feature vector that reflects the imbalance characteristics of the data in the data set ;

[0031] Step S302, based on the SVR proxy model, use the feature vector and the corresponding material stacking height to form the data to train the prediction model, and are the feature vector and height data collected in the th trial respectively, is the total number of times;

[0032] During training, use the radial basis function kernel ; where is the kernel width parameter;

[0033] After training, the prediction model can be based on the actual collected data set. The corresponding eigenvector Output the predicted pile height.

[0034] Beneficial effects: The agricultural material cleaning test bench and cleaning test method of the present invention have the following beneficial effects:

[0035] (1) By redesigning the feeding device, precise adjustment of the material delivered to the screen surface is achieved. The delivery position, delivery amount and flow rate of the material can be adjusted according to different delivery plans. Not only can various feeding plans that are easy to clean be simulated, but also the delivery effect of material accumulation can be simulated. A variety of cleaning plans can be formed by combining operating parameters and fan speed. The controller automatically adjusts various parameters based on the loss rate data to improve the cleaning effect and achieve the purpose of optimizing the cleaning plan.

[0036] (2) It can simulate the placement of accumulated materials and verify the effect and parameters of the material dispersing device installed above the cleaning screen. The sensor group monitors the wind speed and volume data above and below the cleaning screen in real time, providing a basis for adjusting the working parameters of the dispersing device, further optimizing the dispersion of materials and the airflow penetration efficiency, and improving the cleaning quality and efficiency.

[0037] (3) Combining the SVR agent model and the MOPSO algorithm, the speed of the main fan and the front blower, the frequency of the vibrating screen and the feeding parameters are dynamically optimized to generate the Pareto optimal solution set, and the operating parameter combination with the best comprehensive evaluation value is selected. This intelligent dynamic adjustment mechanism not only enables the equipment to maintain efficient operation under different working conditions, but also reduces energy consumption and improves the automation level and adaptability of the entire system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 The first structural diagram of the agricultural material cleaning test bench;

[0039] Figure 2 The second structural diagram of the agricultural material cleaning test bench;

[0040] Figure 3 This is a side view of the agricultural material cleaning test bench;

[0041] Figure 4 It is the structural diagram of the feeding device;

[0042] Figure 5 Schematic diagram of the flow chart of the test method for agricultural material cleaning.

[0043] In the figure: 1-cleaning screen; 11-vibrating screen drive motor; 2-main fan; 3-front blowing fan; 4-feeding device; 41-feeding box; 41a-feeding grid; 42-feeding amount adjustment device; 421-outer frame; 422-baffle; 423-rotating shaft; 424-adjusting handle; 425-connecting rod; 426-connecting arm; 427-handle; 428-bend arm; 429-first spring; 4210-second spring; 4211-fixed handle; 5-loss rate sensor; 61-first wind speed and air volume sensor; 62-second wind speed and air volume sensor; 7-material scattering device; 71-drum; 72-feeding rod; 73-material scattering motor; 8-collecting box; 9-controller; 10-frame. DETAILED DESCRIPTION

[0044] The present invention will be further described below in conjunction with the accompanying drawings.

[0045] like Figures 1 to 3 The agricultural material cleaning test bench shown comprises a frame 10 with a cleaning chamber, on which a cleaning screen 1, a main fan 2, a front blowing fan 3, a feeding device 4, a loss rate sensor 5 and a controller 9 are installed; the cleaning screen 1 comprises a shaking plate, an upper fish scale screen, a finger screen, a lower woven screen and a vibrating screen drive motor 11, and the vibrating screen drive motor 11 can make all the screen bodies vibrate together; a sensor group for detecting wind speed and air volume is also installed on the frame 10; the front blowing fan 3 is used to provide a parallel airflow with high wind pressure to blow away lighter impurities before the material falls on the screen, and the main fan 2 is placed on the lower side of the cleaning screen 1 to provide an oblique upward and backward blowing airflow during the operation of the cleaning screen 1 to blow part of the impurities in the material on the screen surface out of the screen surface and assist the material on the screen surface to move backward; the controller 9 can adjust the rotation speed of the main fan 2 and the front blowing fan 3, as well as adjust the operating parameters of the cleaning screen 1, and can obtain data collected by each sensor;

[0046] The feeding device 4 includes a feeding box 41 arranged in an upper and lower manner and a feeding amount adjusting device 42. The feeding box 41 is connected from top to bottom and has a plurality of feeding grids 41a arranged in a square array inside. The feeding amount adjusting device 42 can adjust the amount or flow rate of materials fed by the feeding grid 41a to the front end of the cleaning screen 1.

[0047] In the above-mentioned cleaning test bench, by redesigning the feeding device 4, the materials fed onto the sieve surface can be precisely adjusted as required to conduct the cleaning test of the feeding scheme. The feeding scheme includes the feeding position (corresponding to the position of the feeding grid 41a), the distribution of the fed materials (the proportion of the feeding amount corresponding to each feeding grid 41a when materials are fed through multiple feeding grids 41a), the feeding amount, and the feeding flow rate. By adjusting the combination of the feeding scheme, operating parameters, and the fan speed to form a cleaning scheme, the controller 9 can judge the material cleaning effect of each cleaning scheme based on the loss rate data generated by the loss rate sensor 5, and adjust the operating parameters of the cleaning sieve 1 and the fan speed based on the loss rate to improve the cleaning effect and obtain an optimized cleaning scheme.

[0048] As Figure 4 shown, the feeding amount adjusting device 42 includes an outer frame 421, and further includes baffles 422 arranged corresponding to the lower part of each row of the feeding grids 41a. A rotating shaft 423 that rotates relative to the outer frame 421 is fixed on the baffle 422; an adjusting handle 424 is fixed on each rotating shaft 423, and all the adjusting handles 424 are connected to the same connecting rod 425; the feeding amount adjusting device 42 further includes an adjusting mechanism for changing the position of the connecting rod 425; the adjusting mechanism includes a connecting arm 426 connected to the connecting rod 425, and further includes a handle 427 rotatably installed relative to the outer frame 421. A bent arm 428 is rotatably installed on the handle 427. A first spring 429 is connected between the bent arm 428 and the connecting arm 426, and a second spring 4210 is connected between the connecting rod 425 and the outer frame 421. The pulling forces exerted on the connecting rod 425 by the first spring 429 and the second spring 4210 are in opposite directions. In addition, a fixing handle 4211 for fixing the adjusting handle 424 is provided on the outer frame 421, and the fixing handle 4211 can be fixed on the outer frame 421 by screws.

[0049] By adjusting the rotation angle of the fixing handle 4211 and fixing the fixing handle 4211 on the outer frame 421 with screws, the inclination angle of the baffle 422 can be adjusted, that is, the opening degree can be adjusted, so as to adjust the feeding flow rate of the materials.

[0050] The first spring 429 and the second spring 4210 form a dynamic balance through reverse pulling forces in the feeding amount adjusting device 42: The first spring 429 connects the bent arm 428 and the connecting arm 426, and is used to buffer the external force during the adjustment of the handle 427 and provide a reset function; the second spring 4210 connects the connecting rod 425 and the outer frame 421, and is used to stabilize the position of the connecting rod 425 and offset the vibration interference of the baffle 422. The two cooperate to ensure the accuracy of the opening degree adjustment of the baffle 422, make the adjustment of the feeding flow rate of the materials sensitive and lockable, avoid the opening degree deviation caused by vibration or external force, and ensure the controllability of the feeding scheme in the cleaning test.

[0051] The sensor group includes a plurality of first wind speed and air volume sensors 61 disposed below the cleaning screen 1, and also includes a plurality of second wind speed and air volume sensors 62 disposed above the cleaning screen 1; all of the first wind speed and air volume sensors 61 are arranged in a linear array in the front-to-back direction; and all of the second wind speed and air volume sensors 62 are arranged in a square array in the front-to-back and left-to-right directions.

[0052] The layout of wind speed and air volume sensors can comprehensively monitor the flow field parameters in the cleaning chamber. Specifically, multiple first wind speed and air volume sensors 61 can accurately capture the changes in airflow at different positions, especially the airflow characteristics on the material falling path, and provide a basis for adjusting the fan speed. The sensor array composed of the second wind speed and air volume sensors 62 not only covers a wider monitoring area, but also can effectively identify the changes in the lateral airflow (generated by the front blowing fan 3) and the longitudinal airflow, which helps to more accurately evaluate the airflow distribution during the entire cleaning process. This layout design allows the controller 9 to perform a comprehensive analysis based on the data collected at multiple points, and then adjust the speed of the main fan 2 and the front blowing fan 3 and the operating parameters of the cleaning screen 1, so as to achieve the purpose of optimizing the material cleaning effect and improve the cleaning efficiency and quality.

[0053] A material dispersing device 7 is also installed above the cleaning screen 1. By installing the material dispersing device 7 above the cleaning screen 1, the dispersion of the material is effectively improved, the material is prevented from agglomerating or piling up, and the material is ensured to be evenly distributed on the screen surface, so that the airflow can better penetrate the material layer, and the cleaning effect is further optimized. Through the wind speed and air volume data at each position above and below the cleaning screen 1 collected by the sensor group, the air volume distribution at each position on the screen surface of the cleaning screen 1 can be known to predict the material thickness at each position on the screen surface, and the working parameters of the dispersing device can be adjusted accordingly to improve the cleaning quality.

[0054] The material scattering device 7 includes a roller 71 , on which a plurality of shifting assemblies arranged in a circular array are fixed, the shifting assemblies including a plurality of material shifting rods 72 dispersedly arranged in the axial direction of the roller 71 , and the roller 71 is driven to operate by a material scattering motor 73 .

[0055] In practical applications, the material dispersion device 7 can work in coordination with the sensor group through the controller 9 to achieve dynamic adjustment. When the feeding device 4 delivers materials to the cleaning sieve 1, the controller 9 analyzes the accumulation degree of materials on the sieve surface and the air flow penetration efficiency based on the air flow distribution data collected by the first air velocity and air volume sensor 61 and the second air velocity and air volume sensor 62, combined with the real-time feedback of the loss rate sensor 5. If it is detected that the material distribution is uneven or the air flow resistance is abnormal, the controller 9 can automatically start the material dispersion motor 73 to drive the roller 71 to rotate at an appropriate speed, driving the axially distributed material guiding rods 72 to flexibly disperse the falling materials. When it is detected that the material quantity suddenly increases, the finger sieve length can be synchronously extended and the rotation speed of the dispersion device can be increased through the linkage mechanism to ensure the uniform distribution of materials under the condition of large flow rate. This technical solution makes the spreading thickness of materials on the sieve surface tend to be uniform, and at the same time reduces the no-load loss of the equipment through dynamic adjustment of energy consumption, providing reliable process parameter support for optimizing the cleaning scheme.

[0056] Below the frame 10, an aggregate box 8 for collecting grains is installed. During operation, grains of known mass are mixed with impurities such as light impurities and straw as cleaning raw materials. Based on the mass of the grains finally obtained in the aggregate box 8 and the mass of the original grains, the data of the loss rate can be obtained.

[0057] An agricultural material cleaning test method, based on the above-mentioned agricultural material cleaning test bench, as Figure 5 shown, the method includes the following steps S101 - S105:

[0058] Step S101: Initialize the cleaning scheme, set the vibration frequency of the cleaning sieve 1 , the rotation speed of the main fan 2 and the rotation speed of the pre-blowing fan 3 through the controller 9, and set the feeding parameters based on the feeding amount adjusting device 42 of the feeding device 4. The feeding parameters include the feeding position, feeding amount ratio and feeding flow rate ;

[0059] Step S102: Start the cleaning test bench, deliver materials to the head end of the cleaning sieve 1 through the feeding device 4, and at the same time, collect the air velocity and air volume data above and below the cleaning sieve 1 in real time through the sensor group, and monitor the cleaning loss rate and impurity content rate through the loss rate sensor 5;

[0060] Step S103: Judge the material distribution uniformity and air flow penetration efficiency according to the air velocity and air volume data; when it is detected that there is material accumulation on the sieve surface or the air flow resistance is abnormal, start the material dispersion device 7 and adjust the rotation speed of the roller 71;

[0061] Step S104: Based on the preset multi-objective optimization algorithm, combining the SVR surrogate model and the MOPSO algorithm, optimize the rotational speed of the main fan 2 , the rotational speed of the pre-blowing fan 3 , the vibrating screen frequency and the feeding parameters dynamically, generate the Pareto optimal solution set, and select the operation parameter combination with the optimal comprehensive evaluation value;

[0062] Step S105: According to the optimized parameter combination, adjust each parameter in real time, and loop to execute S102 to S104 until the cleaning loss rate and impurity content rate are stabilized within the target range.

[0063] The execution process of the multi-objective optimization algorithm in the step S104 includes steps S201 - S203:

[0064] Step S201: Construct a double-objective optimization model of the cleaning loss rate and the impurity content rate , and its mathematical expression is: ;

[0065] Step S202: Construct an SVR surrogate model, and its optimization objective function is:

[0066]

[0067] Where: is the penalty factor, and are slack variables, is the weight vector, is the bias term; each parameter here is a local variable;

[0068] Step S203: Perform global optimization on the cleaning parameters through the MOPSO algorithm, and its fitness function is: ; is the fitness, is the loss rate weight, is the impurity content rate weight;

[0069] Update the particle velocity using the linearly decreasing inertia weight strategy:

[0070]

[0071] Where, is the current iteration number, is the total iteration number, and are the maximum and minimum values of the weight respectively.

[0072] Preferably, the method further includes the following steps S301-S302:

[0073] Step S301: Adjust the feeding ratio and feeding flow rate through the feeding amount adjusting device 42 of the feeding device 4 to simulate the piled-up bulk materials; and collect in real time the data set composed of all the data of the second air velocity and air volume sensors 62. And the feeding flow rate to simulate the piled-up bulk materials; and collect in real time the data set composed of all the data of the second air velocity and air volume sensors 62. Let the data of the second air velocity and air volume sensor 62 in the i-th row and j-th column be , be the i-th row and j-th column of the second air velocity and air volume sensor 62. Calculate the imbalance characteristics of the data, including the global standard deviation , local maximum deviation , and sub-region maximum variance to construct a feature vector ; where:

[0074] The global standard deviation ;

[0075] The average air velocity ;

[0076] The local maximum deviation: ;

[0077] Step S302: Based on the SVR proxy model, use the collected feature vector and the corresponding material stacking height to form data to train a prediction model. Let and be the feature vector and height data collected in the i-th experiment respectively. The optimization objective is;

[0078] ;

[0079] where is the weight vector, is the bias term, is the penalty factor, and are the slack variables; is the total number of experiments; here all parameters are local variables;

[0080] Adopt the radial basis function kernel ; where is the kernel width parameter; after training, the prediction model can output the predicted stacking height corresponding to the feature vector of the actually collected data set ;

[0081] In step S103 and subsequent actual operations, the controller 9 obtains in real time a data set composed of the data of all the second air velocity and air volume sensors 62 , and calculates the feature vector accordingly , and obtains the predicted stacking height of the SVR model ; according to a preset threshold , adjust the rotation speed of the roller 71 through the following control strategy :

[0082] ;

[0083] Wherein: is the proportionality coefficient, is the stacking height threshold, and the specific values of both are determined through experiments;

[0084] Subsequently, the controller drives the material dispersion motor 73 to operate according to the calculated rotation speed , and continuously monitors the data set composed of the data of all the second air velocity and air volume sensors 62 , and can realize the test of the effect of the dispersed material

[0085] The above method effectively solves the problems of low processing efficiency and uneven quality caused by material agglomeration and stacking in the agricultural material cleaning process by intelligently regulating the rotation speed of the roller of the material dispersion device. Using the support vector regression model to predict the material stacking height and dynamically adjusting the roller rotation speed based on real-time monitoring data not only improves the uniformity of material dispersion, but also significantly enhances the stability and working efficiency of the equipment operation. This method realizes the precise control of the material feeding amount and the air velocity and air volume, ensures the best processing state of the material in the cleaning process, and further enhances the automation level and adaptability of the entire system, providing a reliable technical guarantee for agricultural production

[0086] The above are only the preferred embodiments of the present invention. It should be noted that: for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention

Claims

1. An agricultural material cleaning test bench, comprising a frame with a cleaning chamber, on which a cleaning screen, a main fan, a front blowing fan, a feeding device, a loss rate sensor and a controller are installed; the cleaning screen comprises a shaking plate, an upper fish scale screen, a finger screen, a lower woven screen and a vibrating screen drive motor; a sensor group for detecting wind speed and air volume is also installed on the frame; the controller can adjust the speed of the main fan and the front blowing fan, as well as the operating parameters of the cleaning screen, and can obtain data collected by each sensor; it is characterized in that: The feeding device includes a feeding box arranged in an upper and lower manner and a feeding amount adjusting device. The feeding box is connected from top to bottom and has a plurality of feeding grids arranged in a square array inside. The feeding amount adjusting device can adjust the amount or flow rate of materials released by the feeding grid to the head end of the cleaning screen.

2. The agricultural material cleaning test bench according to claim 1, wherein The feeding amount adjustment device includes an outer frame, and also includes a baffle plate corresponding to each row of the feeding grids, and a rotating shaft rotating relative to the outer frame is fixed on the baffle plate; an adjustment handle is fixed on each of the rotating shafts, and all the adjustment handles are connected to the same connecting rod; The feeding amount adjustment device also includes an adjustment mechanism for changing the position of the connecting rod; the adjustment mechanism includes a connecting arm connected to the connecting rod, and also includes a handle rotatably installed relative to the outer frame, a bent arm is rotatably installed on the handle, a first spring is connected between the bent arm and the connecting arm, and a second spring is connected between the connecting rod and the outer frame.

3. The agricultural material cleaning test bench according to claim 1, characterized in that, The sensor group includes a plurality of first wind speed and air volume sensors placed below the cleaning screen, and also includes a plurality of second wind speed and air volume sensors placed above the cleaning screen; all of the first wind speed and air volume sensors are arranged in a linear array in the front-to-back direction; all of the second wind speed and air volume sensors are arranged in a square array in the front-to-back and left-to-right directions.

4. The agricultural material cleaning test bench according to claim 3, characterized in that, A material breaking device is also installed above the cleaning screen.

5. The agricultural material cleaning test bench according to claim 4, characterized in that, The material scattering device comprises a drum, on which a plurality of groups of shifting assemblies arranged in a circular array are fixed, the shifting assemblies comprising a plurality of material shifting rods dispersedly arranged in the axial direction of the drum, and the drum is driven to operate by a material scattering motor.

6. The agricultural material cleaning test bench according to claim 1, characterized in that, A collecting box for collecting grains is installed below the frame.

7. An agricultural material cleaning test method, which is based on the agricultural material cleaning test bench described in claim 5, and is characterized in that, The method comprises the following steps S101-S105: Step S101: Initialize the cleaning solution, set the vibration frequency of the cleaning sieve through the controller , the rotational speed of the main blower and the rotational speed of the pre-blowing blower , and set the feeding parameters based on the feeding amount adjusting device of the feeding device, where the feeding parameters include the feeding position, feeding amount ratio and the feeding flow rate ; Step S102: Start the cleaning test bench, feed materials to the head end of the cleaning screen through the feeding device, and at the same time, collect the wind speed and air volume data above and below the cleaning screen in real time through the sensor group, and monitor the cleaning loss rate through the loss rate sensor and impurity content ; Step S103: judging the material distribution uniformity and air flow penetration efficiency according to the wind speed and air volume data; when detecting that the material is accumulated on the screen surface or the air flow resistance is abnormal, starting the material breaking device and adjusting the rotation speed of the drum; Step S104: Based on a preset multi-objective optimization algorithm, combining the SVR surrogate model and the MOPSO algorithm, dynamically optimize the rotational speed of the main fan , the rotational speed of the pre-blowing fan , the vibrating screen frequency and the feeding parameters, generate a Pareto optimal solution set, and select the operation parameter combination with the optimal comprehensive evaluation value; Step S105: According to the optimized parameter combination, adjust each parameter in real time, and execute S102 to S104 in a loop until the cleaning loss rate and the impurity content are stabilized within the target range.

8. The agricultural material cleaning test method according to claim 7, wherein The execution process of the multi-objective optimization algorithm in step S104 includes steps S201 to S203: Step S201: Construct a dual-objective optimization model for the cleaning loss rate and the impurity content rate , and its mathematical expression is as follows: ; Step S202: construct an SVR agent model, whose optimization objective function is: Wherein: is the penalty factor, and are slack variables, is the weight vector, is the bias term; is the total number of trials; Step S203: Use the MOPSO algorithm to perform global optimization on the cleaning parameters The fitness function is as follows: ; is the fitness, is the loss rate weight, is the impurity rate weight; Update particle velocity using a linearly decreasing inertia weight strategy: Among them, is the current iteration number, is the total number of iterations, and are the maximum and minimum values of the weights, respectively.

9. The agricultural material cleaning test method according to claim 7, characterized in that, The method further comprises the following steps S301-S302: Step S301, adjust the feeding ratio and the feeding flow rate through the feeding amount adjusting device of the feeding device and the feeding flow rate , simulate the agglomerated and piled materials; and collect in real time the data set composed of the data of all the second wind speed and air volume sensors , For the data of the second wind speed and air volume sensor in the row and the column, where are the total number of rows and the total number of columns of the second wind speed and air volume sensors respectively; calculate the eigenvector reflecting the unbalanced characteristics of the data in the data set Step S302: Based on the SVR proxy model, using the feature vector and the corresponding material stacking height to form data to train a prediction model, and are the feature vector and height data collected in the th experiment respectively, being the total number of times. During training, a radial basis function kernel is adopted ; where is the kernel width parameter; After training is completed, the prediction model can be based on the actually collected data set corresponding feature vectors to output the predicted stacking height.

Images