A lever-type lysimeter weight detection method that can suppress soil box tilting and vibration.

Abstract

This invention discloses a lever-type lysimeter weight detection method that can suppress soil box tilting and vibration, the steps of which are as follows: Step 1, install on the lever-type weighing lysimeter x Axis acceleration sensor, y The method comprises: a axial acceleration sensor and two-axis tilt sensors; step 2, calibrating the lever-type weighing lysimeter under different operating conditions using standard weights; step 3, establishing a weighing compensation model and training it to obtain a trained weighing compensation model; step 4, obtaining the final weighing measurement result based on the trained weighing compensation model. This invention can suppress the influence of tilt and wind-induced vibrations on the weighing measurement, greatly improving the accuracy of the lysimeter's weight detection results.

Description

Technical Field

[0001] This invention belongs to the technical field of lysimeter weight detection methods, specifically relating to a lever-type lysimeter weight detection method that can suppress soil box tilting and vibration. Background Technology

[0002] Evaporation and transpiration dominate the exchange of matter and energy between land and atmosphere in an ecosystem and are an important part of the surface water cycle. Measuring evaporation and transpiration is of practical significance for formulating scientific agricultural water use plans and developing water-saving agriculture.

[0003] Among existing methods for measuring evapotranspiration, the lever-type lysimeter is the most effective and is often used as a comparative reference for other methods. The soil box is the weighing object of the lever-type lysimeter. The soil box typically consists of a steel structure container, the soil it holds, and plants planted on its surface, with a rigid connection between the bottom of the soil box and the lever. Ideally, the direction of soil box weight measurement should perfectly coincide with the direction of gravity. A high-precision electromagnetic torque load cell is commonly used to detect the weight of the soil box, thereby determining evapotranspiration by monitoring changes in the soil box weight. Because the soil in the soil box undergoes irrigation, planting, rainfall, cultivation, and human activity, the mass distribution of the soil box will change significantly, easily leading to tilting. When the soil box is tilted, the direction of weight measurement no longer coincides with the direction of the soil box's center of gravity. Without compensation, the measurement result will be underestimated, and the larger the tilt angle, the greater the error. Furthermore, lysimeters are usually placed in outdoor fields, where the soil box will vibrate under the combined effects of natural wind and plants, and the amplitude of the vibration varies with the wind force. This vibration introduces an additional force unrelated to evaporation and transpiration in the direction of weight measurement. Since the actual weight of the soil box is unknown, the magnitude of this additional force is also unknown. For lever-type lysimeters, the additional force caused by tilting and wind-induced vibration causes the weight measurement result to deviate from the true value, thus introducing a significant error in evaporation and transpiration measurement, resulting in a severe decrease in the accuracy of the lysimeter. As water-saving agriculture increasingly demands a clear understanding of water consumption patterns throughout the crop growth cycle, the need for weight measurement using lever-type lysimeters becomes more urgent. Therefore, proposing a weight measurement method that can suppress soil box tilting and vibration to improve the accuracy of lysimeter weight measurement and meet the needs of water-saving agriculture is a pressing issue. Summary of the Invention

[0004] The purpose of this invention is to provide a lever-type lysimeter weight detection method that can suppress soil box tilting and vibration, thereby solving the problem of low accuracy of weight detection results in existing methods.

[0005] The technical solution adopted in this invention is a lever-type lysimeter weight detection method that can suppress soil box tilting and vibration, the steps of which are as follows: Step 1, install on the lever-type weighing lysimeter x Axis acceleration sensor,y Axial acceleration sensor, dual-axis tilt sensor; Step 2: Calibrate the lever-type weighing lysimeter under different operating conditions using standard weights; Step 3: Establish a weighing compensation model and train the weighing compensation model to obtain a trained weighing compensation model. Step 4: Obtain the final weighing measurement result based on the trained weighing compensation model.

[0006] The invention is further characterized by: In step 1, the lever-type weighing lysimeter includes a cylindrical soil box. A lever fixing point is located at the center of the bottom surface of the soil box. Along the circumference of the bottom surface of the soil box, a first elastic support point, a second elastic support point, a third elastic support point, and a fourth elastic support point are sequentially arranged. Each of these four elastic support points is connected to an elastic support. The end of each elastic support furthest from the soil box is connected to a support surface. A lever fixing block is located at the bottom of the lever fixing point. A lever connection point is located at the bottom of the smooth rod fixing block, connecting to one end of the lever. A weighing sensor is connected to the other end of the lever. A lever support block is connected to the bottom of the lever near the lever connection point, serving as the lever fulcrum. The bottom of the lever support block is connected to the support surface. A counterweight is connected to the lever on the side wall near the weighing sensor. x Axis acceleration sensor, y The axial acceleration sensor and the two-axis tilt sensors are installed on the upper part of the outer surface of the soil box, with the two-axis tilt sensors located at... x Directly below the axial accelerometer sensor, y The shaft acceleration sensor is located at x The axial acceleration sensor is positioned between the two axial tilt sensors and on the side closest to the load cell.

[0007] The line connecting the first, second, third, and fourth elastic support points forms a square, and the intersection of the two diagonals of the square coincides with the lever fixing point.

[0008] In step 2, the operating conditions include no tilt and no wind, tilt and wind.

[0009] Condition 1: The calibration experiment under no tilt and no wind conditions is as follows: Choose a windless day, level the soil box, and then adjust the counterweight so that the load cell output is at half full scale. Then, using standard weights, load the soil box from 0 kg to [amount missing] in increments of ΔF. kg, then from The load was reduced from 0 kg to 1 kg, and the output of the weighing sensor was recorded for each loading and unloading operation. and the corresponding weight of the weights , Indicates the calibration serial number. This represents the total number of calibration points. The input-output working curve of the lever-type weighing lysimeter during weighing can be expressed by the formula: (1) In equation (1), A coefficient related to leverage; The offset related to the counterweight; This is the input quantity used when weighing in a lever-type gravimetric lysimeter. This is the output quantity of the lever-type weighing quasi-percolator during weighing; The output of the weighing sensor As the input quantity when weighing in a lever-type gravimetric diatomometer, the corresponding weight is... As the output of the lever-type weighing lysimeter, the input-output working curve of the lever-type weighing lysimeter is fitted using the least squares algorithm. Then, in formula (1) It can be obtained by formula (2). It can be obtained through formula (3), that is: (2) (3).

[0010] Operating Condition 2: The calibration test procedure under inclined and windy conditions is as follows: Step S1, set the calibration experimental conditions: On a windless day, the tilt angle of the soil box was adjusted by changing the position of the standard weights on the surface of the soil box. The tilt angle was measured by a two-axis tilt sensor. An adjustable speed industrial fan was used to simulate the natural wind blowing across the surface of the soil box. The wind direction was determined based on the dominant wind direction in the historical agricultural meteorological data of the area. The wind speed was measured by a wind power sensor, and the wind direction was measured by a wind direction sensor. The tilt includes eight scenarios: the first scenario is tilting towards the first elastic support point; the second scenario is tilting towards the second elastic support point; the third scenario is tilting towards the third elastic support point; the fourth scenario is tilting towards the fourth elastic support point; the fifth scenario is tilting towards both the first and second elastic support points simultaneously; the sixth scenario is tilting towards both the second and third elastic support points simultaneously; the seventh scenario is tilting towards both the third and fourth elastic support points simultaneously; and the eighth scenario is tilting towards both the fourth and first elastic support points simultaneously. The array Dip_case[P] is used to represent the P types of tilt. The tilt angle range is set to 0°~15°. M different tilt angles are selected within the tilt angle range for calibration. The array Dip_angle[M] is used to represent the M different tilt angles. Based on the degree of influence of wind speed on agricultural planting, the wind speed range is determined to be 0m / s to 8m / s. Within the wind speed range, N different wind speeds are selected for calibration, and the array Wind[N] is used to represent the N different wind speeds. Step S2, first let , Indicates the first The first type of tilt is selected. Step S3, determine If yes, proceed to step S5; otherwise, proceed to step S4. Step S4, end calibration; Step S5, set the tilt case Dip_case[ l ],initialization ; Step S6, determine , Indicates the first If the tilt angle is yes, proceed to step S8; otherwise, proceed to step S7. Step S7, This indicates that a tilting condition has been selected, and step S3 is executed; Step S8, set the tilt angle Dip_angle[ m ],initialization ; Step S9, determine , Indicates the first If the wind speed is specified, proceed to step S11; otherwise, proceed to step S10. Step S10, This indicates that the next tilt angle should be selected, and step S6 should be executed; Step S11, set the wind speed Wind[ n ]; Step S12: Using standard weights, load from 0 kg to [amount missing] kg in increments of ΔF. kg, number of calibration points After completion, unload all weights; Step S13, This indicates that the next wind speed should be selected, and step S9 should be executed. Step S14, record the results of each experiment. Axis tilt angle, Axis tilt angle, Axial acceleration, The axial acceleration, wind speed measured by the wind speed sensor, weight of the weight, and output of the weighing sensor.

[0011] Step 3 is as follows: Step 3.1: Construct the dataset and normalize the data in the dataset, then divide the normalized dataset into a training set and a test set. Step 3.2: Train the weighing compensation model using the training set to obtain the trained weighing compensation model. Test the trained weighing compensation model using the test set. If the test results meet the requirements, the trained weighing compensation model is obtained. If the test results do not meet the requirements, retraining is performed. The loss function used during training is: (8) In equation (8), k Indicates the current training sample number. , This represents the total number of training samples. The true values ​​of the current training samples. This is the output value of the weighing compensation model.

[0012] In step 3.1, the dataset is... Serial Number Total number of samples ,in Let be the input vector, where for x Axis tilt angle, for y Axis tilt angle, for x Axial acceleration, for y Axial acceleration, This is the output of the weighing sensor. The output value is the output data of the weighing sensor corresponding to the weights used in the calibration experiment under condition 2 in condition 1.

[0013] In step 3.2, the weighing compensation model uses a backpropagation (BP) neural network. The BP neural network has a 3-layer structure: 5 neurons in the input layer, 1 neuron in the output layer, and [number missing] neurons in the hidden layer. .

[0014] The specific process of step 4 is as follows: When the lever-type weighing lysimeter performs the weighing measurement task, it simultaneously collects data. Axis tilt angle, Axis tilt angle, Axial acceleration, The axial acceleration and the output data of the weighing sensor are normalized. The normalized data is input into the trained weighing compensation model. The output of the trained weighing compensation model is inversely normalized to obtain the inversely normalized weighing sensor output. The inversely normalized weighing sensor output is substituted into formula (1) to obtain the final weighing measurement result.

[0015] The beneficial effects of this invention are as follows: This invention provides a lever-type lysimeter weight detection method that can suppress soil box tilting and vibration. Taking into account the possible tilting of the soil box and the characteristics of natural wind, a two-axis tilt sensor and a two-axis acceleration sensor are installed on the surface of the soil box. The lysimeter is then calibrated within the allowable tilt angle and wind speed range. A compensation model based on the calibration data is established using a backpropagation neural network, thereby effectively suppressing the influence of tilting and wind-induced vibrations on lysimeter weight measurement, greatly improving the accuracy of the lysimeter weight detection results, and demonstrating high practicality. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the lever-type weighing lysimeter in the lever-type lysimeter weight detection method of the present invention, which can suppress soil box tilting and vibration. Figure 2 This is a bottom view of the soil box in the lever-type lysimeter weight detection method of the present invention, which can suppress soil box tilting and vibration. Figure 3 This is a calibration flowchart for working condition 2 in the lever-type lysimeter weight detection method of the present invention, which can suppress soil box tilting and vibration. Figure 4 This invention relates to a lever-type lysimeter weight detection method for suppressing soil box tilting and vibration, in which the soil box is wound around... x (or y ) Schematic diagram of axis tilt; Figure 5 In the lever-type lysimeter weight detection method of this invention, which can suppress soil box tilting and vibration, the soil box simultaneously rotates... x and y Schematic diagram of axis tilt; Figure 6 This invention relates to a lever-type lysimeter weight detection method for suppressing soil box tilting and vibration, in which the soil box is wound around... y When the axis is tilted along x Schematic diagram of the additional force on the shaft.

[0017] In the diagram, 1. First elastic support point, 2. Second elastic support point, 3. Third elastic support point, 4. Fourth elastic support point, 5. Lever fixing point, 6. x 7. Axis accelerometer. y8. Axis acceleration sensor, 9. Two-axis tilt sensor, 10. Soil box, 11. Elastic support, 12. Lever connection point, 13. Lever fulcrum, 14. Counterweight, 15. Lever, 16. Weighing sensor, 17. Support surface, 18. Lever fixing block, 19. Lever support block. Detailed Implementation

[0018] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0019] Example 1 The following are the steps for a lever-type lysimeter weight detection method that can suppress soil box tilting and vibration: Step 1, install on the lever-type weighing lysimeter x 6. Axis acceleration sensor y 7. Axial acceleration sensor; 8. Two-axis tilt sensor; Step 2: Calibrate the lever-type weighing lysimeter under different operating conditions using standard weights; Step 3: Establish a weighing compensation model and train the weighing compensation model to obtain a trained weighing compensation model. Step 4: Obtain the final weighing measurement result based on the trained weighing compensation model.

[0020] Example 2 Based on Example 1, in step 1, as follows: Figure 1 As shown, the lever-type weighing lysimeter includes a cylindrical soil box 9. A lever fixing point 5 is located at the center of the bottom surface of the soil box 9. Along the circumference of the bottom surface of the soil box 9, a first elastic support point 1, a second elastic support point 2, a third elastic support point 3, and a fourth elastic support point 4 are sequentially arranged. The line connecting the first elastic support point 1, the second elastic support point 2, the third elastic support point 3, and the fourth elastic support point 4 forms a square, and the intersection of the two diagonals of the square coincides with the lever fixing point 5. Each of the first elastic support point 1, the second elastic support point 2, the third elastic support point 3, and the fourth elastic support point 4 is connected to an elastic support 10. Each elastic support 10 has its end away from the soil box 9 connected to the support surface 16. A lever fixing block 17 is provided at the bottom of the lever fixing point 5. A lever connection point 11 is provided at the bottom of the smooth rod fixing block 17. The lever connection point 11 is connected to one end of the lever 14. A weighing sensor 15 is connected to the other end of the lever 14. A lever support block 18 is connected to the bottom of the lever 14 near the lever connection point 11. The connection point between the lever support block 18 and the lever 14 serves as the lever fulcrum 12. The bottom of the lever support block 18 is connected to the support surface 16. A counterweight 13 is connected to the side wall of the lever 14 near the weighing sensor 15. Figure 2 As shown, x 6. Axis acceleration sensor yThe axial acceleration sensor 7 and the two-axis tilt sensor 8 are installed on the upper part of the outer surface of the soil box 9. The two-axis tilt sensor 8 is located at... x Directly below axis accelerometer 6, y The shaft acceleration sensor 7 is located in x The axial acceleration sensor 6 is positioned between the two axial tilt sensors 8 and on the side close to the load cell 15.

[0021] Example 3 Based on Example 2, in step 2, the operating conditions include no tilt and no wind, tilt and wind. Condition 1: The calibration experiment under no tilt and no wind conditions is as follows: Choose a windless day, adjust the soil box 9 to be level, then adjust the counterweight 13 so that the output of the weighing sensor 15 is at 1 / 2 of its full range. Then, using standard weights, load the soil box 9 from 0 kg to [amount missing] in increments of ΔF. kg, then from Unload the load from 0 kg to 0 kg, and record the output of the weighing sensor 15 during each loading and unloading cycle. and the corresponding weight of the weights , Indicates the calibration serial number. This represents the total number of calibration points. The input-output working curve of the lever-type weighing lysimeter during weighing can be expressed by formula (1), that is: (1) In equation (1), The coefficient related to lever 14; The bias amount related to the counterweight 13; This is the input quantity used when weighing in a lever-type gravimetric lysimeter. This is the output quantity of the lever-type weighing quasi-percolator during weighing; The output of the weighing sensor 15 As the input quantity when weighing in a lever-type gravimetric diatomometer, the corresponding weight is... As the output of the lever-type weighing lysimeter, the input-output working curve of the lever-type weighing lysimeter is fitted using the least squares algorithm. Then, in formula (1) It can be obtained by formula (2). It can be obtained through formula (3), that is: (2) (3) Operating Condition 2: The calibration test procedure under inclined and windy conditions is as follows: Step S1, set the calibration experimental conditions: On a windless day, the tilt angle of the soil box 9 is adjusted by changing the position of the standard weights on the surface of the soil box 9. The tilt angle is measured by a two-axis tilt sensor 8. An adjustable speed industrial fan is used to simulate the natural wind blowing across the surface of the soil box 9. The wind direction is determined based on the dominant wind direction in the historical agricultural meteorological data of the area. The wind speed is measured by a wind power sensor, and the wind direction is measured by a wind direction sensor. The tilt includes eight scenarios: the first is tilting towards the first elastic support point 1; the second is tilting towards the second elastic support point 2; the third is tilting towards the third elastic support point 3; the fourth is tilting towards the fourth elastic support point 4; the fifth is tilting towards both the first and second elastic support points 1 and 2 simultaneously; the sixth is tilting towards both the second and third elastic support points 2 and 3 simultaneously; the seventh is tilting towards both the third and fourth elastic support points 3 and 4 simultaneously; and the eighth is tilting towards both the fourth and first elastic support points 4 simultaneously. An array Dip_case[P] is used to represent the P types of tilts (P=8). When the tilt angle of the soil box 9 exceeds 20°, the mechanical structure jams, causing the output of the weighing sensor 15 to be zero. Therefore, the tilt angle range is determined to be 0°~15°. Within this range, M different tilt angles are selected for calibration, and an array Dip_angle[M] is used to represent the M different tilt angles. Based on the degree of impact of wind speed on agricultural planting, the wind speed range is determined to be 0m / s to 8m / s. This wind speed range covers low-speed wind, medium-speed wind and high-speed wind. N different wind speeds are selected within the wind speed range for calibration, and the array Wind[N] is used to represent N different wind speeds. Step S2, first let , Indicates the first The first type of tilt is selected. Step S3, determine If yes, proceed to step S5; otherwise, proceed to step S4. Step S4, end calibration; Step S5, set the tilt case Dip_case[ l ],initialization ; Step S6, determine , Indicates the first If the tilt angle is yes, proceed to step S8; otherwise, proceed to step S7. Step S7, This indicates that a tilting condition has been selected, and step S3 is executed; Step S8, set the tilt angle Dip_angle[ m ],initialization ; Step S9, determine , Indicates the first If the wind speed is specified, proceed to step S11; otherwise, proceed to step S10. Step S10, This indicates that the next tilt angle should be selected, and step S6 should be executed; Step S11, set the wind speed Wind[ n ]; Step S12: Using standard weights, load from 0 kg to [amount missing] kg in increments of ΔF. kg, number of calibration points After completion, unload all weights; Step S13, This indicates that the next wind speed should be selected, and step S9 should be executed. Step S14, record the results of each experiment. Axis tilt angle, Axis tilt angle, Axial acceleration, The axial acceleration, wind speed measured by the wind speed sensor, weight of the weight, and output of the weighing sensor can obtain 8*M*N*T sets of data.

[0022] Example 4 Based on Example 3, the specific process of step 3 is as follows: Step 3.1: Construct the dataset and normalize the data in the dataset, then divide the normalized dataset into a training set and a test set. Let the dataset be Serial Number Total number of samples ,in Let be the input vector, where for x Axis tilt angle, for y Axis tilt angle, for x Axial acceleration, for y Axial acceleration, This is the output of the weighing sensor. The output value is the output data of the weighing sensor corresponding to the weight used in the calibration experiment of working condition 2 in working condition 1. For dataset The expression for normalizing each data variable in the data is: (4) In equation (4), This represents the current value of the variable in the dataset. For data centralization and The minimum value of the corresponding variable. Data centralization and The maximum value of the corresponding variable, The data is the normalized version of the variables; make The input vector is the normalized form. The normalized output is the dataset used for training the neural network. , The ratio of training set to test set is 2:1; Step 3.2: Train the weighing compensation model using the training set to obtain the trained weighing compensation model. Test the trained weighing compensation model using the test set. If the test results meet the requirements, the trained weighing compensation model is obtained. If the test results do not meet the requirements, retraining is performed. The weighing compensation model uses a backpropagation (BP) neural network. The BP neural network has a 3-layer structure: 5 neurons in the input layer, 1 neuron in the output layer, and [number missing] neurons in the hidden layers. ; make , For the input layer The input of each neuron, Indicates the current moment. Represents input layer neurons To hidden layer neurons The weights of the hidden layer are then... Input of each neuron Represented as: (5) Then the hidden layer j The output of each neuron for: (6) Define hidden layer neurons The weights to the output layer neurons are The output of the weighing compensation model is then... Represented as: (7) Define the current training sample number , This represents the total number of training samples. The true values ​​of the current training samples. The loss function is used to evaluate the total training error for the output value of the weighing compensation model. Refer to the following formula: (8) Define momentum factor learning rate The weight update algorithm for the weighing compensation model is as follows: (9) (10).

[0023] Example 5 Based on Example 4, the specific process of step 4 is as follows: When the lever-type weighing lysimeter performs the weighing measurement task, it simultaneously collects data. Axis tilt angle, Axis tilt angle, Axial acceleration, The shaft acceleration and the output data from the weighing sensor 15 are normalized. The normalized data is then input into the trained weighing compensation model, and the output is... ,Will After inverse normalization, the inverse normalized load cell output is obtained. ,Will Substituting into formula (1) yields the final weighing measurement result. , Used for subsequent evaporation and transpiration calculations; in, The expression after inverse normalization is:

[0024] In the formula, and Each dataset Medium variables The maximum and minimum values; Final weighing measurement results The expression is: (12).

[0025] The weight detection method of the lever-type lysimeter in this invention, which can suppress soil box tilting and vibration, uses the data recorded in the calibration experiment under working condition 2. Axis tilt angle, Axis tilt angle, Axial acceleration, The shaft acceleration and the output of the load cell are used as the input vectors of the weighing compensation model. The output of the load cell corresponding to the calibration experiment under working condition 1 is used as the output variable of the weighing compensation model. The principle that can obtain weighing measurement results that eliminate or reduce the effects of tilt angle and wind is as follows: The lever-type gravimetric lysimeter is currently recognized as the most accurate method for measuring evaporation and transpiration. ) or evaporation ( The method involves using a lysimeter to control the weighing sensor 15 to measure the mass of the soil box 9 at different times according to a certain sampling cycle, and then calculating the mass of the soil box 9 within the measured time period according to the water balance equation. or The water balance equation is defined as follows: (13) (14) In equations (13) and (14), The sampling interval is... and They are respectively Internal evaporation and transpiration, in mm; and They are respectively Irrigation and rainfall amounts within the area, in mm; and They are respectively The amount of groundwater exchange and surface runoff within the area, in mm; For lysimeter in Time and Constant water level difference.

[0026] From equations (13) and (14), we can see that: (15) Let the cross-sectional area of ​​the soil column be Then the mass of the soil column It can be represented as: (16) Among them, the mass of the soil column The result is obtained by converting the output of the weighing sensor into formula (1) under the condition that the soil column is neither tilted nor disturbed by wind. It should be noted that... It refers to the mass within the weighing range of the lever-type elution meter, not the absolute mass of the soil column.

[0027] The effect of the tilt of the soil column on the symmetry of gravity can be explained in two cases: (1) Earth column around or Axis tilt Establish a three-dimensional rectangular coordinate system along the surface of the soil column. When the soil column rotates... Axis tilt At an angle, such as Figure 4 As shown in the figure , and The coordinate system before tilting. , and The coordinate system is tilted. , For the weight of the soil column, For gravitational acceleration, in units Due to the structural design of the lysimeter, the weighing measurement direction also changes with the inclination of the soil column. Axis tilt Angle, such as Figure 4 As shown, the actual force to be measured at this time is Obviously For soil columns around Axis tilt At an angle, the situation is the same as when turning around. The shaft analysis results are consistent, only... .

[0028] (2) The soil column is surrounded at the same time and Axis tilt Establish a three-dimensional rectangular coordinate system along the surface of the soil column, when the soil column simultaneously rotates around... shaft and When the axis tilts, the weight of the soil column Will no longer be with shaft and The plane containing the axis is perpendicular. For simplicity, the three-dimensional coordinates can be kept fixed, and the weight of the soil column can be considered. Around each shaft and Axis rotation and Angle, such as Figure 5 As shown, at this time For Right angle The triangle is obviously ,at the same time For Right angle The triangle is obviously At this time, the force to be measured Obviously .

[0029] The lyometer is placed on a surface covered with plants of a certain height and a considerable windward area. The wind causes the plants to sway, resulting in a horizontal impact force on the soil container. Without loss of generality, the soil container is surrounded... Axis tilt Angle and with edge x The wind on the axis causes an impact force For example, Figure 6 As shown in the figure For the actual force measurement direction, for The component of the force along the direction of the measured force is visible. Let the actual mass of the soil box be... The horizontal acceleration caused by the wind is c ,but The true mass in a lever-type weighing lysimeter Since it is unknown and there is no need to measure it, the component force is... It will be unknown.

[0030] In summary, regardless of whether the soil column is tilted, whether there is wind or plants exerting a horizontal force on the soil column, or whether both are acting together, the mass of the soil box measured by the lever-type weighing lysimeter through the weighing sensor is consistent. All values ​​are not true values, and the greater the tilt angle and wind speed, the more significant the impact, leading to substantial measurement errors. Clearly, the true mass of the soil box to be measured is... With and around Axis tilt angle , around Axis tilt angle , Axial acceleration , Axial acceleration Actual mass of soil box Both are related, and it can be assumed that there is a functional relationship between them, that is:

[0031] But functions The mathematical expression for this function is unknown and difficult to establish. Regarding function approximation, a BP neural network trained using a global approximation algorithm can approximate any nonlinear function with arbitrarily high precision. Therefore, if we consider... Axis tilt angle , around Axis tilt angle , Axial acceleration , Axial acceleration Measured mass of soil box For input vectors, the true mass For output, a function is built based on the training data. The neural network-based weighing compensation model can obtain weighing measurement results that eliminate or reduce the effects of tilt angle and wind.

[0032] Example 6 The lever-type weighing lysimeter used in this embodiment is a cylinder with a height of 4m and a radius of 1000mm, and the two-axis tilt sensor is QKQH2-20. shaft and The shaft acceleration sensor is SCA3400-D01, the load cell is BCE4202I-1, the wind speed sensor is NBL_W_SS, the wind direction sensor is NBL_W_DS, and the industrial fan is FS-65.

[0033] There are P=8 tilt cases, which can be represented by the array Dip_case[8]={1, 2, 3, 4, 12, 23, 34, 41} (tilting towards the first elastic support point 1, tilting towards the second elastic support point 2, tilting towards the third elastic support point 3, tilting towards the fourth elastic support point 4, tilting towards both the first and second elastic support points 1 and 2, tilting towards both the second and third elastic support points 2 and 3, tilting towards both the third and fourth elastic support points 3 and 4, and tilting towards both the fourth and first elastic support points 1). The tilt angle range is 0 to 15°. Within this range, M=4 different angles are selected, and the specific angles can be represented by the array Dip_angle[4]={0°, 5°, 10°, 15°}. The wind speed range is 0 to 8 m / s. Within this wind speed range, N=4 different wind speeds are selected, and the specific wind speeds can be represented by the array Wind[4]={0 m / s, 2 m / s, 5 m / s, 8 m / s}. The standard weight is F=20kg, and the step size is ΔF=20kg. During calibration under condition 1, the load is increased from 0kg to C=200kg and then decreased back to 0kg. The number of calibration points obtained is... =21. Condition 2 calibration uses the same weights and step size as step 1, loading from 0 kg to C=200 kg. =21, a total of 2688 sets of data can be obtained.

[0034] Based on the calibration data without wind or tilt, the least squares method can be used to determine , .

[0035] The recompensation model employs a three-layer structure: a 5-neuron input layer, a 1-neuron output layer, and the number of hidden layer neurons determined through trial and error. Momentum factor during training Learning rate The training sample consisted of 1792 groups, the test sample consisted of 896 groups, the training number was 200, and the performance index MSE=1E-5.

[0036] Weighing tests were conducted on a lever-type gravimetric lysimeter. On a sunny day with no natural wind, the soil column of the lysimeter was artificially tilted at a 5° angle. The exposed soil surface was covered with mulch to prevent moisture evaporation. To ignore the influence of plant transpiration on the gravimetric results, the test was conducted in the morning. The test weight was 30 kg, the test duration was 2 hours, and the sampling interval was 5 minutes. To quantitatively illustrate the measurement effect of the method of this invention, the root mean square error (RMSE) and the mean absolute error (MAE) were defined as follows:

[0037]

[0038] in, and These are the expected value and the calculated value after compensation, respectively. For measuring sample serial number, This represents the total number of measurement sample points.

[0039] The test results are shown in Table 1: Table 1 Test Comparison Results

[0040] As can be seen from Table 1, when there is no wind and no tilt, the RMSE and MAE of the method of the present invention are consistent with the performance before compensation. However, for the cases of no wind and tilt, wind and no tilt, and wind and tilt, the method of the present invention can reduce RMSE by 77.46%, 97.86% and 97.99% respectively, and reduce MAE by 76.10%, 97.54% and 97.77% respectively, with significant improvement effect.

Claims

1. A lever-type lysimeter weight detection method that can suppress soil box tilting and vibration, characterized in that, The steps are as follows: Step 1, install on the lever-type weighing lysimeter x Shaft acceleration sensor (6) y Axial acceleration sensor (7), two-axis tilt sensor (8); Step 2: Calibrate the lever-type weighing lysimeter under different operating conditions using standard weights; Step 3: Establish a weighing compensation model and train the weighing compensation model to obtain a trained weighing compensation model. Step 4: Obtain the final weighing measurement result based on the trained weighing compensation model.

2. The lever-type lysimeter weight detection method for suppressing soil box tilting and vibration according to claim 1, characterized in that, In step 1, the lever-type weighing lysimeter includes a cylindrical soil box (9). A lever fixing point (5) is set at the center of the bottom surface of the soil box (9). A first elastic support point (1), a second elastic support point (2), a third elastic support point (3), and a fourth elastic support point (4) are set sequentially along the circumference of the bottom surface of the soil box (9). Each of the first elastic support point (1), the second elastic support point (2), the third elastic support point (3), and the fourth elastic support point (4) is connected to an elastic support (10). The end of each elastic support (10) away from the soil box (9) is connected to the support surface (16). The lever fixing point (5) A lever fixing block (17) is provided at the bottom of the lever (14), and a lever connection point (11) is provided at the bottom of the lever fixing block (17). The lever connection point (11) is connected to one end of the lever (14), and a weighing sensor (15) is connected to the other end of the lever (14). A lever support block (18) is connected to the bottom of the lever (14) near the lever connection point (11). The connection between the lever support block (18) and the lever (14) serves as the lever fulcrum (12). The bottom of the lever support block (18) is connected to the support surface (16). A counterweight (13) is connected to the side wall of the lever (14) near the weighing sensor (15). x Shaft acceleration sensor (6) y The axial acceleration sensor (7) and the two-axis tilt sensor (8) are set on the upper part of the outer surface of the soil box (9), and the two-axis tilt sensor (8) is located at... x Directly below the axial accelerometer (6), y The shaft acceleration sensor (7) is located at x The axial acceleration sensor (6) is positioned between the two axial tilt sensors (8) and on the side close to the load cell (15).

3. The lever-type lysimeter weight detection method for suppressing soil box tilting and vibration according to claim 2, characterized in that, The line connecting the first elastic support point (1), the second elastic support point (2), the third elastic support point (3), and the fourth elastic support point (4) is a square, and the intersection of the two diagonals of the square coincides with the lever fixing point (5).

4. The lever-type lysimeter weight detection method for suppressing soil box tilting and vibration according to claim 1, characterized in that, In step 2, the operating conditions include no tilt and no wind, tilt and wind.

5. The lever-type lysimeter weight detection method for suppressing soil box tilting and vibration according to claim 4, characterized in that, Condition 1: The calibration experiment under no tilt and no wind conditions is as follows: Choose a windless day, adjust the soil box (9) to be level, then adjust the counterweight (13) so that the output of the weighing sensor (15) is at 1 / 2 of full scale. Then, using standard weights, load the soil box (9) from 0 kg to [amount missing] in increments of ΔF. kg, then from The load was reduced from 0 kg to 0 kg, and the output of the weighing sensor (15) was recorded during each loading and unloading. and the corresponding weight of the weights , Indicates the calibration serial number. This represents the total number of calibration points. The input-output working curve of the lever-type weighing lysimeter during weighing can be expressed by formula (1), that is: （1） In equation (1), The coefficient related to leverage (14); The bias amount related to the counterweight (13); This is the input quantity used when weighing in a lever-type gravimetric lysimeter. This is the output quantity of the lever-type weighing quasi-percolator during weighing; The output of the weighing sensor (15) As the input quantity when weighing in a lever-type gravimetric diatomometer, the corresponding weight is... As the output of the lever-type weighing lysimeter, the input-output working curve of the lever-type weighing lysimeter is fitted using the least squares algorithm. Then, in formula (1) It can be obtained by formula (2). It can be obtained through formula (3), that is: （2） （3）。 6. The lever-type lysimeter weight detection method for suppressing soil box tilting and vibration according to claim 5, characterized in that, Operating Condition 2: The calibration test procedure under inclined and windy conditions is as follows: Step S1, set the calibration experimental conditions: On a windless day, the tilt angle of the soil box (9) is adjusted by changing the position of the standard weight on the surface of the soil box (9). The tilt angle is measured by a two-axis tilt sensor (8). An adjustable speed industrial fan is used to simulate the natural wind blowing across the surface of the soil box (9). The wind direction is determined according to the dominant wind direction in the historical agricultural meteorological data of the area. The wind speed is measured by a wind power sensor and the wind direction is measured by a wind direction sensor. The tilt includes eight scenarios: the first scenario is tilting towards the first elastic support point (1), the second scenario is tilting towards the second elastic support point (2), the third scenario is tilting towards the third elastic support point (3), the fourth scenario is tilting towards the fourth elastic support point (4), the fifth scenario is tilting towards both the first elastic support point (1) and the second elastic support point (2) simultaneously, the sixth scenario is tilting towards both the second elastic support point (2) and the third elastic support point (3) simultaneously, the seventh scenario is tilting towards both the third elastic support point (3) and the fourth elastic support point (4) simultaneously, and the eighth scenario is tilting towards both the fourth elastic support point (4) and the first elastic support point (1) simultaneously. The array Dip_case[P] is used to represent the P types of tilting. The tilt angle range is set to 0°~15°. M different tilt angles are selected within the tilt angle range for calibration. The array Dip_angle[M] is used to represent the M different tilt angles. Based on the degree of influence of wind speed on agricultural planting, the wind speed range is determined to be 0m / s to 8m / s. Within the wind speed range, N different wind speeds are selected for calibration, and the array Wind[N] is used to represent the N different wind speeds. Step S2, first let , Indicates the first The first type of tilt is selected. Step S3, determine If yes, proceed to step S5; otherwise, proceed to step S4. Step S4, end calibration; Step S5, set the tilt case Dip_case[ l ],initialization ; Step S6, determine , Indicates the first If the tilt angle is yes, proceed to step S8; otherwise, proceed to step S7. Step S7, This indicates that a tilting condition has been selected, and step S3 is executed; Step S8, set the tilt angle Dip_angle[ m ],initialization ; Step S9, determine , Indicates the first If the wind speed is specified, proceed to step S11; otherwise, proceed to step S10. Step S10, This indicates that the next tilt angle should be selected, and step S6 should be executed; Step S11, set the wind speed Wind[ n ]; Step S12: Using standard weights, load from 0 kg to [amount missing] kg in increments of ΔF. kg, number of calibration points After completion, unload all weights; Step S13, This indicates that the next wind speed should be selected, and step S9 should be executed. Step S14, record the results of each experiment. Axis tilt angle, Axis tilt angle, Axial acceleration, The axial acceleration, wind speed measured by the wind speed sensor, weight of the weight, and output of the weighing sensor.

7. The lever-type lysimeter weight detection method for suppressing soil box tilting and vibration according to claim 6, characterized in that, Step 3 is as follows: Step 3.1: Construct the dataset and normalize the data in the dataset, then divide the normalized dataset into a training set and a test set. Step 3.2: Train the weighing compensation model using the training set to obtain the trained weighing compensation model. Test the trained weighing compensation model using the test set. If the test results meet the requirements, the trained weighing compensation model is obtained. If the test results do not meet the requirements, retraining is performed. The loss function used during training is: （8） In equation (8), k Indicates the current training sample number. , This represents the total number of training samples. The true values ​​of the current training samples. This is the output value of the weighing compensation model.

8. The lever-type lysimeter weight detection method for suppressing soil box tilting and vibration according to claim 7, characterized in that, In step 3.1, the dataset is... Serial Number Total number of samples ,in Let be the input vector, where for x Axis tilt angle, for y Axis tilt angle, for x Axial acceleration, for y Axial acceleration, This is the output of the weighing sensor. The output value is the output data of the weighing sensor corresponding to the weights used in the calibration experiment under condition 2 in condition 1.

9. The lever-type lysimeter weight detection method for suppressing soil box tilting and vibration according to claim 7, characterized in that, In step 3.2, the weighing compensation model uses a backpropagation (BP) neural network. The BP neural network has a 3-layer structure: 5 neurons in the input layer, 1 neuron in the output layer, and [number missing] neurons in the hidden layer. .

10. The lever-type lysimeter weight detection method for suppressing soil box tilting and vibration according to claim 7, characterized in that, The specific process of step 4 is as follows: When the lever-type weighing lysimeter performs the weighing measurement task, it simultaneously collects data. Axis tilt angle, Axis tilt angle, Axial acceleration, The axial acceleration and the output data of the weighing sensor (15) are normalized. The normalized data is input into the trained weighing compensation model. The output of the trained weighing compensation model is inversely normalized to obtain the inversely normalized weighing sensor output. The inversely normalized weighing sensor output is substituted into formula (1) to obtain the final weighing measurement result.

Images