Researchers and planter manufacturers have been working closely to develop an automated system for evaluating performance of seeding. In the present study, an innovative use of acoustic signal for laboratory evaluation of seeding-machine application is described. Seed detection technique of the proposed system was based on a rising voltage value that a microphone sensed in each impaction of seeds to a steel plate. Online determining of seed spacing was done with a script which was written in MATLAB software. To evaluate the acoustic system with desired seed spacing, a testing rig was designed. Seeds of wheat, corn and pelleted tomato were used as experimental material. Typical seed patterns were positioned manually on a belt stand with different spacing patterns. When the belt was running, the falling seeds from the end point of the belt impacted to the steel plate, and their acoustic signal was sensed by the microphone. In each impact, data was processed and spacing between the seeds was automatically obtained. Coefficient of determination of gathered data from the belt system and the corresponding seeds spacing measured with the acoustic system in all runs was about 0.98. This strong correlation indicates that the acoustic system worked well in determining the seeds spacing.
Precision spacing of seeds can provide maximum space for each plant, reducing intra-specific competition and increasing yields (
In laboratory tests, sticky belt stand is one of the most frequently used methods (
Among other alternative techniques, an opto-electronic seed spacing determination system has been developed (
With current advances in digital video technology, machine vision has shown potential as a sensing technology for seeding performance evaluation. In this case, spatial distribution of seeds is measured with a digital camera for recording of passing seeds and a computer for data processing and monitoring. Results from this machine vision technique have shown a good accuracy (
Lately, non-destructive acoustical experiments have been progressively executed in agricultural engineering and its accuracy is proven in detection and classification of agricultural products (
Acoustic signals characteristic of an impacted seed is in relation with the seed physical and mechanical properties. In order to evaluate the devised system performance with different seed shapes, seeds of wheat (
A vernier caliper was used to determine length (L), width (W), and thickness (T) of about 50 randomly selected seeds of each sample. Axial dimensions were used to determine sphericity (Sp) using the
Thousand seed weight (TSW) was measured by sampling 50 seeds and weighing them in an electronic balance (accuracy 0.001). This weight was then multiplied by 20 to give the mass of 1000 seeds. The mean values and standard errors of acquired properties of the seeds in ten repetitions are presented in
To evaluate the acoustic system with desired seed spacing, a testing rig was designed. The testing rig was composed of a belt and an acoustical system (
Seeds spacing was determined based on a rising voltage value that the microphone sensed in each seed’s impact to the plate. For this purpose, MATLAB software was used. This software in data-acquisition toolbox allows to acquire data from sensors and to send out electrical signals that can be used to control or drive external devices (
The device was put into a continuous acquisition mode, and acquisition began when start commend was set. The collected data were analyzed in order to detect the specified trigger condition. If the data did not contain the trigger condition, they were discarded. When the trigger condition was met, the engine begins storing data. These data were retrieved subsequently. By an adjustment using several trails, it was found that the most appropriate threshold point that could sense the three kinds of seeds at one time is 0.4 volts. Primary experiments also showed that putting 0.4 volts for triggering sound card caused environmental sounds not being able to start data acquisition except impaction of seeds. Moreover, for all three kinds of seeds, when the logging data for each seed’s impact sound continued for 0.015 sec after passing this point, no rising in the voltage value equal or greater than 0.4 volts would happen. Therefore, setting data acquisition on logging data with 700 samples for each impact sound (when the sound card frequency is set to 44.1 kHz) provided system to sense each seed one time. Data acquisition was configured with a sound card based on voice activation. The sample rate was set to 44.1 kHz, and data was logged when an acquired sample had a value equal to or greater than 0.4 volts and a rising slope. After triggering, upon getting a trigger signal the computer acquired 700 data points from every sample in the time-domain. Typical data acquired from pelleted tomato seed with desired condition is shown in
According to the described principles, a script was written in MATLAB software, in each loop, trigger time and data for each trigger was obtained. Flowchart for written script is demonstrated in
For evaluation of the acoustical technique, seeds were positioned manually in the trajectory line on the belt with distinctive spaces (10 to 40 cm intervals). For each space range, the experiment was replicated 5 times. The seeds were positioned just in the first 3.5 m of the belt so that the running belt had time to reach to a steady state operating condition. When the belt was running, the positioned seeds on the trajectory line were falling from the end point of belt, so seeds were impacting to the plate one by one and related seed spacing data were acquired by the acoustical system. In each stage of the experiment, the belt simulated travel speed was set up to 0.5 m/s.
The mean absolute percentage error (MAPE), also known as mean absolute percentage deviation (MAPD), is a quantity used to measure how close forecasts or predictions are to the eventual outcomes (
where
Linear regression was used to model the relationship between seeds spacing on belt system as an independent variable and acoustic system data as a dependent variable. Results of regression analysis and correlation coefficient of variables for corn, pelleted tomato and wheat seeds are presented in
MAPE results obtained for corn, pelleted tomato and wheat seeds in different spacing ranges are shown in
Sticky belt stand is the method commonly used by researchers as a reference technique to test the seed spacing of each planter configuration (
Furthermore, the benefits of the proposed acoustic system in comparison with the sticky belt method are: (1) unlimited number of data can be obtained by acoustic system; acoustic system seed spacing determination is not restricted due to infinite data acquisition possibility provided with PC and MATLAB software; whereas, in the sticky belt method, the length of the belt limits number of data that can be obtained (
Moreover, significant simplicity and low cost of proposed acoustic system is also expected in comparison with opto-electronic and machine vision systems; such that opto-electronic seeds spacing determination systems are costly and include complex hardware and software (
Another advantage of the acoustic system is easy calibration process. By determining proper threshold point adjusted with several trails, no irrelevant sound except seed’s impact sound can run the system. instead, the machine vision system requires defined and consistent lighting (
As
In this study, primary tests were done through trial and error method to investigate appropriate data acquisition setting to sense all utilized seeds at once. In other words, same data acquisition setting (including threshold point) was successfully found for all utilized seeds. Our results have shown that with advanced setting in data acquisition of impact sound signal for different kinds of seeds with various shapes used in this study, good accuracy of seed spacing can be determined by the designed acoustic seed spacing evaluation system. Nevertheless, seeds with different physical properties may need special setting for impact sound data acquisition.
According to the mentioned results, it is clearly argued that good performance for the acoustic system is expected for seeds which are greater than the small sized seeds of this study. Extension of this method for smaller sizes and weights of seeds could be continued up to a point where ability of smaller seeds to make impact sound signals is higher than entering ambient noises into the isolated chamber. In this case, determining proper threshold point to start data acquisition of seed impaction would not be possible.
However, improving the accuracy of the acoustical system seems to be possible if the following suggestions would be followed: utilizing a sound card with more sampling frequency (> 44.1 kHz), increasing the attenuation of the impact sound signal by using thicker stainless steel blocks as impact plates and reinforcing the chamber’s isolation ability to reduce the environmental noise.
In addition, the main efforts in this study were made to develop a system as simple and cheap as possible. Some investigations are recommended in future works about acoustic seeds spacing determination system, as developing an adaptive detection algorithm and using a data conditioning method to eliminate the ambient noise.
Furthermore, to the best of our knowledge, no successful automatic system for determination of damaged seeds caused by malfunction of seeding mechanism except the acoustic technique has been reported in the scientific literature. Seed breakage can be a problem in planters with mechanical metering devices and it is an important parameter in the evaluation of seeding machine application performance. In a previous research (
Advantages of the proposed system, besides its good accuracy, are automatism and online seed spacing determination, low cost, simplicity and easy calibration in laboratory conditions. Considering these benefits, the extensive use of the acoustical system in laboratory evaluation of planter would be conceivable in future.