Self-propelled mulching film sampling detection equipment and detection method
By using self-propelled plastic film sampling and detection equipment and a deep neural network model, automated sampling, rapid on-site detection, and identification of plastic film residue from different years in farmland have been achieved. This solves the problems of high labor intensity and low detection efficiency of traditional methods and provides accurate residual film assessment data.
Patent Information
- Application Number
- CN202610206130.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-02-12
- Publication Date
- 2026-05-19
AI Technical Summary
Existing technologies make it difficult to automate the sampling, rapid on-site detection, and identification and assessment of residual film from different years in farmland. Traditional methods are labor-intensive and inefficient, automated devices lack specificity, and existing equipment is expensive and unsuitable for rapid field detection.
A self-propelled mulch film sampling and testing equipment was designed, including an equipment chassis, a rotary tillage and excavation sampling device, a conveying device, and a storage and testing device. Combining an image acquisition mechanism and a deep neural network model, it realizes automated soil collection, image recognition, and quality calculation. Through the coordinated actions of rotary tillage, excavation, conveying, storage, leveling, and image acquisition, it identifies and calculates the amount of mulch film residue in different years.
It has achieved full mechanization and automation of the sampling process for plastic film testing, reducing labor intensity, improving work efficiency, providing accurate residual film testing data, and solving the problems of existing methods being affected by human factors and unable to be evaluated on an annual basis.
Smart Images
Figure CN122062932A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the technical field of automated testing in agricultural engineering, and in particular to a self-propelled mulch film sampling and testing equipment and method. Background Technology
[0002] In modern agricultural production, plastic film mulching technology has been widely used due to its significant effects in warming, conserving moisture, controlling weeds, and promoting crop growth. However, with the increase in the amount and years of use of plastic film, a large amount of residual film accumulates in the soil, forming "white pollution," damaging soil structure, hindering the transport of water and nutrients, reducing soil fertility, and seriously affecting crop growth and sustainable agricultural development.
[0003] Current methods for detecting residual plastic film in soil have the following limitations: Traditional gravimetric methods require manual soil sample collection, followed by tedious processes such as sorting, washing, drying, and weighing in the laboratory, resulting in high labor intensity, low efficiency, and susceptibility to human factors; while pyrolysis-gas chromatography / mass spectrometry can accurately analyze the composition of plastic film, the equipment is expensive, the detection cycle is long, and environmental requirements are stringent, making it unsuitable for rapid field detection. Furthermore, existing automated soil sampling devices only focus on soil collection and lack specific identification and detection capabilities for plastic film residue.
[0004] While some technologies have attempted image-based identification of agricultural film, such as detecting thin soil cross-sections or classifying film in cotton impurities, these methods often lack the ability to automate soil sampling and assess residual film levels from different years. Furthermore, their application scenarios differ significantly from field film residue detection, failing to meet the needs of large-scale, automated, and yearly assessments of agricultural film pollution. Therefore, a complete technical solution integrating self-propelled sampling, rapid on-site detection, and residual film identification from different years is currently lacking. Summary of the Invention
[0005] The purpose of this invention is to provide a self-propelled mulch film sampling and testing equipment and method, which solves the technical problems of how to achieve automated sampling, rapid on-site testing, and effective identification and evaluation of mulch film residue from different years in farmland.
[0006] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: In a first aspect, the present invention provides a self-propelled mulch film sampling and detection device, comprising: The equipment chassis is used to support various working devices and provide the power for the entire machine to move; A rotary tillage and excavation sampling device is installed at the front of the equipment chassis and is used to loosen the soil at a specified depth on the ground during the movement of the equipment. A conveying device, installed on the chassis of the equipment and located behind the rotary tillage excavation and sampling device, is used to convey the loosened soil upwards. A storage and testing device, installed on the equipment chassis and located at the discharge end of the chain conveyor, is used to receive, store, and test the conveyed soil, and includes: Storage container, used to receive and store soil; The bottom plate lifting mechanism of the storage box is connected to the bottom moving plate of the storage box, and can drive the bottom moving plate of the storage box and the soil placed on it to be lifted sequentially in the vertical direction. A soil scraping mechanism is installed above the storage box and is used to cooperate with the bottom plate lifting mechanism of the box to scrape the surface layer of soil that is lifted layer by layer in the storage box. An image acquisition mechanism, located above the storage box, is used to acquire images of the leveled soil surface for subsequent detection and analysis.
[0007] Furthermore, the rotary tillage excavation sampling device includes: The gasoline rotary tiller assembly is rotatably connected to the chassis of the equipment. A rotary tillage height adjustment hydraulic cylinder is installed on the equipment chassis and connected to the gasoline rotary tiller assembly. It is used to drive the gasoline rotary tiller assembly to rotate vertically to adjust its tillage depth.
[0008] Furthermore, the conveying device includes: The conveying device body is rotatably connected to the equipment chassis; The lifting hydraulic cylinder of the conveying device is installed on the chassis of the equipment and connected to the body of the conveying device. It is used to adjust the height of the front end of the conveying device body so that its lowering depth is adapted to the lowering depth of the rotary tiller blades of the gasoline rotary tiller assembly.
[0009] Furthermore, the box bottom plate lifting mechanism includes: An electric push rod for lifting the bottom plate of the box is installed on the equipment chassis and connected to the movable plate at the bottom of the box. The bottom plate of the box is lifted and supported by a load-bearing plate, which is connected to the output end of the electric push rod for lifting the bottom plate of the box, and is connected to the bottom moving plate of the box through the lifting seam on the side wall of the storage box.
[0010] Furthermore, the storage detection device also includes: The image processing unit is configured to identify and classify residual plastic film from different years in the images based on the acquired multi-layer soil surface images using a pre-trained deep neural network model, and calculate the quality of residual plastic film from different years based on the identification results.
[0011] Furthermore, the pre-trained deep neural network model is a multi-class target detection model trained based on the differences in wrinkles, dirt, and color difference of residual plastic film from different years.
[0012] Furthermore, the image processing unit is also configured to: perform edge detection and image segmentation on the image area where residual plastic film is identified, extract the residual film image and calculate its pixel area, and calculate the mass of residual plastic film for each year based on the sum of the pixel areas of plastic film fragments from each year, combined with preset residual film thickness, density parameters and area conversion coefficients.
[0013] Secondly, the present invention also provides a self-propelled mulch film sampling and detection method based on the detection equipment described in any one of the above claims, comprising the following steps: S1: The loosened soil is transported to the storage and testing device for storage; S2: Gradually lift the soil inside the box and scrape the soil surface layer, and collect images of the soil surface layer after each scraping. S3: Input the acquired multi-layer soil surface images into a pre-trained deep neural network model to identify and classify residual plastic film from different years in the images; S4: Based on the identification results, calculate the mass of residual plastic film from different years within a single sampling area.
[0014] Furthermore, in step S3, the pre-training of the deep neural network model includes the following steps: Constructing a training dataset: Acquire and sort residual plastic film samples from different years; randomly bury the samples in the soil and turn them over to simulate field conditions; remove the topsoil layer by layer in the buried soil area and take repeated photos until the rotary tillage depth is reached to obtain a series of soil images at different depths; label the images according to the year of the plastic film samples and their corresponding wrinkles, dirt, and color difference features to construct a training dataset. Model training: The deep neural network model is trained based on the training dataset.
[0015] Furthermore, step S4 specifically includes: S4.1: Perform image segmentation on the identified plastic film fragments in each image and calculate their pixel area; S4.2: Based on the total pixel area of the plastic film fragments from each year, combined with the preset residual film area conversion coefficient, the thickness and density of the plastic film in that year, calculate the residual mass of plastic film per unit area for each year.
[0016] Compared with the prior art, the present invention has at least the following beneficial effects: The self-propelled mulch film sampling and testing equipment of this invention realizes the full mechanization and automation of the mulch film testing and sampling process. Relying on a tracked chassis, the whole machine can operate on its own. The rotary tillage and excavation sampling device can precisely adjust the soil loosening depth and complete the automatic loosening. The conveying device can transport the soil to the storage and testing device in sync with the loosening action. No manual participation is required in the excavation of sample plots and soil transportation throughout the process, which completely changes the traditional manual sampling mode and greatly reduces the labor intensity of mulch film testing personnel. At the same time, the hydraulic synchronous adjustment design of each device ensures the standardization and consistency of the sampling process and effectively improves the work efficiency of sampling residual mulch film in farmland.
[0017] This invention achieves accurate on-site detection of residual agricultural film in farmland and classification and identification of film residue from different years through the deep integration of mechanical structure and image detection algorithm. The storage and detection device provides a standardized image acquisition basis for film residue detection through the coordinated actions of quantitative lifting of the base plate, soil scraping and leveling, and layer-by-layer photography. The detection algorithm based on a deep neural network model takes film wrinkles, dirt, and color difference as core features to accurately identify residual film from different years. Then, through image segmentation, pixel area calculation, and mass conversion, it can accurately obtain the amount of residual film from each year and the total amount of residual film per unit area. This solves the problems of the accuracy of existing detection methods being affected by human factors and the inability to evaluate residual film by year, and provides accurate and reliable measured data for large-scale surveys of agricultural film residue. Attached Figure Description
[0018] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0019] Figure 1 This is a schematic diagram of the structure of the self-propelled mulch film sampling and testing equipment of the present invention; Figure 2 for Figure 1 A schematic diagram of the tracked chassis in the structure shown; Figure 3 for Figure 2 A schematic diagram of the right track drive device in the structure shown; Figure 4 for Figure 2 A schematic diagram of the main support frame in the structure shown; Figure 5 for Figure 1 A schematic diagram of the rotary tillage excavation sampling device in the structure shown; Figure 6a for Figure 1A schematic diagram of the chain conveyor device shown in the diagram; Figure 6b for Figure 1 A partial structural diagram of the chain conveyor device shown in the diagram. Figure 7a for Figure 1 A schematic diagram of the storage and detection device from a certain perspective in the structure shown; Figure 7b for Figure 1 A schematic diagram of the storage and detection device from another perspective in the structure shown; Figure 8 for Figure 7a A schematic diagram of the wedge-shaped scraper plate in the structure shown; Figure 9 for Figure 7b The diagram shows the structural schematic of the lifting support plate of the box bottom plate in the system shown. Figure 10 A control logic diagram for the mulch film testing process of a self-propelled mulch film sampling and testing equipment; Figure 11 This is a schematic diagram of soil transport. Figure 12 Flowchart of the method for constructing the training dataset; Figure 13 Flowchart for building the object detection model; Figure 14 This is a basic structural diagram of a neural network model; Figure 15 This describes the principle of an image-based algorithm for detecting actual residual plastic film.
[0020] Figure label: 1- Tracked chassis; 2- Rotary tillage and excavation sampling device; 3- Chain conveyor device; 4- Storage and testing device; 11-Left track drive unit; 12-Right track drive unit; 13-Main support frame; 121-Fixing position of the mounting rod for the gasoline rotary tiller; 122-Connection point of the hydraulic cylinder mounting plate; 123-Mounting point of the main support frame; 131-Adjusting hydraulic cylinder connecting lug; 132-Fixing bearing connection; 133-Hydraulic cylinder support; 134-Base plate of the main support frame; 21-Gasoline rotary tiller assembly; 22-Rotary tillage digging height adjustment hydraulic cylinder; 23-Lifting rod; 24-Mounting support rod; 300-Wedge-shaped shovel; 301-Drive chain; 302-Chain plate; 303-Lifting rod of conveying device; 304-Lifting hydraulic cylinder of conveying device; 305-Hydraulic cylinder mounting plate; 306-Conveying power motor; 307-Conveying power chain; 308-Mounting support rod of conveying device; 309-Conveying device frame; 310-Rear drive shaft; 311-Front drive shaft; 312-Mounting plate of conveying device; 401-Storage box; 402-Box support frame; 403-Scraping guide shaft; 404-Box bottom plate lifting guide bearing; 405-Box bottom plate lifting guide shaft; 406-Box bottom plate lifting electric push rod; 407-Box bottom plate lifting support plate; 408-Scraping linear bearing; 409-Wedge-shaped scraper blade; 410-Scraping hydraulic cylinder; 411-Box bottom plate lifting guide shaft fixed support bearing seat; 412-Electric... 413 - Camera and flash assembly; 414 - Camera and flash mounting rod; 415 - Nylon sheet; 416 - Bottom moving plate of the enclosure; 4011 - Lifting joint; 4071 - Load-bearing lug; 4072 - Load-bearing plate electric push rod; 4073 - Bottom plate lifting guide bearing mounting hole of the enclosure; 4091 - Scraper blade; 4092 - Linear bearing mounting point; 4093 - Hydraulic cylinder mounting point. Detailed Implementation
[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0022] The following detailed description of some embodiments of the present invention is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.
[0023] This embodiment provides a self-propelled mulch film sampling and testing equipment and method. The core of the testing equipment consists of an equipment chassis, a rotary tillage and excavation sampling device, a conveying device (such as a chain conveyor), and a storage and testing device. The equipment chassis carries all devices and provides the power for movement. The rotary tillage and excavation sampling device loosens the soil to a specified depth during the equipment's movement. The conveying device transports the loosened soil upwards to the storage and testing device. The storage box of the storage and testing device receives and stores the soil. Its bottom plate lifting mechanism can drive the soil to be lifted vertically and sequentially. A scraping mechanism works in conjunction to scrape the surface layer of the lifted soil layer by layer. The image acquisition mechanism then acquires images of the scraped soil surface layer, providing a basis for subsequent testing. The detection method is based on the aforementioned sampling and detection equipment. The core process is to first use the equipment to automatically transport and store the loosened soil, and then use the coordinated actions of the equipment's bottom plate lifting, scraping, and image acquisition mechanisms to lift and level the soil layer by layer in the box and acquire surface images of each layer. Subsequently, the multi-layer images are input into a pre-trained deep neural network model to identify and classify residual plastic film from different years in the images. Finally, the residual plastic film mass from different years in a single sampling area is calculated based on the identification results.
[0024] Furthermore, this self-propelled mulch film sampling and testing equipment and method are implemented through the following core technical solutions, as detailed below: I. Self-propelled Mulch Film Sampling and Testing Equipment This self-propelled mulch film sampling and testing equipment, such as Figure 1 As shown, the system includes a tracked chassis 1, a rotary tillage and sampling device 2, a chain conveyor device 3, and a storage and detection device 4. The tracked chassis 1 carries each working device and provides the power for the entire machine to move forward. The rotary tillage and sampling device 2 is responsible for loosening the soil at a specified depth on the ground as the chassis moves along the sampling path. The chain conveyor device 3 transports the loosened soil to the storage and detection device 4 for soil storage. After sampling, the storage and detection device, in conjunction with its related mechanical structure and image acquisition camera, takes orderly, layer-by-layer photos of the stored soil and uses an image detection algorithm to accurately identify residual plastic film and determine the amount of residual plastic film. The tracked chassis 1, such as Figure 2 , Figure 3 and Figure 4 As shown, it includes a left track drive unit 11, a right track drive unit 12, and a main support frame 13. The main support frame 13 spans across the left track drive unit 11 and the right track drive unit 12, and achieves relative fixation of the left track drive unit 11 and the right track drive unit 12 through the main support frame 13, and provides mounting space for the working device. The left track drive unit 11 and the right track drive unit 12 are driven independently, and rely on their own differential motion to realize the forward, backward, left turn, and right turn of the tracked chassis 1. The upper part of the right track drive device 12 has a gasoline rotary tiller mounting support rod fixing position 121, a hydraulic cylinder fixing mounting plate connection point 122, and a main support frame mounting point 123; the left track drive device 11 is structurally symmetrical to the right track drive device 12; the main support frame 13 includes an adjusting hydraulic cylinder connecting lug 131, a fixed bearing connection 132, a cylinder support 133, and a main support frame base plate 134; the main support frame 13 is connected to the main support frame mounting points on the left and right track drive devices through the main support frame base plate 134.
[0025] The rotary tillage excavation sampling device 2, such as Figure 5 As shown, the system includes a gasoline rotary tiller assembly 21, a rotary tillage height adjustment hydraulic cylinder 22, a lifting rod 23, and a mounting support rod 24. One end of the bottom structural component of the gasoline rotary tiller assembly 21 is rigidly connected to the mounting support rod 24. The mounting support rod 24 is connected to the bearing seats on both sides and is respectively installed at the gasoline rotary tiller mounting support rod fixing positions on the left track drive device 11 and the right track drive device 12. Figure 5As shown; the middle part of the bottom structural component of the gasoline rotary tiller assembly 21 is connected to the lifting rod 23. The two ends of the lifting rod 23 are respectively hinged to the ends of the action rods of the rotary tillage height adjustment hydraulic cylinder 22. The other end of the rotary tillage height adjustment hydraulic cylinder 22 is hinged to the connecting lug 131 of the front adjustment hydraulic cylinder of the main support frame 13. During operation, under the action of the hydraulic system, the rotary tillage height adjustment hydraulic cylinders 22 on both sides can extend and shorten synchronously. Therefore, the gasoline rotary tiller assembly 21 can rotate around the center of the mounting rod 24, pushing the lifting rod 23 up and down, realizing the adjustment of the position height of the front rotary tiller blade, and thus adjusting the soil loosening depth. The gasoline rotary tiller assembly 21 is a mature mechanical component, which drives the rotary tiller blade to rotate at high speed through the operation of its gasoline engine.
[0026] The chain conveyor device 3, as shown in Figures 6(a, b), includes a wedge-shaped shovel 300, a drive chain 301, a chain plate 302, a lifting rod 303, a lifting hydraulic cylinder 304, a hydraulic cylinder mounting plate 305, a conveyor power motor 306, a conveyor power chain 307, a conveyor mounting support rod 308, a conveyor frame 309, a rear drive shaft 310, a front drive shaft 311, and a conveyor mounting plate 312. The wedge-shaped shovel 300 is bolted to the front end of the conveyor frame 309. The front and rear parts of the conveyor frame 309 are respectively connected to... A front drive shaft 311 and a rear drive shaft 310 are mounted on a bearing. The front drive shaft 311 and the rear drive shaft 310 can rotate freely around their axes relative to the conveyor frame 309. Sprockets are mounted at both ends of the front drive shaft 311 and the rear drive shaft 310. A transmission chain 301 is installed between the corresponding sprockets at both ends of the front drive shaft 311 and the rear drive shaft 310 to synchronize the movement of the corresponding sprockets at both ends. The chain plate 302 is respectively attached to the transmission chain 301 on both sides of the conveyor frame 309. The movement of the chain 301 drives the movement of the chain plate 302. The conveyor fixed mounting plate 312 and... The rear end of the conveying device frame 309 is welded and fixed. The conveying device mounting support rod 308 is installed on the conveying device fixed mounting plate 312 via bearing connection. Both ends of the conveying device mounting support rod 308 are further installed at the fixed bearing connection 132 at the rear of the main support frame 13 via fixed bearings. The conveying device lifting rod 303 is fixedly welded and installed in the middle of the conveying device frame 309. Both ends of the conveying device lifting rod 303 are respectively hinged to the end of the rod of the conveying device lifting hydraulic cylinder 304. The other end of the conveying device lifting hydraulic cylinder 304 is connected to the protruding pin of the hydraulic cylinder fixed mounting plate 305. The hydraulic cylinder mounting plate 305 is fixedly installed on the connection point of the hydraulic cylinder mounting plate inside the left track drive device 11 and the right track drive device 12 by bolts. With the help of the conveying device, the hydraulic cylinder 304 is extended and shortened. The action rod of the conveying device lifting hydraulic cylinder 304 pushes the chain conveyor 3 to rotate around the central axis of the conveying device mounting support rod 308, realizing the height change of the front of the chain conveyor 3, realizing the matching of the height of the wedge shovel 300 with the soil loosening depth, and ensuring that the soil loosened by the rotary tillage excavation sampling device 2 can be smoothly conveyed upward by the chain conveyor 3.
[0027] The storage and detection device 4, as shown in Figures 7(a, b), includes a storage box 401, a box support frame 402, a scraping guide shaft 403, a box bottom plate lifting guide bearing 404, a box bottom plate lifting guide shaft 405, a box bottom plate lifting electric push rod 406, a box bottom plate lifting support plate 407, a scraping linear bearing 408, a wedge-shaped scraping blade 409, a scraping hydraulic cylinder 410, a box bottom plate lifting guide shaft fixed support bearing seat 411, an electric push rod mounting connection plate 412, and a camera. The camera and flash assembly 413, the camera and flash mounting rod 414, the nylon plate 415, and the bottom movable plate of the box 416; the wedge-shaped scraper 409 includes a scraper 4091, a linear bearing mounting point 4092, and a hydraulic cylinder mounting point 4093; the storage box 401 has two lifting slots 4011 on each side; the bottom plate lifting support plate 407 of the box has a support lug 4071, a support plate electric push rod 4072, and a bottom plate lifting guide bearing mounting hole 4073.
[0028] The storage box 401 is mounted across the rear of the left track drive unit 11 and the right track drive unit 12 via a box support frame 402, providing basic support for the box structure. Two scraping guide shafts 403 are longitudinally arranged above the box support frame 402. Two scraping linear bearings 408 are respectively fitted into the scraping guide shafts 403 on both sides. The scraping linear bearings 408 and the scraping guide shafts 403 are commonly used complete sets of equipment, enabling precise relative sliding. A wedge-shaped scraper blade 409, such as... Figure 8 As shown, its linear bearing mounting point 4092 is connected to the scraping linear bearing 408, thus the scraping linear bearings 408 on both sides and the wedge-shaped scraping plate 409 are relatively fixed. The working end of the scraping hydraulic cylinder 410 is connected to the hydraulic cylinder mounting point 4093 of the wedge-shaped scraping plate 409. The cylinder body of the scraping hydraulic cylinder 410 is fixed to the main support frame 13 through two cylinder supports 133. Finally, the extension and retraction of the working end of the scraping hydraulic cylinder 410 drives the wedge-shaped scraping plate 409 to reciprocate along the scraping guide shaft 403. The wedge-shaped scraping plate 409 is located on the upper part of the storage box 401 and retains a certain gap; the bottom plate of the box lifts the support plate 407, as shown. Figure 9As shown, its two load-bearing ears 4071 pass through the two lifting slots 4011 on one side of the storage box 401 and are connected to the bottom of the bottom moving plate 416 of the box. The electric push rod 4072 of the load-bearing plate is connected to the extended end of the electric push rod 406 of the bottom plate of the box. The box bottom plate lifting guide bearing mounting holes 4073 on both sides of the box bottom plate lifting load-bearing plate 407 are fixedly connected to the box bottom plate lifting guide bearings 404. A total of four box bottom plate lifting guide bearings 404 installed on the box bottom plate lifting load-bearing plates 407 on both sides are respectively matched with four box bottom plate lifting guide shafts 405. The box bottom plate lifting guide bearings 404 and the box bottom plate lifting guide shafts 405 are a set of devices, which can achieve precise fitting and relative sliding. The nylon plate 415 is slightly smaller than the bottom moving plate 416 of the box and is placed directly on the bottom moving plate 416 of the box. The box bottom plate lifting guide bearings 404 and the box bottom plate lifting guide shafts 405 are a set of devices, which can achieve precise fitting and relative sliding. The fixed end of the electric push rod 406 is connected and fixed to the corresponding position of the main support frame 13 through the electric push rod mounting connection plate 412, ultimately achieving the following effect: the two box bottom plate lifting electric push rods 406 move precisely and synchronously. Their contraction and extension actions drive the box bottom plate lifting support plates 407 on both sides to move up and down synchronously through the support plate electric push rod action rod 4072. The box bottom plate lifting support plates 407 drive the box bottom moving plate 416 to move up and down through the support ears 4071, and then drive the nylon plate 415 to move up and down. The existence of the lifting gap 4011 ensures that there is no obstruction to the up and down movement of the box bottom plate lifting support plates 407 and the box bottom moving plate 416. The four box bottom plate lifting guide shafts 405 are installed in parallel. Under the action of the box bottom plate lifting guide bearing 404, the smoothness of the up and down movement of the box bottom moving plate 416 and the nylon plate 415 is ensured.
[0029] II. Self-propelled Mulch Film Sampling and Testing Method This self-propelled mulch film sampling and testing method is controlled based on the mulch film testing process control logic of the self-propelled mulch film sampling and testing equipment, such as... Figure 10 As shown, it includes the following: 1) After the machine is installed, place it in the plot of land where the mulch film testing and sampling are required, and plan the sampling area and sampling path in advance; 2) Adjust the extension length of the hydraulic cylinder 22 by adjusting the rotary tillage excavation height through the hydraulic system, thereby determining the depth of rotary tillage loosening. According to the technical requirements of conventional sampling, this depth generally needs to be set at 10~20cm. During the depth adjustment process, the gasoline rotary tiller assembly 21 is opened simultaneously to ensure that the soil can be loosened during the downward movement of the rotary tillage blades and the sampling depth is set accurately. 3) The extension length of the hydraulic cylinder 304 of the conveying device is adjusted by the hydraulic system so that the depth of the wedge shovel 300 is consistent with the depth of the rotary tiller. Therefore, the chain conveying device 3 needs to move synchronously with the rotary tillage and excavation sampling device 2. This process can be achieved by setting the hydraulic system control logic. That is, the hydraulic cylinder 304 of the conveying device and the hydraulic cylinder 22 of the rotary tillage excavation height adjustment can be synchronously oiled or returned to ensure the consistency of the action. The depth of the wedge shovel 300 is consistent with the depth of the rotary tiller through reasonable structural parameter control. 4) The tracked chassis 1 moves forward, driving the rotary tillage and sampling device 2 to loosen the soil. At the same time, the conveying power motor 306 of the chain conveyor device 3 works. The conveying power motor 306 drives the rear transmission shaft 310 to rotate through the conveying power chain 307 and the corresponding sprocket. Then, through another pair of sprockets, it drives the two-sided transmission chains 301 to move, which in turn drives the transmission chain to drive the chain plate 302 to move. Thus, the soil loosened by the rotary tillage and sampling device 2 is conveyed upward through the wedge shovel 300 and the main body of the chain conveyor device 3. 5) In the storage and testing device 4, the hydraulic system controls the extension of the hydraulic cylinder 410 for scraping soil, which pushes the wedge-shaped scraper 409 to the rear of the machine. At the same time, the electric push rod 406 for lifting the bottom plate of the box extends, which pushes the bottom moving plate 416 of the box to the bottom through the action of the lifting support plates 407 on both sides of the bottom plate of the box, thus leaving enough space for storing soil. This action must be completed before soil excavation and transportation, and the storage and testing device 4 is ready to accommodate the sampled soil. 6) When the soil conveyed by the chain conveyor 3 reaches the top of the chain conveyor 3, it is turned over and poured into the box 401 of the storage and detection device 4. As the soil conveying process continues, the soil will form a "piled-up peak" in the box 401. At this time, the hydraulic system can control the scraping action hydraulic cylinder 410 to shorten to the limit position and then extend to the limit position, so that the wedge scraper 409 moves back and forth once above the box 401 to achieve the purpose of smoothing out the "piled-up peak". 7) The equipment continues to move forward along the planned path until sampling is completed, and the container 401 is filled with the sampled soil; the soil transport path is as follows: Figure 11 As shown.
[0030] 8) The equipment stops moving, and the sampling and testing phase begins. The system workflow diagram for the testing phase is as follows: Figure 10 As shown, the hydraulic system controls the shortening of the hydraulic cylinder 410 for the soil scraping action, which pulls the wedge-shaped scraper 409 closer to a position away from the tail of the machine. During the movement of the wedge-shaped scraper 409, the excess soil in the box 401 is scraped off, thereby leveling the soil on the top of the box 401. 9) In the camera and flash assembly 413, the flash moves synchronously with the camera to take the first picture of the surface soil; 10) The electric push rod 406 for lifting the bottom plate of the box shortens by a certain distance, preferably 2cm, so that the bottom moving plate 416 of the box is lifted by a certain distance (2cm) by the action of the two bottom plate lifting support plates 407 on both sides. After the lifting is completed, the electric push rod 406 for lifting the bottom plate of the box stops moving. 11) The hydraulic system controls the extension of the scraping hydraulic cylinder 410, which pushes the wedge scraper 409 to the tail of the machine. During the movement of the wedge scraper 409, the excess soil in the box 401 is scraped off, so that the soil on the top of the box 401 is leveled again. Then the scraping hydraulic cylinder 410 shortens, which pulls the wedge scraper 409 closer to the tail of the machine, in preparation for the next photo. 12) In the camera and flash assembly 413, the flash moves synchronously with the camera to take another picture of the soil on the surface. 13) Start the next lifting-scraping-photographing cycle until all the soil stored in box 401 is discharged and photographed layer by layer, completing the soil sampling work at one point.
[0031] This invention provides a specialized detection method and algorithm for image-based detection of residual plastic film, including a method for constructing a training dataset, a method for constructing a target detection model, and an image-based algorithm for actual detection of residual plastic film. Regarding the construction of the training dataset, for example... Figure 12 As shown, firstly, various types of residual plastic film from different years over the past three years were collected from a plastic film recycling station and sorted by year to represent residual plastic film in the field under actual conditions. The differences between plastic film from different years are mainly reflected in the degree of wrinkles, soiling, and color difference, providing a training basis for the algorithm to identify the amount of residual plastic film in different years. Then, the pre-made plastic film was randomly buried in the field soil, and a small rotary tiller was used to naturally turn over the soil area where the plastic film was buried to construct the real appearance of the soil in the actual test. The interference of factors such as dead branches and gravel was fully considered, and the surface layer of the soil was photographed. After the photographing, the top 2cm was removed. To increase the amount of data, the experiment was repeated until the rotary tillage depth was reached. This process was repeated until sufficient image samples were obtained. Image preprocessing was then performed, including image cropping, position correction, and noise reduction. Specialized data labeling software was used to label the training data. The labels were assigned to the year based on the different degrees of degradation, weathering, wrinkling, staining (degradation), and color difference of the residual plastic film from different years. The specific year label values were determined based on human experience and the history of plastic film use in previous plots in the region. This completed the construction of the training dataset. The model building process for object detection is as follows: Figure 13As shown, first, the training dataset is loaded, then the candidate box size is set to the size of the storage device box surface, and the training parameters of the deep neural network model are set. The deep neural network model used can adopt a relatively simple basic structure to achieve high training efficiency and prediction speed. The model grid is mainly constructed based on the following work: (1) image enhancement preprocessing, scaling and region division; (2) using convolutional neural networks to extract image features of each region; (3) fully connected layers predict the location of the residual film and the probability of the residual film appearing in each region image; (4) further identify the residual film feature type of each region, i.e., the year. During the training process, the model is considered to be completed by judging whether the loss function has converged, thereby outputting the target detection model; the specific form of the neural network model is as follows. Figure 14 As shown in Table 1, the specific parameter settings for each layer are as follows. The basic model is a series of multiple convolutional layers (conv), normalization layers (bn), and activation layers (ReLU). After the final dropout layer, a multi-classification layer is connected (merging detection and classification: 7 classes - 0 = no residual membrane, 1-5 = year, 6 = other). This model has the advantages of lightweight design, fast training loading, and fast prediction speed.
[0032] Table 1 Summary of parameters for each layer of the convolutional neural network model
[0033] For image-based algorithms for detecting actual residual plastic film, such as Figure 15 As shown, after inputting surface images of the storage device box at different depths obtained from a single detection and loading the target detection model, the location and classification of residual film in each frame image are predicted based on the trained deep neural network model, and the year of film residue is identified. Then, based on the location of the residual film in the detection image, residual film edge detection is performed, and image segmentation is performed to extract it from the background. By calculating the "quality" (pixel stacking) of the residual film image, its residual area is measured, thereby obtaining the content (residual area) of residual film in each year in each frame image, and further converting it into the actual mass of residual film. Secondly, data synthesis was performed to obtain the residual amount of plastic film from different years corresponding to the soil samples in the storage device box obtained from a single sampling. Data conversion was then performed, transforming the sampling area corresponding to a single sampling into a unit area, thus obtaining the residual amount of plastic film in each year and the total residual amount of plastic film per square meter of the plot obtained from a single sampling. The average of multiple sampling and evaluation results was then used to obtain the final values of the residual plastic film in the most recent year and the total amount for the plot. Overall, the residual film mass per unit area... m The theoretical calculation formula is shown below:
[0034] In the formula, , F Total number of soil stratification images (total number of frames). It is the first f The number of residual films detected in the frame, It is the first f The first frame of the image i The pixel area of the residual film; It refers to the thickness of the residual film in that year. The density of the residual film in that year was obtained through prior actual measurement. k This is the residual film area conversion factor, i.e., the actual residual film area. residual film area conversion factor k In practice, the value is generally taken as 1.2 to 1.4.
[0035] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A self-propelled mulch film sampling and testing equipment, characterized in that, include: The equipment chassis is used to support various working devices and provide the power for the entire machine to move; A rotary tillage and excavation sampling device is installed on the chassis of the equipment and is used to loosen the soil at a specified depth on the ground during the movement of the equipment. A conveying device, installed on the chassis of the equipment and located behind the rotary tillage excavation and sampling device, is used to convey the loosened soil upwards. A storage and testing device, mounted on the equipment chassis and located at the discharge end of the conveying device, is used to receive, store, and test the conveyed soil, and includes: Storage container, used to receive and store soil; The bottom plate lifting mechanism of the storage box is connected to the bottom moving plate of the storage box, and can drive the bottom moving plate of the storage box and the soil placed on it to be lifted sequentially in the vertical direction. A soil scraping mechanism is installed above the storage box and is used to cooperate with the bottom plate lifting mechanism of the box to scrape the surface layer of soil that is lifted layer by layer in the storage box. An image acquisition mechanism, located above the storage box, is used to acquire images of the leveled soil surface for subsequent detection and analysis.
2. The self-propelled mulch film sampling and testing equipment according to claim 1, characterized in that, The rotary tillage excavation sampling device includes: The gasoline rotary tiller assembly is rotatably connected to the chassis of the equipment. A rotary tillage height adjustment hydraulic cylinder is installed on the equipment chassis and connected to the gasoline rotary tiller assembly. It is used to drive the gasoline rotary tiller assembly to rotate vertically to adjust its tillage depth.
3. The self-propelled mulch film sampling and testing equipment according to claim 2, characterized in that, The conveying device includes: The conveying device body is rotatably connected to the equipment chassis; The lifting hydraulic cylinder of the conveying device is installed on the chassis of the equipment and connected to the body of the conveying device. It is used to adjust the height of the front end of the conveying device body so that its lowering depth is adapted to the lowering depth of the rotary tiller blades of the gasoline rotary tiller assembly.
4. The self-propelled mulch film sampling and testing equipment according to claim 1, characterized in that, The box bottom plate lifting mechanism includes: An electric push rod for lifting the bottom plate of the box is installed on the equipment chassis and connected to the movable plate at the bottom of the box. The bottom plate of the box is lifted and supported by a load-bearing plate, which is connected to the output end of the electric push rod for lifting the bottom plate of the box, and is connected to the bottom moving plate of the box through the lifting seam on the side wall of the storage box.
5. The self-propelled mulch film sampling and testing equipment according to claim 1, characterized in that, The storage detection device further includes: The image processing unit is configured to identify and classify residual plastic film from different years in the images based on the acquired multi-layer soil surface images using a pre-trained deep neural network model, and calculate the quality of residual plastic film from different years based on the identification results.
6. The self-propelled mulch film sampling and testing equipment according to claim 5, characterized in that, The pre-trained deep neural network model is a multi-class target detection model trained based on the differences in wrinkles, dirt, and color difference of residual plastic film from different years.
7. The self-propelled mulch film sampling and testing equipment according to claim 5, characterized in that, The image processing unit is further configured to: perform edge detection and image segmentation on the image area where residual plastic film is identified, extract the residual film image and calculate its pixel area, and calculate the mass of residual plastic film for each year based on the sum of the pixel areas of plastic film fragments from each year, combined with preset residual film thickness, density parameters and area conversion coefficients.
8. A self-propelled mulch film sampling and testing method based on the testing equipment according to any one of claims 1-7, comprising the following steps: S1: The loosened soil is transported to the storage and testing device for storage; S2: Gradually lift the soil inside the box and scrape the soil surface layer, and collect images of the soil surface layer after each scraping. S3: Input the acquired multi-layer soil surface images into a pre-trained deep neural network model to identify and classify residual plastic film from different years in the images; S4: Based on the identification results, calculate the mass of residual plastic film from different years within a single sampling area.
9. The self-propelled mulch film sampling and detection method according to claim 8, characterized in that, In step S3, the pre-training of the deep neural network model includes the following steps: Constructing a training dataset: Acquire and sort residual plastic film samples from different years; randomly bury the samples in the soil and turn them over to simulate field conditions; remove the topsoil layer by layer in the buried soil area and take repeated photos until the rotary tillage depth is reached to obtain a series of soil images at different depths; label the images according to the year of the plastic film samples and their corresponding wrinkles, dirt, and color difference features to construct a training dataset. Model training: The deep neural network model is trained based on the training dataset.
10. The self-propelled mulch film sampling and detection method according to claim 8, characterized in that, Step S4 specifically also includes: S4.1: Perform image segmentation on the identified plastic film fragments in each image and calculate their pixel area; S4.2: Based on the total pixel area of the plastic film fragments from each year, combined with the preset residual film area conversion coefficient, the thickness and density of the plastic film in that year, calculate the residual mass of plastic film per unit area for each year.