A processing method, system and program product for pulp slurry defect screening

Abstract

The application relates to the field of vegetable and fruit processing, and provides a processing method, a system and a program product for screening defects of fruit pulp slurry, which comprises the following steps: spreading the fruit pulp slurry on the surface of a conveying belt and continuously conveying; photographing the slurry image at a visual detection station, identifying defects through a visual recognition algorithm, and recording the position information of the defects on the conveying belt plane; determining the execution time sequence of each defect reaching a vacuum suction station according to the position information, and generating an action sequence table containing the starting sequence and starting time of each vacuum suction head; and starting the corresponding vacuum suction head according to the action sequence table, and performing pinpoint suction on the defects. The application realizes accurate positioning and pinpoint screening of defects of the wet fruit pulp slurry, avoids waste of qualified materials, and improves the yield of raw materials.

Description

Technical Field

[0001] This application relates to a treatment method, system, and procedure for screening defects in fruit pulp for use in the field of fruit and vegetable processing. Background Technology

[0002] In the fruit and vegetable deep processing industry, after raw materials are pressed or crushed to extract juice, a large amount of fruit pulp rich in dietary fiber, flavor substances, and nutrients is produced as a byproduct. This fruit pulp is typically used as a food ingredient, beverage base, or dietary fiber supplement for further processing. However, due to factors such as the quality of incoming materials and processing techniques, various undesirable components or foreign objects inevitably contaminate the fruit pulp, including discolored fruit pulp with abnormal color, non-fruit pulp skin tissue, black spots caused by mold, and non-fruit pulp foreign objects introduced during harvesting or processing, such as branches, leaves, and insects. The presence of these defective components will seriously affect the sensory quality, taste uniformity, and food safety of the final product. Therefore, before the fruit pulp enters further processing, an efficient online detection and separation process is needed to effectively remove these defective components.

[0003] Currently, the screening and grading technologies for solid fruit and vegetable raw materials are relatively mature. Existing technologies typically employ machine vision-based detection systems to acquire images and identify features of individual items (such as whole fruits, nuts, or sliced ​​materials) arranged sequentially on the conveyor channel, and classify them according to preset grading rules such as color, size, and shape. The rejection mechanisms often employ instantaneous mechanical force, such as using high-pressure airflow to directionally blow the target individual, using a swing arm to strike it, or using contact components such as push rods or forks to push the target individual away from its original conveying trajectory, thereby separating defective materials from qualified materials, or allowing materials of different grades to enter designated collection channels. Meanwhile, in the field of fruit pulp conveying and processing, there is also a defective material rejection scheme based on the principle of physical diversion. This scheme, upon detecting a defective area in the fruit pulp on the conveyor channel, activates a diversion mechanism located downstream of the material channel, switching the entire section of pulp containing the defective area to the waste collection channel, thereby separating the defective material.

[0004] However, the aforementioned prior art has at least the following technical problems: First, because the pulpy pulp produced after juicing is wet and has a certain viscosity and fluidity, it usually needs to be spread evenly on the conveyor belt surface of a certain thickness for transport in actual production. Its material form is completely different from discrete rigid solids. Therefore, the "air blowing" or "impact" screening methods mentioned above for individual fruits and vegetables cannot be effectively applied to wet pulp layers with adhesiveness and fluidity. Forcing their use can easily lead to pulp splashing, deformation, or failure to overcome adhesion to achieve separation, so they are not suitable for the fine screening of this specific pulpy material.

[0005] Secondly, regarding the aforementioned slurry defect removal scheme based on the principle of physical diversion, due to the localized and discrete distribution of defects within the slurry layer, this scheme results in a large amount of qualified slurry surrounding the defects being discharged as waste during the diversion operation. This not only causes a significant waste of qualified materials, significantly reducing the comprehensive utilization rate of raw materials and product yield, but also increases the burden of subsequent wastewater treatment, impacting the economic efficiency of production. Summary of the Invention

[0006] The purpose of this application is to overcome the shortcomings of the prior art and provide a treatment method, system and program product for screening defects in fruit pulp, which can accurately identify and precisely remove defects and impurities in fruit pulp.

[0007] Firstly, this application provides a method for screening defects in fruit pulp, the technical solution of which includes the following steps: S100, the pulp is spread evenly on the surface of the conveyor belt, and the conveyor belt is controlled to continuously convey the pulp at a preset speed; S200, when the conveyor belt transports the pulp to the visual inspection station, the pulp is photographed, defects in the pulp are identified by a visual recognition algorithm, and the position information of each defect on the conveyor belt plane is recorded; the position information includes a first coordinate along the length direction of the conveyor belt and a second coordinate along the width direction of the conveyor belt. S300, based on the location information, determines the execution sequence of each defect arriving at the vacuum suction station, specifically including: Based on the first coordinate, the timing of each defect reaching the corresponding vacuum suction station is calibrated; Based on the second coordinate, each defect is assigned to and calibrated to a plurality of vacuum suction head channels arranged along the width of the conveyor belt; Based on the calibration results, an action sequence table containing the start-up order and timing of each vacuum head is generated; S400, according to the action sequence table, when the pulp material including the defect arrives at the vacuum suction station, the vacuum suction head of the corresponding vacuum suction head channel is controlled to start, so as to perform targeted suction of the defect in the pulp material on the conveyor belt.

[0008] By adopting the above technical solution, this application establishes an accurate mapping of the spatial location of fruit pulp defects in the visual inspection station and forms an action sequence table for vacuum suction head activation. This enables the precise and targeted extraction of local defect areas by a single row of vacuum suction heads, effectively meeting the real-time defect removal requirements in high-speed continuous production scenarios.

[0009] Preferably, S100 further includes a preliminary step on the characteristics of the fruit pulp: before or during the feeding of the fruit pulp, the solid content and / or fluid viscosity of the batch of fruit pulp are detected; in S400, the suction time after the vacuum suction head is started is adjusted according to the detection results of the solid content and / or fluid viscosity of the fruit pulp.

[0010] By adopting the above technical solution and introducing a priori detection and dynamic control mechanism for the solid content and fluid viscosity of the fruit pulp, this solution can adaptively match the absorption energy input according to the physical characteristics of the fruit pulp itself. This avoids the problems of defective pulp being missed due to insufficient absorption time or fruit pulp adhering to the inner wall of the vacuum suction head, as well as the problem of excessive absorption of undefective and intact fruit pulp due to excessive absorption time. This improves the system's universal adaptability to different batches and varieties of fruit pulp and the stability of the defect screening process.

[0011] Preferably, in S100, the preset speed of the conveyor belt is adjusted according to the suction duration of the vacuum suction head; in S200, the shooting interval for shooting the fruit pulp is adjusted according to the preset speed of the conveyor belt.

[0012] By adopting the above technical solution, the running speed of the conveyor belt is matched with the suction time of the vacuum suction head, ensuring that the conveying speed of the pulp on the conveyor belt and the suction time of the vacuum suction head can be adapted. At the same time, the shooting interval is adjusted synchronously with the speed of the conveyor belt, ensuring that each frame of image can be seamlessly connected in the length direction of the conveyor belt. This avoids missed detections due to too low shooting frequency, and also avoids regional overlap and redundant data due to too high shooting frequency.

[0013] Preferably, the drive shaft of the conveyor belt is equipped with an encoder, which is used to output periodic pulse signals generated as the drive shaft rotates; In S300, the step of calibrating the timing of each defect reaching directly below the corresponding vacuum suction head based on the first coordinate specifically includes: Based on the pre-calibrated distance between the visual inspection station and the vacuum suction station and the first coordinate, the countdown value of the pulse signal when each defect reaches directly below the corresponding vacuum suction head is calibrated, and the countdown is performed based on the periodic pulse signal; when the countdown value of the pulse signal returns to zero, it is the start-up time of the corresponding vacuum suction head in the action sequence table.

[0014] By adopting the above technical solution, the encoder pulse signal directly reflects the actual physical displacement of the conveyor belt. Using this as a time reference for the reciprocal calibration of the start-up timing, the problem of inconsistency between theoretical speed and actual displacement caused by factors such as output power fluctuations or speed regulation deviations of the conveyor belt can be effectively overcome. Thus, the process of defect transmission from the visual inspection station to the vacuum suction station can be accurately tracked in the form of a displacement pulse chain. Compared with calibration schemes that rely solely on time delay or speed conversion, this can significantly improve the synchronization accuracy between the action triggering timing and the actual arrival position of the material.

[0015] Preferably, in S400, when the pulp containing the defect arrives at the vacuum suction station according to the action sequence table, the step of controlling the vacuum suction head of the corresponding vacuum suction head channel to start also includes setting an advance amount based on the inherent response time of the vacuum suction head and its driving mechanism, and subtracting the advance amount from the start time in the action sequence table to obtain the actual start time for controlling the corresponding vacuum suction head; the advance amount is adjusted according to the verification results of vacuum suction.

[0016] By adopting the above technical solution and introducing advance compensation, the solution eliminates the inherent response delay caused by the mechanical action delay of the solenoid valve, so that the actual suction action of the suction head is precisely aligned with the arrival time of the defect. Furthermore, the verification and adjustment mechanism further endows the system with the ability to optimize compensation accuracy online, thereby significantly improving the stability and controllability of defect screening in engineering practice.

[0017] Preferably, the visual recognition algorithm is a target detection model or image classification model based on a neural network; the neural network includes one of the YOLO series models, CNN, CNN-Transformer, or a network structure based on the above models and improved thereon.

[0018] By adopting the above technical solution and selecting a visual neural network model that combines local feature extraction and global context modeling capabilities, we can effectively cope with the task of identifying minute defects in complex backgrounds, providing a reliable technical guarantee for achieving high accuracy and low false negative rate in defect identification and localization.

[0019] Preferably, the defects include one or more of the following: discolored pulp, peel or peel fragments, leucocephala or leucocephala fragments, seeds or seed fragments, black spots, mold, and non-pulp foreign matter.

[0020] By adopting the above technical solution, the above defect types cover the defective components that are common in the actual production process of fruit pulp and have the most significant impact on the quality of the final product. The technical solution limits the actual problems and the adaptability boundaries of the solution in the field of fruit pulp screening.

[0021] Secondly, this application provides a processing system for screening defects in fruit pulp, used to implement the above-mentioned processing method, and employing the following technical solution, including: The material conveying module includes a feeding device, a conveyor belt, and an encoder; the feeding device is used to spread fruit pulp evenly on the surface of the conveyor belt; the encoder is mounted on the drive shaft of the conveyor belt and is used to output pulse signals. A visual inspection module, located at the visual inspection station, includes an image acquisition device and a visual processing unit; the image acquisition device is used to photograph the pulp on the conveyor belt; the visual processing unit is used to identify defects in the pulp through a visual recognition algorithm and output the location information of the defects. A vacuum suction module is installed at the vacuum suction station and includes multiple vacuum suction heads arranged along the width of the conveyor belt, a vacuum pump that provides negative pressure to each vacuum suction head, a solenoid valve group that controls the air passage opening and closing of each vacuum suction head, and a waste liquid tank connected to each vacuum suction head. The PLC controller is connected to the material conveying module, the vision inspection module, and the vacuum suction module, respectively. It is used to receive the pulse signal and the location information of the defect, control the corresponding vacuum suction head to perform fixed-point suction of the defect, and control the conveying speed of the conveyor belt, the shooting interval of the vision inspection module, and the on / off duration of the solenoid valve group.

[0022] Preferably, the vacuum suction module further includes a vertical displacement mechanism connected to each of the vacuum suction heads. The vertical displacement mechanism is used to drive the corresponding vacuum suction head to descend to a position close to the surface of the conveyor belt when the defect reaches the vacuum suction station, and to control the vacuum suction head to reset after vacuum suction is completed.

[0023] Thirdly, this application provides a computer program product, including a computer program or instructions, wherein the technical solution adopted is that when the computer program or instructions are executed by a processor, they implement the steps of the processing method described above.

[0024] In summary, this application includes at least one of the following beneficial technical effects: 1. This application addresses the specific material form of pulp after juicing by employing vacuum suction as the power source for defect removal, achieving effective separation of the pulp layer. Simultaneously, by establishing a precise mapping relationship between the spatial location of defects and the actuator, it enables targeted and accurate suction of localized defect areas, fundamentally avoiding the problem of discarding large amounts of qualified pulp along with defects, and significantly improving the comprehensive utilization rate of raw materials and product yield.

[0025] 2. This application introduces a priori detection steps for the solid content and / or fluid viscosity of fruit pulp, and dynamically adjusts the suction duration after the vacuum suction head is started based on the detection results. Furthermore, the preset speed of the conveyor belt is adjusted in conjunction with the suction duration, and the shooting interval of the visual inspection module is also synchronously adapted to the conveyor belt speed. Through the logical link of material characteristic perception—suction duration determination—conveyor speed adjustment—shooting interval adaptation, a dynamic and coordinated control mechanism for fruit pulp suction is formed. This allows the system to adaptively match the varying suction and separation difficulties caused by differences in solid content and fluid viscosity in different batches, varieties, and states of fruit pulp, avoiding mismatches between the vacuum suction head's suction duration and the fruit pulp, thereby significantly improving the system's versatility and operational stability in diverse production scenarios.

[0026] 3. This application uses an encoder mounted on the conveyor belt drive shaft to generate periodic pulse signals, which are used to calibrate the conveyor belt's conveying distance. The countdown value of the pulse signal when each defect reaches directly below the corresponding vacuum suction head is used, and the start-up timing of the corresponding vacuum suction head is determined when the countdown value reaches zero. This mechanism directly reflects the actual physical displacement of the conveyor belt using the encoder pulse signal, effectively overcoming the problem of inconsistency between theoretical speed and actual displacement caused by factors such as power fluctuations or speed regulation deviations, thus achieving precise defect tracking. Furthermore, advance compensation is set according to the inherent response time of the vacuum suction head and its drive mechanism, ensuring precise alignment between the actual suction action of the vacuum suction head and the arrival time of the defect.

[0027] 4. In view of the characteristics of defects in fruit pulp, such as small size, irregular shape, and subtle differences in texture and color from the normal fruit pulp background, this application adopts a target detection model or image classification model based on neural network as a visual recognition algorithm. It can effectively deal with the task of detecting small defects in complex backgrounds, and provide high accuracy and low false negative rate defect identification and localization support for online detection, thereby ensuring the effective execution of subsequent fixed-point absorption process and the engineering feasibility of the overall solution. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the architecture of a processing system for screening defects in fruit pulp, as described in an embodiment of this application. Figure 2This is a schematic flowchart of a method for screening defects in fruit pulp according to an embodiment of this application; Figure 3 This is a schematic diagram of the camera field of view of the visual inspection module 200 in a processing system for screening defects in fruit pulp according to an embodiment of this application. Figure 4 This is a detailed architectural diagram of a processing system for screening defects in fruit pulp, as described in an embodiment of this application. Detailed Implementation

[0029] The specific embodiments are merely explanations of this application and are not intended to limit it. After reading this specification, those skilled in the art can make modifications to these embodiments without contributing any inventive step, but such modifications are protected by patent law as long as they are within the scope of this application.

[0030] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. It should be noted that in the optional embodiments of this application, the object information and other related data involved require the permission or consent of the object when the embodiments of this application are applied to specific products or technologies, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. That is to say, if the embodiments of this application involve data related to the object, it needs to be obtained with the authorization and consent of the object, the authorization and consent of the relevant departments, and in compliance with the relevant laws, regulations, and standards of the country and region. If personal information is involved in the embodiments, the acquisition of all personal information requires the consent of the individual. If sensitive information is involved, the separate consent of the information subject is required, and the embodiments also need to be implemented with the authorization and consent of the object.

[0031] The embodiments of this application will now be described in further detail with reference to the accompanying drawings. Example 1:

[0032] Please see Figure 1 and Figure 2 This embodiment provides a method for screening defects in fruit pulp. This method relies on the coordinated operation of a material conveying module 100, a vision inspection module 200, a vacuum suction module 300, and a PLC controller 400 to achieve precise positioning and targeted screening of defective components in wet fruit pulp. Specifically, the method includes the following steps: S100: The pulp is spread evenly on the surface of the conveyor belt, and the conveyor belt is controlled to continuously convey the pulp at a preset speed.

[0033] In this embodiment, the material conveying module 100 evenly spreads the fruit pulp from the storage tank onto the conveyor belt, forming a pulp layer of uniform thickness. This even spreading is a prerequisite for accurate imaging during subsequent visual inspection, avoiding occlusion or shadow interference caused by pulp accumulation. The conveyor belt runs continuously at a preset speed driven by a motor, carrying the pulp layer through the visual inspection station and the vacuum suction station in sequence.

[0034] More specifically, S100 also includes a preliminary step on the characteristics of the pulp. Before or during the feeding of the pulp, the solids content and / or fluid viscosity of the batch of pulp are tested.

[0035] In actual industrial production, different types of fruit pulp, or different batches of the same type of fruit pulp, or even different stages of the same batch of fruit pulp, often have significantly different fluid viscosity characteristics, and the fluid viscosity of fruit pulp is significantly correlated with the solid content of fruit pulp.

[0036] Meanwhile, the suction action of the subsequent vacuum suction head also needs to be adapted to the fluid viscosity of the fruit pulp. Specifically, when the fluid viscosity of the fruit pulp is high, it is more difficult to change its position and movement state. Therefore, the vacuum suction head needs to use a longer suction action to ensure that the defective fruit pulp is sucked away from the conveyor belt and that there is sufficient time to transport it to the waste collection point, so as not to stick to the edge and inner wall of the vacuum suction head, causing contamination or blockage.

[0037] One specific implementation involves a pre-installed model mapping the viscosity of the fruit pulp to the suction time of the vacuum suction head. When high viscosity or high solids content is detected, the system automatically extends the opening and holding time of the vacuum suction head's solenoid valve assembly to ensure the vacuum suction head has a sufficient working window to completely remove the target defects. Conversely, under low viscosity conditions, the suction time is appropriately shortened to avoid excessive suction, unnecessary energy waste, or disturbance to surrounding qualified pulp. This dynamic control mechanism effectively solves the problem of unstable defect removal rate caused by fluctuations in material properties.

[0038] In another embodiment, the power of the vacuum suction head can also be adjusted synchronously.

[0039] To obtain the solids content and / or fluid viscosity of fruit pulp, various detection methods can be employed. For example, an online viscometer or refractometer can be installed at the outlet of the storage tank of the feeding device to collect rheological data of the pulp flowing through it in real time; alternatively, offline sampling can be used, where operators periodically sample and measure the solids content or viscosity grade of the current batch via a human-machine interface. Regardless of the detection method used, the core purpose is to provide the system with a priori benchmark value reflecting the current material handling difficulty, enabling the control system to perceive changes in the material state and guide the selection of subsequent process parameters. The above-mentioned methods for collecting the physical properties of the solids content and / or fluid viscosity of fruit pulp are existing mature technologies and will not be elaborated upon here.

[0040] It should also be noted that, based on the solids content and / or fluid viscosity of the obtained fruit pulp, the primary process parameter to control is the suction time of the vacuum suction head. Simultaneously, the preset speed of the conveyor belt also needs to be adjusted accordingly to match the suction time of the vacuum suction head. If the conveyor belt speed is too fast, the vacuum suction head may not be able to complete the suction of the target defect area of ​​the fruit pulp, as the conveyor belt has already transported the pulp backward, resulting in insufficient retention time for defect removal and potential defect residue. Conversely, if the conveyor belt speed is too slow, the vacuum suction head may suction a large amount of intact, defect-free fruit pulp, leading to waste. One method to adjust the preset speed of the conveyor belt is to establish a mapping table relating the conveyor belt speed to the suction duration of the vacuum suction head, enabling autonomous adjustment of the conveyor belt speed based on this mapping relationship.

[0041] S200: When the conveyor belt transports the pulp to the visual inspection station, the pulp is photographed, and defects in the pulp are identified by a visual recognition algorithm. The position information of each defect on the conveyor belt plane is recorded. The position information includes a first coordinate along the width direction of the conveyor belt and a second coordinate along the length direction of the conveyor belt.

[0042] Specifically, the vision inspection module 200 is located at the vision inspection station and can be captured using an industrial camera. When the pulp layer enters the camera's field of view, the industrial camera is triggered to capture a high-resolution image. The image is then input into a pre-trained visual recognition algorithm for frame-by-frame analysis.

[0043] As a preferred implementation, visual recognition algorithms can be based on neural network-based object detection models or image classification models, such as the YOLO series, CNN, CNN-Transformer, or improved network structures based on the above models to adapt to the complex texture backgrounds and non-standard defect morphologies of fruit pulp. Among them, the YOLO series, as a single-stage object detection algorithm, has achieved a good balance between real-time performance, detection accuracy, and computational efficiency, and has been widely used in the field of fruit defect detection, including subsequent iterations such as YOLOv5, YOLOv7, YOLOv11, and YOLOv12.

[0044] All the aforementioned models employ a two-stage implementation strategy of general pre-training and domain-specific fine-tuning. First, pre-trained weights on a general dataset are loaded, enabling the model to extract general low-level visual features such as edges, textures, and colors. Second, defect image samples from fruit pulp production sites are collected. After manually labeling the defect categories and locations, a dedicated dataset is constructed. The pre-trained model is then fine-tuned, freezing the shallow network parameters and retraining only the deep networks to adapt them to the unique textures and defect morphologies of fruit pulp. Identified defect types can include one or more of the following: discolored pulp, peel or peel fragments, white peel or white peel fragments, seeds or seed fragments, black spots, mold, and non-pulp foreign objects. The fine-tuned model is then deployed to an industrial control computer or edge computing device at the visual inspection station to perform real-time inference on images captured by industrial cameras, outputting the category and location coordinates of each defect.

[0045] After detecting a defect, the system establishes a two-dimensional coordinate system with the conveyor belt's direction of travel as the horizontal axis and the width direction as the vertical axis. It records the first coordinate (X value) and the second coordinate (Y value) of the centroid of each defect in this coordinate system, thereby realizing the mapping from the image pixel space to the physical conveyor space and providing a data foundation for subsequent precise execution.

[0046] Please see Figure 3 This diagram illustrates the field of view of the industrial camera in the vision inspection module 200 of this embodiment. The center of the camera's field of view is taken as the origin O, and the coordinates of the defect are the positive and negative offset values ​​of the origin in the X and Y directions. Those skilled in the art will understand that this coordinate system is merely one implementation method of this embodiment and should not be construed as limiting the scope of protection of this application. It should also be noted that the speed adjustment of the conveyor belt based on the detection results of the solid content and / or fluid viscosity of the fruit pulp must be synchronously transmitted to the shooting interval of the industrial camera in the vision inspection module 200. When the conveyor belt speed decreases, the distance traveled by the conveyor belt per unit time shortens. If the shooting frequency remains unchanged, this will result in excessive overlap of the field of view between adjacent frames, causing repeated location of defects and generating a large amount of redundant data, increasing the computational load of the vision processing unit. Conversely, if the belt speed increases but the shooting frequency does not keep up, blind spots will appear in the field of view, resulting in missed detections. Since the distance of the industrial camera's field of view along the length of the conveyor belt can be pre-calibrated, the timing of the next frame shooting by the vision inspection module 200 can be accurately calculated by combining the conveyor belt's speed or distance. This ensures that regardless of the conveyor belt's speed, the vision inspection module 200 can continuously scan the surface of the fruit pulp in a continuous, end-to-end manner.

[0047] This multi-parameter linkage adjustment strategy not only ensures the success rate of defect screening under high viscosity conditions, but also takes into account the production efficiency under low viscosity conditions. At the same time, it maintains the integrity and accuracy of visual inspection across the entire speed range, demonstrating the system's deep adaptability to complex working conditions.

[0048] S300 determines the execution sequence of each defect arriving at the vacuum suction station based on the location information. This includes the following steps.

[0049] First, based on the second coordinate, each defect is assigned to and calibrated to one of multiple vacuum suction head channels arranged along the width of the conveyor belt. Specifically, the vacuum suction module 300 has N vacuum suction heads arranged in the width direction, each vacuum suction head corresponding to an independent channel. The system determines which vacuum suction head's effective coverage area the defect falls within based on its Y-value, thus marking the defect as belonging to a specific numbered vacuum suction head channel.

[0050] Secondly, based on the first coordinate, the timing of each defect reaching the corresponding vacuum suction station is calibrated. Since the distance between the visual inspection station and the vacuum suction station—that is, the distance from the center point of the camera's field of view to the vacuum suction station (e.g., the center line of the vacuum suction head's installation position)—is a fixed value that can be obtained through pre-calibration, the actual positional deviation of the defect from the vacuum suction station can be converted by increasing or decreasing the first coordinate X value. Simultaneously, combined with the conveyor belt's speed or distance, the timing of the target defect reaching the vacuum suction station can be determined.

[0051] Finally, based on the calibration results of the second and first coordinates, an action sequence table containing the start-up order and timing of each vacuum suction head is generated. This action sequence table is essentially a timing control queue, and its data structure can include fields such as suction head ID, trigger time, and suction duration. For example, an entry in the table might record "Vacuum suction head 3, activated at T+30s, suction duration 100ms". If the defect location spans the channels of adjacent vacuum suction heads, the two adjacent vacuum suction heads can be activated synchronously at a preset time point; if the defect area is large, the suction duration can be increased. For the suction time of a single small-area defect, as mentioned above, it can be adjusted accordingly based on the detection results of the solid content and / or fluid viscosity of the pulp.

[0052] Through the above approach, the solution in this embodiment transforms spatially discrete defects into control commands arranged in an orderly manner along the time axis.

[0053] S400, according to the action sequence table, when the pulp material including defects arrives at the vacuum suction station, the vacuum suction head of the corresponding vacuum suction head channel is activated to perform targeted suction of defects in the pulp material on the conveyor belt.

[0054] Specifically, the PLC controller 400 reads the action sequence table in real time and monitors the operating status of the conveyor belt. When it is determined that a defect is about to reach the vacuum suction station, the PLC controller 400 sends an opening signal to the corresponding solenoid valve group, causing the negative pressure generated by the vacuum pump to act instantaneously on the designated vacuum suction head. The sucked-up defective pulp then detaches from the surface of the conveyor belt and enters the waste liquid tank for collection through the pipeline, while the surrounding qualified pulp continues to be conveyed forward unaffected.

[0055] The fixed-point suction mechanism of this application embodiment can retain qualified raw materials to the greatest extent, significantly improving the yield. At the same time, since the action sequence table is generated based on precise location information, even if multiple defects are located at different width positions at the same time, the system can achieve synchronous defect screening by triggering multiple suction heads in parallel, ensuring that production efficiency does not decrease due to the increase in the number of defects. Example 2:

[0056] This embodiment further refines the timing control logic in S300 and S400 to address the issues of fluctuations in conveyor belt motor power due to power supply influences or adjustments in conveyor belt speed during the process.

[0057] Specifically, the conveyor belt is equipped with an encoder on its drive shaft, which is used to output periodic pulse signals generated as the drive shaft rotates.

[0058] In this embodiment, the location tracking of defects is directly linked to the actual physical displacement of the conveyor belt. Since the encoder is mounted on the drive shaft, the number of pulses it outputs is strictly proportional to the actual distance the conveyor belt travels. Therefore, regardless of whether the conveyor belt experiences instantaneous speed fluctuations due to motor speed regulation or power fluctuations, the pulse count can accurately reflect how far the material has moved, thereby fundamentally eliminating the timing accumulation error caused by speed uncertainty.

[0059] Therefore, in S300, the timing of each defect reaching the corresponding vacuum suction head is calibrated based on the first coordinate. Specifically, this includes: calibrating the countdown value of the pulse signal when each defect reaches the corresponding vacuum suction head based on the pre-calibrated distance between the visual inspection station and the vacuum suction station and the first coordinate, and counting down based on the periodic pulse signal. When the countdown value of the pulse signal reaches zero, it is the start-up timing of the corresponding vacuum suction head in the action sequence table.

[0060] Specifically, during the commissioning phase, the physical distance between the center of the camera's field of view and the vacuum suction station is first determined in advance, and converted into a fixed reference pulse count based on the encoder's resolution. When the vision processing unit identifies a defect and outputs its first coordinate (i.e., the pixel distance away from the center of the field of view along the conveyor belt length direction), the PLC controller 400 converts this pixel distance into an offset pulse count, and then subtracts the offset pulse count from the reference pulse count to obtain the countdown value of the pulse signal specific to that defect. Subsequently, the PLC controller 400 uses the encoder's pulse signal as a clock source to count down the countdown value of the pulse signal in real time until it reaches zero, triggering the vacuum suction head to start. This countdown-to-zero triggering mechanism essentially converts the target position in space into the remaining count in the counter, so that the triggering action only occurs when the defect physically reaches the suction point precisely, achieving true displacement synchronization.

[0061] In another embodiment, during the process of controlling the start-up of the corresponding vacuum nozzle according to the action sequence table in S400, the method further includes setting an advance amount according to the inherent response time of the vacuum nozzle and its drive mechanism. For the start-up timing in the action sequence table, the advance amount is deducted as the actual start-up timing for controlling the corresponding vacuum nozzle.

[0062] Although encoders solve the problem of accurate displacement tracking, there is still a significant physical delay at the actuator end. For example, it takes about 10ms-20ms for the solenoid valve assembly to fully open after receiving an electrical signal, and about 5ms-15ms for the negative pressure to be established in the air path of the vacuum suction head and transmitted to the suction head inlet. If a vertical displacement mechanism is equipped, its downward movement also takes a certain amount of time. Without compensation, when the countdown reaches zero and the command is issued, the defect has actually already passed the optimal suction position, resulting in defect screening failure or the removal of too much qualified slurry.

[0063] Therefore, this embodiment of the application presets a lead time before the theoretical trigger point to compensate for physical delay. In this embodiment, since the vacuum suction head's action sequence table uses pulse count as a calibration parameter for conveyor belt displacement, this lead time is also calibrated using a preset pulse count to ensure alignment of control dimensions. More importantly, this lead time is not static but adjustable. Specifically, the system can determine whether defects have been accurately removed by manual inspection or by using a high-speed camera installed downstream of the vacuum suction station to capture images of the defect removal effect. If there is a deviation, the system observes whether the removal position is in front of or behind the defect and then fine-tunes the lead time parameter. For example, if continuous defect removal lag is detected, the lead time value is increased; conversely, if the suction head is found to move too early, causing the suction head to pick up qualified material in front, the lead time is decreased. This closed-loop calibration mechanism based on measured feedback ensures that the system maintains optimal timing matching during long-term operation, effectively overcoming response characteristic drift caused by component aging, air pressure fluctuations, or mechanical wear, and significantly improving the system's robustness and environmental adaptability. Example 3:

[0064] like Figure 1 As shown, this embodiment provides a processing system for screening defects in fruit pulp, which provides a specific hardware carrier and execution basis for the processing method in the aforementioned embodiments. The processing system includes a material conveying module 100, a vision inspection module 200, a vacuum suction module 300, and a PLC controller 400.

[0065] Specifically, such as Figure 4 As shown, the material conveying module 100 includes a feeding device 110, a conveyor belt 120, and an encoder 130. The feeding device 110 is used to spread the fruit pulp evenly on the surface of the conveyor belt 120 to form a pulp layer of uniform thickness; the encoder 130 is mounted on the drive shaft of the conveyor belt 120 and is used to output a pulse signal that strictly corresponds to the actual displacement of the conveyor belt, providing a high-precision position reference for the system. The material conveying module 100 may also include a detection module 140 for detecting the solid content and / or fluid viscosity of the fruit pulp.

[0066] A vision inspection module 200 is installed at the vision inspection station and includes an image acquisition device 210 and a vision processing unit 220. The image acquisition device 210 is used to capture high-resolution images of the fruit pulp on the conveyor belt 120, for example, a high-resolution industrial camera can be used. The vision processing unit 220 is used to identify defects in the fruit pulp using a vision recognition algorithm and output the location information of the defects.

[0067] The vacuum suction module 300 is installed at the vacuum suction station and includes multiple vacuum suction heads 310 arranged along the width direction of the conveyor belt 120, a vacuum pump 320 that provides negative pressure to each vacuum suction head 310, a solenoid valve group 330 that controls the air passage opening and closing of each vacuum suction head 310, and a waste liquid tank 340 connected to each vacuum suction head 310.

[0068] The PLC controller 400 is connected to the material conveying module 100, the vision inspection module 200, and the vacuum suction module 300, forming the control center of the system. It not only receives pulse signals from the encoder 130 and defect location information output by the vision processing unit 220, but also generates an action sequence table based on this information, thereby controlling the corresponding vacuum suction head 310 to precisely pick up defects. Simultaneously, the PLC controller 400 also uniformly regulates the conveying speed of the conveyor belt 120, the shooting interval of the vision inspection module 200, and the on / off duration of the solenoid valve group 330, ensuring that each subsystem maintains strict timing synchronization and parameter matching during dynamic operation. This modular system architecture design allows each functional unit to be relatively independent for easy maintenance and replacement, while also achieving efficient collaboration under the overall coordination of the PLC controller 400, providing reliable hardware support for the fine screening of wet slurry.

[0069] In another embodiment, to further improve the accuracy of point-to-point suction, the vacuum suction module 300 also includes a vertical displacement mechanism 350 connected to each vacuum suction head 310. The vertical displacement mechanism 350 is used to drive the corresponding vacuum suction head 310 down to a position close to the surface of the conveyor belt 120 when a defect reaches the vacuum suction station, and to control the vacuum suction head 310 to reset after vacuum suction is completed.

[0070] Specifically, if the vacuum suction head 310 is always kept in a low position close to the conveyor belt 120, it is prone to continuous friction with the pulp layer, which will accelerate the wear of the suction head and contaminate the qualified pulp. Therefore, this embodiment introduces a vertical displacement mechanism 350 to realize the dynamic switching between the vacuum suction head standby state and the close suction state.

[0071] During standby, the vertical displacement mechanism 350 raises the vacuum suction head 310 to a preset safe height, maintaining a sufficient gap between it and the slurry layer on the conveyor belt 120 to avoid any physical contact. Only when the PLC controller 400 determines that a defect is about to arrive and the corresponding suction head is triggered, does the vertical displacement mechanism 350 quickly drive the suction head downward until its port approaches or lightly touches the slurry surface, forming an effective sealed suction area. After suction is completed and a short suction time has elapsed, the vertical displacement mechanism 350 drives the suction head upward again to reset and detach it from the slurry layer.

[0072] The above-mentioned action cycle is controlled by the PLC controller 400 to interlock the action of the solenoid valve group 330 in a time sequence. It follows the control logic of first lowering to the position and then opening the vacuum, and first closing the vacuum and then rising to reset, so as to prevent premature air intake before the suction head is in position, which would cause slurry splashing, or carry away excess material before the suction head leaves the position.

[0073] The specific implementation of the vertical displacement mechanism 350 is not limited to a single form in this application. For example, it can be driven by a cylinder, utilizing the rapid response characteristics of the cylinder to achieve high-speed production; it can also be driven by a servo electric cylinder or a linear motor to achieve more precise position control and flexible acceleration and deceleration curves. Regardless of the driving method used, as long as the vacuum suction head 310 can achieve controlled reciprocating motion in the vertical direction, it falls within the protection scope of this application. By introducing the vertical displacement mechanism 350, this embodiment effectively solves the technical problems of easy damage to the suction head and difficulty in ensuring sealing in wet fruit pulp environments. While ensuring the success rate of defect screening, it significantly extends the service life of key components and reduces the risk of cross-contamination, demonstrating the deep adaptation of the system design to actual working conditions.

[0074] Example 4: This embodiment provides a computer program product, including a computer program or instructions. When the computer program or instructions are executed by a processor, they implement the steps of the processing method as described in the foregoing embodiment.

[0075] Specifically, the computer program product can be stored in various forms of computer-readable storage media. For example, the storage medium can be volatile memory, such as random access memory; it can also be non-volatile memory, such as read-only memory, electrically erasable programmable read-only memory, or flash memory; it can also be magnetic storage devices (such as hard drives or magnetic tapes) or optical storage devices (such as optical discs). Furthermore, with the development of Industrial Internet technology, the computer program product can also manifest as data signals transmitted over a network or software service packages on a cloud server. When a terminal device downloads or remotely invokes the data signal or service package, it is considered to have implemented the computer program product.

[0076] When the aforementioned computer program or instructions are loaded into the processor (such as the computing core within the aforementioned vision processing unit 220 or PLC controller 400) and run, the processor will automatically execute complete control logic, including slurry conveying control, defect visual recognition, generation of action sequence tables based on position information, and fixed-point triggering of the vacuum suction module. It should be understood that solidifying the technical solution of this application into a computer program product, thereby decoupling the control method from the mechanical structure, allows the technical solution to be sold and licensed as an independent industrial software product, adapting to different customer procurement models. Furthermore, it facilitates the intelligent transformation of existing equipment; users only need to update the control program to improve the accuracy and adaptability of fruit pulp defect screening, without requiring expensive hardware replacement of the conveyor belt or vacuum module, thus enhancing the flexibility of the technical solution.

[0077] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. 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 of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for screening defects in fruit pulp, characterized in that, Includes the following steps: S100, the pulp is spread evenly on the surface of the conveyor belt, and the conveyor belt is controlled to continuously convey the pulp at a preset speed; S200, when the conveyor belt transports the pulp to the visual inspection station, the pulp is photographed, defects in the pulp are identified by a visual recognition algorithm, and the position information of each defect on the conveyor belt plane is recorded; the position information includes a first coordinate along the length direction of the conveyor belt and a second coordinate along the width direction of the conveyor belt. S300, based on the location information, determines the execution sequence of each defect arriving at the vacuum suction station, specifically including: Based on the first coordinate, the timing of each defect reaching the corresponding vacuum suction station is calibrated; Based on the second coordinate, each defect is assigned to and calibrated to a plurality of vacuum suction head channels arranged along the width of the conveyor belt; Based on the calibration results, an action sequence table containing the start-up order and timing of each vacuum head is generated; S400, according to the action sequence table, when the pulp material including the defect arrives at the vacuum suction station, the vacuum suction head of the corresponding vacuum suction head channel is controlled to start, so as to perform targeted suction of the defect in the pulp material on the conveyor belt.

2. The treatment method for screening defects in fruit pulp according to claim 1, characterized in that, S100 also includes a preliminary step on the characteristics of the pulp: before or during the feeding of the pulp, the solid content and / or fluid viscosity of the batch of pulp are detected; in S400, the suction time after the vacuum suction head is started is adjusted according to the detection results of the solid content and / or fluid viscosity of the pulp.

3. The treatment method for screening defects in fruit pulp according to claim 2, characterized in that, In S100, the preset speed of the conveyor belt is adjusted according to the suction time of the vacuum suction head; in S200, the shooting interval of the fruit pulp is adjusted according to the preset speed of the conveyor belt.

4. The treatment method for screening defects in fruit pulp according to claim 1, characterized in that, The drive shaft of the conveyor belt is equipped with an encoder, which is used to output periodic pulse signals generated as the drive shaft rotates. In S300, the step of calibrating the timing of each defect reaching directly below the corresponding vacuum suction head based on the first coordinate specifically includes: Based on the pre-calibrated distance between the visual inspection station and the vacuum suction station and the first coordinate, the countdown value of the pulse signal when each defect reaches directly below the corresponding vacuum suction head is calibrated, and the countdown is performed based on the periodic pulse signal; when the countdown value of the pulse signal returns to zero, it is the start-up time of the corresponding vacuum suction head in the action sequence table.

5. The treatment method for screening defects in fruit pulp according to claim 1, characterized in that, In S400, when the pulp containing the defect arrives at the vacuum suction station according to the action sequence table, the vacuum suction head of the corresponding vacuum suction head channel is controlled to start. The method also includes setting an advance amount according to the inherent response time of the vacuum suction head and its driving mechanism, and deducting the advance amount from the start time in the action sequence table as the actual start time for controlling the corresponding vacuum suction head. The lead time is adjusted based on the verification results of vacuum aspiration.

6. The treatment method for screening defects in fruit pulp according to claim 1, characterized in that, The visual recognition algorithm is a target detection model or image classification model based on a neural network; the neural network includes one of the YOLO series models, CNN, CNN-Transformer, or a network structure based on the above models and improved thereon.

7. The treatment method for screening defects in fruit pulp according to claim 1, characterized in that, The defects include one or more of the following: discolored pulp, peel or peel fragments, leucocephala or leucocephala fragments, seeds or seed fragments, black spots, mold, and foreign matter that is not pulp.

8. A treatment system for screening defects in fruit pulp, used to implement the treatment method as described in any one of claims 1-7, characterized in that, include: Material conveying module, including feeding device, conveyor belt and encoder; The feeding device is used to spread the fruit pulp evenly on the surface of the conveyor belt; the encoder is set on the drive shaft of the conveyor belt and is used to output pulse signals. A visual inspection module, located at the visual inspection station, includes an image acquisition device and a visual processing unit; the image acquisition device is used to photograph the pulp on the conveyor belt; the visual processing unit is used to identify defects in the pulp through a visual recognition algorithm and output the location information of the defects. The vacuum suction module is set at the vacuum suction station and includes multiple vacuum suction heads arranged along the width direction of the conveyor belt, a vacuum pump that provides negative pressure to each vacuum suction head, a solenoid valve group that controls the opening and closing of the air passage of each vacuum suction head, and a waste liquid tank connected to each vacuum suction head. The PLC controller is connected to the material conveying module, the vision inspection module, and the vacuum suction module, respectively. It is used to receive the pulse signal and the location information of the defect, control the corresponding vacuum suction head to perform fixed-point suction of the defect, and control the conveying speed of the conveyor belt, the shooting interval of the vision inspection module, and the on / off duration of the solenoid valve group.

9. The treatment system for screening defects in fruit pulp according to claim 8, characterized in that, The vacuum suction module also includes a vertical displacement mechanism connected to each of the vacuum suction heads. The vertical displacement mechanism is used to drive the corresponding vacuum suction head to descend to a position close to the surface of the conveyor belt when the defect reaches the vacuum suction station, and to control the vacuum suction head to reset after vacuum suction is completed.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the processing method as described in any one of claims 1-7.

Images