Intelligent control system and method for cake production line

Abstract

This invention relates to the field of industrial automatic control technology, and in particular to an intelligent control system and method for a cake production line based on machine vision feedforward and cross-process temporal decoupling. The intelligent control system for a cake production line adopts a layered architecture of industrial PC and PLC, including: a full-dimensional perception layer for collecting operating status and physical characteristic data of the controlled object, including at least a vision sensor for acquiring image data of the cake surface after baking and an infrared thermometer for acquiring surface temperature; an edge collaborative control layer, communicatively connected to the full-dimensional perception layer, including: a feature extraction unit for receiving the image data, extracting chromaticity feature values, and calculating a ripeness index in combination with the surface temperature; a temporal decoupling and compensation decision unit for comparing the ripeness index with a preset threshold and calculating the control parameter compensation amount for downstream processes based on the degree of deviation; and an instruction generation and issuance unit for superimposing the reference parameters and the compensation amount to generate control instructions.

Description

Technical Field

[0001] This invention relates to the field of industrial automatic control technology, and in particular to an intelligent control system and method for a cake production line based on machine vision feedforward and cross-process timing decoupling, which belongs to the category of industrial automatic control system device manufacturing in the intelligent manufacturing equipment industry. Background Technology

[0002] In modern food processing, especially the baking industry, large-scale cake production typically relies on continuous production lines, encompassing multiple sequential processes such as batter filling, baking, demolding, cooling, and packaging. However, current cake production line control systems generally suffer from the following technical challenges: First, existing control systems mostly employ a hybrid architecture of independent closed-loop control for individual machines and rigid global linkage. Each piece of equipment, such as the baking oven, cooling tower, and demolding machine, is equipped with an independent controller to achieve localized constant temperature or speed control; the host computer system is only responsible for global sequential start-up and shutdown. This "information silo" control architecture severs the strong physical coupling between processes. As a typical high-moisture, heat-sensitive controlled object, a cake's minute thermodynamic deviations during baking (such as changes in moisture loss caused by fluctuations in surface heat) will be amplified in subsequent processes along the material flow direction. When a cake exiting the oven experiences excessive caramelization on the surface and low internal moisture due to a momentary high oven temperature, if the cooling section continues to cool at the baseline airflow, it will further dry out the cake, causing it to become brittle and subsequently crack extensively when flipped by the robotic arm during the demolding process.

[0003] Second, existing quality inspection methods are severely outdated and lack feedforward control capabilities. Currently, the industry typically uses manual sampling or end-of-line visual sorting to remove defective products after the cakes are baked. This method is essentially "post-production feedback," meaning that defective products have already been generated and solidified. Inspection and removal can only prevent defective products from entering the market, but cannot promptly correct production process parameters to stop the continuous generation of defective products. Even if a few production lines have introduced visual monitoring, they merely use the visual results for simple pass / fail binary judgment to drive the rejection cylinder, failing to quantify visual features as control variables and input them into the automatic control system to form a closed loop.

[0004] Third, there is a lack of dynamic decoupling mechanism for cross-process parameter adjustments. When abnormal operating conditions occur, the existing system usually resorts to a complete shutdown or manual adjustment of the inverter frequency based on experience. Manual intervention has a significant time delay, and the adjustment accuracy varies from person to person, which can easily cause system oscillations, resulting not only in a large waste of raw materials but also in a significant decrease in overall equipment efficiency (OEE).

[0005] In summary, existing industrial automatic control systems suffer from technical defects when dealing with continuous production processes like cake making, which involve complex phase transitions and rheological characteristics. These defects include blind spots in inter-process coordination, a lack of feedforward compensation mechanisms based on online quality perception, and the difficulty in dynamically demodulating multivariable strongly coupled systems. Summary of the Invention

[0006] The purpose of this invention is to overcome the technical defects of existing cake production lines, such as independent control of each process and lack of cross-process collaborative linkage based on quality perception, which leads to low yield rates, and to provide an intelligent control system and method for cake production lines.

[0007] The technical solution adopted by this invention to solve its technical problem is: an intelligent control system for a cake production line, employing a layered architecture of industrial PC and PLC, including: a full-dimensional perception layer, used to collect operating status and physical characteristic data of the controlled object, including at least a visual sensor for acquiring image data of the cake surface after baking and an infrared thermometer for acquiring surface temperature; and an edge collaborative control layer, communicatively connected to the full-dimensional perception layer, including: a feature extraction unit, used to receive the image data, extract chromaticity feature values, and calculate the ripeness index in combination with the surface temperature. The timing decoupling and compensation decision unit is used to process the maturity index. Compared with a preset threshold, the control parameter compensation amount of the downstream process is calculated according to the degree of deviation; the instruction generation and issuance unit is used to superimpose the reference parameters and the compensation amount to generate control instructions; the precision execution layer is communicatively connected to the edge collaborative control layer and is used to drive the downstream actuator to act according to the control instructions.

[0008] In the aforementioned intelligent control system for the cake production line, the feature extraction unit maps the RGB image to the HSV space to extract the V channel image, performs threshold segmentation to remove the background, extracts the region of interest on the cake surface, and calculates the arithmetic mean of the V channel pixels within the region as the chromaticity feature value. The maturity index ,in, For standard chromaticity, The projected area of ​​the top surface. For standard area, For surface temperature, Standard surface temperature These are the weighting coefficients.

[0009] The aforementioned intelligent control system for the cake production line, when Greater than the over-baking threshold At that time, the decision unit calculates the cooling fan frequency reduction compensation to make the target frequency [value missing]. Calculate the conveyor belt speed-up compensation to shorten cooling retention, and calculate the speed attenuation coefficient of the demolding robot. Make the servo flip speed as and slightly reduce clamping pressure, This is the compensation coefficient.

[0010] The aforementioned intelligent control system for the cake production line, when Less than the under-baking threshold At that time, the decision unit calculates the frequency upscaling compensation of the cooling fan to make the target frequency [value missing]. Calculate the conveyor belt speed reduction compensation to extend the cooling and shaping time, and calculate the demolding torque limiting proportional coefficient. The maximum output torque of the servo is clamped to This is the compensation coefficient.

[0011] The aforementioned intelligent control system for the cake production line also includes a dynamic weighing module in its all-dimensional sensing layer; when If the overheating is indicated but the mass data obtained by the weighing module does not show any reduction, the decision unit triggers a correction mechanism to weaken the compensation amount; when this is detected a predetermined number of times consecutively... When the same threshold is exceeded, the edge collaborative control layer writes a feedforward bias to the front-end baking oven temperature control PLC to fine-tune the oven temperature.

[0012] The aforementioned intelligent control system for the cake production line incorporates ramp generator logic in its instruction generation and distribution unit. Based on the real-time displacement of the encoder, it calculates the timestamp of abnormal batches entering the downstream action zone. When the time node is reached, it updates the target value and writes it into the PLC. The PLC ramp instruction module then smoothly transitions the actual output parameters to the target value.

[0013] A method for intelligent control of a cake production line includes: acquiring image data and surface temperature data of cakes exiting the oven at the end of the baking zone; processing the image data to extract chromaticity feature values ​​and calculating a ripeness index M in combination with the surface temperature; comparing the ripeness index M with a preset threshold and calculating the control parameter compensation amount for downstream processes based on the degree of deviation; superimposing the reference parameters and the compensation amount to generate control commands and sending them to drive the downstream actuators to move.

[0014] The aforementioned intelligent control method for cake production lines extracts chromaticity feature values ​​as follows: mapping the image to the HSV space to extract the V channel, thresholding to extract the region of interest (ROI) of the cake, and calculating the arithmetic mean of the V channel values ​​of the pixels within the region. The maturity index .

[0015] The above-mentioned intelligent control method for cake production lines, when Greater than the over-baking threshold At that time, calculate the compensation amount for cooling fan frequency reduction and conveyor belt speed increase, and reduce the speed and clamping pressure of the demolding robot; when Less than the under-baking threshold At the same time, calculate the compensation amount for the frequency increase of the cooling fan and the speed reduction of the conveyor belt, and clamp the maximum output torque of the demolding servo motor.

[0016] The intelligent control method for the cake production line described above introduces ramp generator logic when issuing control commands. It calculates the timestamp of abnormal batches entering the downstream action zone based on the real-time displacement of the encoder. When the time node is reached, the updated target value is written to the PLC, and the PLC smoothly transitions the actual output parameters to the target value.

[0017] The beneficial effects of this invention are: 1. Breaking through the limitations of traditional single-machine closed-loop production lines, a feedforward decoupled control architecture of "upstream quality perception - downstream parameter reconstruction" has been constructed. By deeply integrating visual perception technology into the automatic control loop, visual signals that were originally only used for quality inspection are transformed into feedforward inputs for the control system. Before the controlled object undergoes irreversible physical deformation, the downstream actuators are intervened in advance, fundamentally eliminating the defect amplification effect caused by control delay.

[0018] 2. A quantitative model for the "maturity index" based on the fusion of HSV color space luminance characteristics and infrared temperature is proposed. This model has clear physical meaning, low computational complexity, and can complete the entire chain of calculation from image acquisition to compensation output within a few hundred milliseconds, meeting the stringent real-time requirements of continuous production lines.

[0019] 3. Achieved refined and flexible control for multiple abnormal operating conditions. Differentiated control strategies were formulated for the two distinct physical states of over-baking (brittle structure) and under-baking (soft structure), including fan speed reduction / acceleration, conveyor belt dwell time shortening / extending, and robot speed reduction / torque reduction. These strategies effectively avoided defects such as cake breakage and shrinkage, significantly improving the flexibility and yield of the production line. Attached Figure Description

[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0021] Figure 1 This is a structural block diagram of an intelligent control system for a cake production line according to the present invention.

[0022] Explanation of reference numerals in the attached figures: 1-Full-dimensional sensing layer; 11-Out-of-furnace quality inspection unit; 111-Vision sensor; 112-Infrared thermometer; 12-Dynamic weighing module; 13-Environment and condition monitoring unit; 131-Temperature and humidity transmitter; 132-Rotary encoder; 133-Photoelectric sensor; 2-Edge Collaborative Control Layer; 21-Industrial PC; 211-Feature Extraction Unit; 212-Temporal Decoupling and Compensation Decision Unit; 213-Instruction Generation and Issuance Unit; 22-PLC; 3-Precision execution layer; 31-Cooling fan array; 32-Main conveyor belt drive; 33-Demolding robot; 4-Upstream baking oven temperature control system. Detailed Implementation

[0023] This invention provides an intelligent control system and method for a cake production line. In order to enable those skilled in the art to better understand and accurately reproduce the technical solution of this invention, the invention will be further described in detail below in conjunction with specific application scenarios.

[0024] This embodiment provides an intelligent control system for a cake production line, deployed in a typical large-scale continuous baking production line. This system abandons the isolated architecture of traditional production lines where each process controller operates independently, adopting a hierarchical control system that coordinates an industrial personal computer and a programmable logic controller (PLC). The hardware topology of the entire system is divided into a full-dimensional perception layer, an edge collaborative control layer, and a precise execution layer. The full-dimensional perception layer is responsible for collecting multi-process operating status and physical characteristic data of the controlled object. The edge collaborative control layer communicates with the full-dimensional perception layer and is responsible for performing complex cross-process timing decoupling and compensation decision calculations based on the perceived data. The precise execution layer communicates with the edge collaborative control layer and includes multiple frequency converters and servo drives, responsible for driving the corresponding actuators according to the control commands issued by the edge collaborative control layer.

[0025] In terms of hardware deployment, the core computing unit of the edge collaborative control layer uses a high-performance industrial PC, equipped with a multi-core processor and large-capacity memory, running a real-time operating system, responsible for deploying vision processing algorithms and cross-process collaborative control models. The logic control unit uses a high-performance PLC with nanosecond-level instruction execution speed, responsible for low-level logic operations and hard real-time input / output control. The two exchange data at high speed via industrial Ethernet, using shielded twisted-pair cable as the physical medium, and employing an industrial protocol that supports real-time communication to ensure zero-delay transmission of large amounts of data and control commands.

[0026] The core of the all-dimensional sensing layer lies in the oven-out quality inspection unit, located directly above the conveyor belt between the baking zone exit and the cooling zone entrance. This unit serves as the outpost breaking down information barriers between processes. It includes a gigabit Ethernet industrial camera equipped with a fixed-focus lens and a ring-shaped flicker-free LED light source. Mounted approximately half a meter above the conveyor belt surface, its field of view covers multiple cake mold cavities in a single row, acquiring high-resolution image data of the cake surfaces after baking. Adjacent to the camera is an infrared thermometer, whose temperature measurement range covers the typical temperature range of baked goods. Its emissivity is calibrated based on the radiation characteristics of the cake surface, and its output analog signal is connected to the PLC's analog input module. Furthermore, the sensing layer also deploys temperature and humidity transmitters within the cooling tunnel to monitor the cooling environment, and a high-resolution incremental rotary encoder is installed at the tail end of the main conveyor belt drive motor to track the conveyor belt's displacement and speed in real time.

[0027] The precision execution layer encompasses the drive equipment for the two key processes of cooling and demolding. The cooling section is equipped with a high-power axial flow fan array, driven by a corresponding frequency converter, which can accept frequency commands from 0 to 50 Hz output from the PLC to achieve stepless adjustment of airflow. The demolding section is equipped with a tilting robotic arm driven by a servo motor. This servo system supports seamless online switching between position control mode and torque limiting mode. Meanwhile, the clamping pressure of the clamping cylinder is finely adjusted via an electro-proportional valve, accepting analog voltage signals from the PLC.

[0028] After the system is powered on, the feature extraction unit and the timing decoupling and compensation decision unit in the edge collaborative control layer begin to work together. When the baked cake moves with the mold to the center of the vision sensor's field of view, the photoelectric sensor installed next to the track is triggered, generating a rising edge electrical signal. After the PLC captures this signal, it immediately sends a trigger command to the industrial PC via Ethernet. Upon receiving the trigger command, the industrial PC controls the industrial camera to expose and capture a frame of image, and transmits the image data stream to the industrial PC's memory via a gigabit Ethernet port.

[0029] The feature extraction unit then preprocesses and performs multi-dimensional feature calculations on the original image. Since slight fluctuations in ambient lighting are unavoidable in workshop environments, direct color recognition in the RGB color space is highly prone to misjudgment. Therefore, this system first normalizes the R, G, and B components of each pixel in the acquired RGB image to the range of zero to one, and then maps them to the HSV color space according to the standard color space conversion formula. In this space, H represents hue, S represents saturation, and V represents lightness. The most direct manifestation of caramelization and Maillard reactions during baking is the change in reflectivity and brightness of the cake surface; therefore, the system discards the easily interfered H and S channels, extracting only the V channel image matrix representing lightness for subsequent analysis.

[0030] After extracting the V-channel image, the system uses the Otsu algorithm to automatically find a binarization segmentation threshold, dividing the image into foreground and background. Since the cake surface exhibits high brightness in the V-channel image, while the stainless steel mold and background conveyor belt exhibit low brightness, this threshold segmentation effectively removes the background. To further eliminate potential reflective noise on the cake surface, the system performs morphological opening operations on the binarized image: first, erosion is performed to remove isolated noise, and then dilation is performed to restore the original outline of the cake. Subsequently, connected component filling is performed on the processed binarized image to extract the complete cake surface outline as a region of interest mask. A bitwise AND operation is performed between the original V-channel matrix and this mask to extract a set of pure pixels containing only the cake surface brightness information. The system calculates the arithmetic mean of the V values ​​of all pixels in this set and quantizes it as a surface chromaticity feature value. Simultaneously, the system counts the total number of pixels in the region of interest and, combined with pre-calibrated pixel physical size coefficients, calculates the projected area of ​​the cake's top surface. This is used to characterize the volume expansion of the cake during the baking process.

[0031] At the same time, the PLC linearizes the analog signal collected by the infrared thermometer to obtain the center temperature data of the cake surface. To eliminate reference drift caused by different batches of raw materials, the system performs self-learning calibration during the first ten standard cycles of each production start-up. It calculates the average values ​​of the surface colorimetric characteristic values, top surface projected area, and surface temperature of the first ten qualified products, and stores these values ​​as standard baking colorimetric reference values. Standard area and standard furnace surface temperature .

[0032] Based on this, the system deeply integrates color deviation, volume expansion deviation, and temperature deviation to calculate a ripeness index that reflects the current physical state of the cake. The quantitative calculation formula for this index is as follows: ,.in, This is the chromaticity weighting coefficient. This is the expansion degree weighting coefficient. This is the temperature weighting coefficient, and The value of is determined through orthogonal experiments based on the specific baking process. In a typical embodiment, it can be set to . When the calculated maturity index When the value is zero, it indicates that the cake is currently in an ideal, standard ripening state; when... A value greater than zero indicates increased cake surface brightness, excessive volume expansion, and a higher temperature, suggesting overbaking, significant internal moisture loss, and a brittle structure; when... When the value is less than zero, it indicates low brightness, insufficient volume expansion, and low temperature, with a tendency to be underbaked, incomplete gelatinization in the center, and a soft structure with insufficient support.

[0033] The timing decoupling and compensation decision unit built into the edge collaborative control layer continuously monitors this maturity index. And compare it with the preset over-baking threshold. Underbaking threshold A comparison was made. Due to the inherent time lag in material transport on the production line—specifically, the several seconds required for transport from the visual inspection point to the cooling fan's operating area, and the additional time from the cooling zone exit to the demolding robot's operating area—traditional feedback control cannot be effective in a timely manner, easily leading to compensatory actions being applied to the wrong batch of cakes. Therefore, this system employs a mechanism combining feedforward and material tracking.

[0034] maturity index Exceeding the over-roasting threshold At this point, the system determines that the current batch is overbaked. Given the brittle and easily broken nature of overbaked cakes, the system activates its first compensation strategy. In the cooling process, the system calculates the frequency reduction compensation for the cooling fans, setting the target frequency of the fans to F_target = F_base - K1·(M - M_high), where F_base is the base frequency and K1 is the cooling compensation coefficient. This aims to reduce the penetrating power of the cold air, creating a moderately moist surface environment and maintaining the cake's resilience. Simultaneously, the system calculates the speed increase compensation for the main conveyor belt ΔV_c = K5·(M - M_high), increasing the frequency of the conveyor belt inverter to shorten the residence time of this batch of cakes in the cooling tower and inhibit continuous moisture evaporation. In the demolding process, the system estimates the time for the batch to arrive at the demolding area based on the real-time displacement and conveyor belt speed fed back by the encoder, calculates the demolding robot speed attenuation coefficient λ=1-K2·(M-M_high), where K2 is the demolding speed compensation coefficient, and then sets the servo flipping speed to V_arm=V_arm_base·λ to reduce the impact acceleration caused by sudden starts and stops, and simultaneously slightly reduces the pressure of the electro-proportional valve of the clamping cylinder to achieve flexible demolding.

[0035] maturity index Falling below the underbaking threshold At this point, the system determines that the current batch is underbaked. Considering the physical characteristics of underbaked cakes—soft structure, prone to collapse and shrinkage—the system activates a second compensation strategy. Regarding the cooling process, the system calculates the frequency increase compensation for the cooling fan and sets the target frequency of the fan to [value missing]. ,in To compensate for the cooling effect, forced rapid cooling is used to help the underbaked cake's surface solidify quickly, forming a supporting shell; simultaneously, the conveyor belt speed reduction compensation is calculated. This extends the cooling and setting time to compensate for insufficient baking time in the early stages. Regarding the demolding process, underbaked cakes are most susceptible to damage from pressure; the system calculates the torque limiting ratio coefficient of the demolding servo motor. ,in This is the demolding torque compensation coefficient, which further clamps the maximum output torque of the servo to... Even if the cake is too soft and its position deviates slightly, the system can use its flexible force control boundary to prevent the cake from crumbling.

[0036] maturity index lie in and During this period, the system maintains the formula baseline parameters without any additional intervention.

[0037] To prevent compensation commands from causing a step-like mechanical impact on downstream actuators, the command generation and distribution unit of the edge collaborative control layer incorporates ramp generator logic when issuing commands to the PLC. This logic is used when the industrial PC calculates the new target frequency for the wind turbine. After setting the servo operating parameters, instead of directly overwriting the PLC's setting register, the compensation requirement is recorded. Based on the real-time feedback of the conveyor belt displacement from the encoder, the exact timestamps of abnormal batches entering the cooling and demolding zones are precisely calculated. Only when this time point is reached does the industrial PC write the updated target value to the PLC. The PLC's internal ramp instruction module then linearly transitions the actual output frequency or speed from the current value to the target value within a two-second smoothing time, thus avoiding mechanical resonance of the fan or vibration of the conveyor belt.

[0038] Meanwhile, to address the complex operating conditions of multiple batches of continuous anomalies, the system maintains a first-in-first-out (FIFO) queue cache in the industrial PC's memory. Whenever the vision system detects an abnormal cake, its corresponding compensation parameter sequence and estimated arrival time are pushed into this queue. As the conveyor belt runs, when the system timestamp matches the timestamp at the head of the queue, the queue triggers a release control command. This ensures that under conditions of dense incoming materials, each compensation command can accurately match its corresponding target object, completely solving the problem of spatial temporal misalignment between detection and execution in production lines with large lags.

[0039] In more extreme operating conditions, the judgment of a single sensor may have limitations. Therefore, a high-precision dynamic weighing module is added below the conveyor belt to acquire the weight data of the cakes after baking. As a redundancy check. When the maturity index... The indicator shows overbaking, but the quality data is incorrect. If no significant reduction is observed, the system infers that the surface may be contaminated with syrup, causing the color characteristic value to be artificially high. At this point, a correction mechanism is triggered to reduce the compensation amplitude of the cooling fan and conveyor belt by a preset ratio to prevent miscontrol from causing new operational imbalances.

[0040] Furthermore, when the system detects the maturity index five times consecutively... When the same judgment threshold is exceeded, it indicates that a systematic thermal field shift has occurred in the upstream baking oven. At this time, the cross-process time-series decoupling and compensation decision model not only adjusts the downstream cooling and demolding parameters, but also writes the feedforward bias to the temperature control PLC of the front-end baking oven through the industrial Ethernet communication protocol, fine-tuning the set temperature of the top or bottom fire, and realizing a closed-loop whole-link system from passive defense to active correction.

[0041] To verify the technical effectiveness of this system, the applicant conducted a comparative experiment lasting hundreds of hours on the same large-scale sponge cake production line. In the control group using traditional control methods, each process's PLC operated independently, and the cooling fan and demolding robot operated at constant speed and torque. When encountering oven temperature fluctuations, the demolding breakage rate averaged over 6%, reaching as high as 15% during extreme fluctuation periods, and the equipment stopped multiple times due to material jamming. In the experimental group using the system of this invention, facing the same intensity of oven temperature disturbance, the system was able to complete the process from image acquisition to multi-dimensional feature fusion calculation within hundreds of milliseconds, and instantly generate cross-process compensation instructions. The demolding breakage rate was stably controlled within 1%, and no material jamming stoppages due to deformation occurred, resulting in a significant improvement in the overall efficiency of the equipment. This experiment fully demonstrates that the cross-process collaborative control strategy provided by this invention can effectively absorb upstream thermal disturbances, giving traditional rigid production lines extremely high flexibility and adaptability.

[0042] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent control system for a cake production line, characterized in that: It adopts a layered architecture of industrial PC and PLC, including: The full-dimensional perception layer is used to collect data on the operating status and physical characteristics of the controlled object, including at least a visual sensor for acquiring image data of the surface of the cake after baking and an infrared thermometer for acquiring the surface temperature. An edge collaborative control layer, communicatively connected to the full-dimensional perception layer, includes: a feature extraction unit, used to receive the image data, extract chromaticity feature values, and calculate the ripening index in conjunction with surface temperature. The timing decoupling and compensation decision unit is used to process the maturity index. Compared with a preset threshold, the control parameter compensation amount of the downstream process is calculated according to the degree of deviation; the instruction generation and issuance unit is used to superimpose the reference parameter and the compensation amount to generate a control instruction; The precision execution layer is communicatively connected to the edge collaborative control layer and is used to drive the downstream actuators to move according to the control commands.

2. The intelligent control system for a cake production line according to claim 1, characterized in that: The feature extraction unit maps the RGB image to the HSV space to extract the V channel image, performs threshold segmentation to remove the background, extracts the region of interest on the cake surface, and calculates the arithmetic mean of the V channel pixels within the region as the chromaticity feature value. The maturity index ,in, For standard chromaticity, The projected area of ​​the top surface. For standard area, For surface temperature, Standard surface temperature These are the weighting coefficients.

3. The intelligent control system for a cake production line according to claim 2, characterized in that: when Greater than the over-baking threshold At that time, the decision unit calculates the cooling fan frequency reduction compensation to make the target frequency [value missing]. Calculate the conveyor belt speed-up compensation to shorten cooling retention, and calculate the speed attenuation coefficient of the demolding robot. Make the servo flip speed as and slightly reduce clamping pressure, This is the compensation coefficient.

4. The intelligent control system for a cake production line according to claim 2, characterized in that: when Less than the under-baking threshold At that time, the decision unit calculates the frequency upscaling compensation of the cooling fan to make the target frequency [value missing]. Calculate the conveyor belt speed reduction compensation to extend the cooling and shaping time, and calculate the demolding torque limiting proportional coefficient. The maximum output torque of the servo is clamped to This is the compensation coefficient.

5. The intelligent control system for a cake production line according to claim 1, characterized in that: The full-dimensional sensing layer also includes a dynamic weighing module; when If the overheating is indicated but the mass data obtained by the weighing module does not show any reduction, the decision unit triggers a correction mechanism to weaken the compensation amount; when this is detected a predetermined number of times consecutively... When the same threshold is exceeded, the edge collaborative control layer writes a feedforward bias to the front-end baking oven temperature control PLC to fine-tune the oven temperature.

6. The intelligent control system for a cake production line according to claim 1, characterized in that: The instruction generation and issuance unit incorporates ramp generator logic. It calculates the timestamp of abnormal batches entering the downstream action zone based on the real-time displacement of the encoder. When the time node is reached, the updated target value is written to the PLC, and the PLC ramp instruction module smoothly transitions the actual output parameters to the target value.

7. A method for intelligent control of a cake production line, characterized in that: include: Acquire image data and surface temperature data of the cake exiting the oven at the end of the baking zone; The image data is processed to extract chromaticity feature values, and the maturity index is calculated in conjunction with the surface temperature. ; The maturity index Compared with a preset threshold, the compensation amount of the control parameters of the downstream process is calculated based on the degree of deviation; The reference parameters are superimposed with the compensation amount to generate control commands, which are then sent to drive the downstream actuators to perform actions.

8. The intelligent control method for a cake production line according to claim 7, characterized in that: The extracted chromaticity feature values ​​are: mapping the image to the HSV space to extract the V channel, thresholding to extract the region of interest (ROI) of the cake, and calculating the arithmetic mean of the V channel values ​​of the pixels within the region. The maturity index .

9. The intelligent control method for a cake production line according to claim 8, characterized in that: when Greater than the over-baking threshold At that time, calculate the compensation amount for cooling fan frequency reduction and conveyor belt speed increase, and reduce the speed and clamping pressure of the demolding robot; when Less than the under-baking threshold At the same time, calculate the compensation amount for the frequency increase of the cooling fan and the speed reduction of the conveyor belt, and clamp the maximum output torque of the demolding servo motor.

10. The intelligent control method for a cake production line according to claim 7, characterized in that: When issuing control commands, a ramp generator logic is introduced to calculate the timestamp of abnormal batches entering the downstream action zone based on the real-time displacement of the encoder. When the time node is reached, the updated target value is written to the PLC, and the PLC smoothly transitions the actual output parameters to the target value.

Images