Energy-saving dehydration method and system for carrot slices

Abstract

The application discloses a carrot slice energy-saving dehydration method and system, and belongs to the technical field of hot air dehydration of fruits and vegetables and energy-saving drying control of agricultural products. The method obtains batch state data of carrot slices, generates water content state parameters and stage discrimination results, and forms a control sequence of air temperature, air speed, relative humidity, exhaust valve opening degree and circulating air proportion according to the water content state parameters and the stage discrimination results, so that the technical effects of adapting to batch differences, reducing the risk of over-drying and reducing invalid energy consumption are realized.

Description

Technical Field

[0001] This invention relates to the field of hot air dehydration of fruits and vegetables and energy-saving drying control technology for agricultural products, and particularly to an energy-saving dehydration method and system for carrot slices. Background Technology

[0002] Dehydration of carrot slices is a common process in fruit and vegetable processing. Hot air drying equipment typically removes moisture by setting the air temperature, air speed, and processing time. However, due to batch-to-batch variations in slice thickness, initial moisture content, packing density, and surface water after washing, it is difficult to accurately match the moisture migration status of different batches using fixed process parameters.

[0003] Existing methods often rely on uniform temperature profiles or empirical dehumidification strategies, which can easily lead to insufficient surface water removal in the early stages of drying, or excessively dry edges and high moisture content in the center during the final drying stage. These differences further affect rehydration dispersibility, finished product brittleness, and moisture consistency. Furthermore, maintaining a high hot air output even when the rate of moisture loss has changed results in high energy consumption per unit of dehydration.

[0004] Therefore, there is a need for an energy-saving dehydration method and system for carrot slices that can overcome the shortcomings of the existing technology. Summary of the Invention

[0005] One objective of this invention is to propose an energy-saving dehydration method and system for carrot slices. Addressing the shortcomings of existing technologies where uniform air temperature, air velocity, or processing time is insufficient to accommodate batch variations in carrot slices, and where phased energy-saving control based on moisture content changes is lacking, this invention proposes a closed-loop control solution that integrates batch status acquisition, moisture content estimation, phase discrimination, energy-saving predictive control, and execution adjustment. This invention achieves the technical effects of improving the consistency of finished product moisture content and reducing ineffective hot air energy consumption.

[0006] This invention provides an energy-saving dehydration method for carrot slices, comprising: S1, acquiring batch status data of the batch to be dehydrated, wherein the batch status data includes carrot slice thickness distribution, loading density, initial weight, current acquisition cycle weight, surface image reflective grayscale value, near-infrared moisture characteristics, material surface temperature, inlet air temperature, return air temperature, exhaust air humidity, relative humidity change rate, and wind speed; S2, generating thickness dispersion coefficient, surface water index, total water estimate, free water estimate, bound water estimate, moisture descent rate, surface evaporation intensity, material temperature hysteresis, internal migration restriction coefficient, and edge-center moisture difference estimate based on the batch status data; S3, generating thickness dispersion coefficient, surface water index, total water estimate, free water estimate, bound water estimate, moisture descent rate, surface evaporation intensity, material temperature hysteresis, internal migration restriction coefficient, and edge-center moisture difference estimate based on the surface water index and moisture descent rate; The current dehydration stage is determined by the descent rate, the relative humidity change rate, the material temperature hysteresis, and the internal migration restriction coefficient. The current dehydration stage is either the surface water peeling stage, the internal moisture migration maintenance stage, or the balanced final drying stage. S4. A control sequence is generated based on the current dehydration stage, the estimated edge-center moisture content difference, the internal migration restriction coefficient, the target endpoint moisture content, the unit dehydration energy consumption, and the preset brittleness retention constraint. The control sequence includes air temperature, air velocity, relative humidity, exhaust valve opening, and circulating air ratio. S5. The heater, fan, dehumidifier, exhaust valve, and circulating air valve are adjusted according to the control sequence, and the control sequence is continuously corrected based on the batch status data obtained in the next acquisition cycle.

[0007] Optionally, S1 includes:

[0008] After the carrot slices are loaded onto the tray and before the heater enters the dehydration power output, the thickness measurement values ​​at different locations on the tray, the loading weight corresponding to the unit tray area, the initial weight of the whole batch, the weight of each collection cycle, the material surface temperature, visible light images and near-infrared spectral data are collected.

[0009] The visible light image is divided into an edge region image and a central region image based on the tray position, and the ratio of the exhaust humidity difference between two consecutive acquisition cycles to the acquisition cycle duration is taken as the relative humidity change rate.

[0010] Optionally, S2 includes:

[0011] The thickness dispersion coefficient is determined based on the maximum thickness, minimum thickness, and average thickness of the thickness distribution.

[0012] The surface water adhesion index is determined based on the grayscale difference between the grayscale value of the surface image and the grayscale value of the reference grayscale value after tray loading.

[0013] The estimated total water, the estimated free water, and the estimated bound water are determined based on the near-infrared moisture characteristics, respectively.

[0014] The moisture descent rate and the surface evaporation intensity are determined based on the initial weight, the weight of the current collection cycle, and the relative humidity change rate.

[0015] The material temperature hysteresis is determined based on the difference between the material surface temperature and the return air temperature;

[0016] The internal migration restriction coefficient is a parameter obtained by weighting the ratio of the bound water estimate to the total water estimate, the material temperature hysteresis, and the thickness dispersion coefficient with a preset weight normalization.

[0017] Furthermore, it also includes: determining the edge water content estimate based on the color change of the edge region image, the shrinkage rate of the edge region image, the estimated near-infrared total water and free water of the corresponding edge region;

[0018] The estimated water content at the center is determined based on the estimated total water content, estimated free water content, and thickness measurement values ​​corresponding to the central region.

[0019] After converting the estimated edge water content and the estimated center water content to the same water content unit, the difference between the estimated center water content and the estimated edge water content is calculated to obtain the estimated edge-center water content difference. The estimated edge-center water content difference is then mapped to a uniformity constraint coefficient according to a preset mapping table.

[0020] Optionally, S3 includes:

[0021] When the surface water index is greater than a first preset threshold, the relative humidity change rate is greater than a second preset threshold, and the moisture loss rate is greater than the initial water loss rate threshold, the current dehydration stage is determined to be the surface water peeling stage.

[0022] When the estimated free water value is not greater than the third preset threshold, the internal migration restriction coefficient is greater than the fourth preset threshold, and the water decrease rate is not greater than the migration maintenance rate threshold, the current dehydration stage is determined to be the internal water migration maintenance stage.

[0023] When the estimated free water value is less than the fifth preset threshold, the estimated total water value is not greater than the upper limit of the target endpoint moisture content, and the estimated edge-center moisture content difference value is not greater than the preset uniformity threshold, the current dehydration stage is determined to be the balanced final drying stage.

[0024] Optionally, S4 includes:

[0025] The current dehydration stage, the uniformity constraint coefficient, the internal migration restriction coefficient, the unit dehydration energy consumption, the target endpoint moisture content, and the preset brittleness retention constraint are input into the prediction controller.

[0026] The preset brittleness retention constraints include the upper limit of wind temperature, the boundary of single-cycle wind temperature change, and the lower limit of edge moisture content corresponding to the brittleness evaluation index of the target product;

[0027] The predictive controller is constrained by the fact that the changes in wind temperature, wind speed, exhaust valve opening and recirculation ratio between adjacent control cycles do not exceed the corresponding preset change boundaries, and outputs the control sequence that covers at least two subsequent control cycles.

[0028] Furthermore, it also includes: inputting the near-infrared predicted moisture content obtained from the near-infrared moisture characteristics and the total water estimate, the weighing-based moisture content obtained from the initial weight and the current acquisition cycle weight, the cumulative exhaust moisture content obtained from the exhaust humidity, the wind speed and the acquisition cycle, the edge moisture content estimate, the center moisture content estimate, the edge-center moisture content difference estimate, and the prediction deviation of the most recent acquisition cycle into the endpoint confidence calibration unit, wherein the prediction deviation of the most recent acquisition cycle is the difference between the near-infrared predicted moisture content and the weighing-based moisture content;

[0029] The endpoint confidence calibration unit generates an upper confidence boundary for the central moisture content based on the estimated central moisture content and the deviation quantile corresponding to the preset confidence level, and generates a lower confidence boundary for the edge moisture content based on the estimated edge moisture content, the estimated edge-central moisture content difference, and the deviation quantile.

[0030] When the upper confidence boundary of the central moisture content is not greater than the upper limit of the target endpoint moisture content and the lower confidence boundary of the edge moisture content is not less than the lower limit of the edge moisture content in the preset brittleness retention constraint, the high-power dehydration stop instruction will be written into the control sequence.

[0031] Optionally, S5 includes:

[0032] After each collection cycle, the batch status data is reacquired and the surface water index, the free water estimate, the bound water estimate, the edge-center water content difference estimate, the upper confidence boundary of the center water content, and the lower confidence boundary of the edge water content are updated.

[0033] When the updated edge-center moisture content difference estimate is greater than the preset uniformity threshold, the opening of the exhaust valve is reduced in the subsequent control cycle, and the proportion of circulating air is increased after the opening of the exhaust valve reaches the preset lower limit.

[0034] When the updated confidence boundary of the central moisture content is greater than the upper limit of the target endpoint moisture content, the control sequence corresponding to the internal moisture migration maintenance phase is extended.

[0035] When the updated lower confidence boundary of edge moisture content is less than the lower limit of edge moisture content in the preset brittleness retention constraint, reduce the wind temperature or wind speed in subsequent control cycles.

[0036] Furthermore, it also includes: when the current dehydration stage is the surface water stripping stage and the surface water index is greater than the sixth preset threshold, low-temperature pulse dehumidification control is executed;

[0037] The low-temperature pulse dehumidification control includes limiting the air temperature to within the upper limit of the initial air temperature, limiting the air speed to within the initial air speed setting range, and setting the dehumidification valve opening to the initial dehumidification opening within a preset pulse cycle, until the surface water index is not greater than the sixth preset threshold, and then switching to the control sequence.

[0038] On the other hand, the present invention also provides an energy-saving dehydration system for carrot slices, comprising:

[0039] A batch status acquisition module is used to acquire batch status data of the batch to be dehydrated; a moisture content estimation module is used to generate, based on the batch status data, thickness dispersion coefficient, surface attached water index, total water estimate, free water estimate, bound water estimate, moisture decline rate, surface evaporation intensity, material temperature hysteresis, internal migration restriction coefficient, and edge-center moisture content difference estimate, wherein the internal migration restriction coefficient is a parameter obtained by weighting the bound water estimate to the total water estimate, the material temperature hysteresis, and the thickness dispersion coefficient using a preset weight normalization; a stage discrimination module is used... The current dehydration stage is determined based on the surface water index, the rate of moisture decrease, the rate of change of relative humidity, the material temperature hysteresis, and the internal migration restriction coefficient. An energy-saving prediction and control module generates a control sequence based on the current dehydration stage, the estimated edge-to-center moisture difference, the internal migration restriction coefficient, the target endpoint moisture content, the unit dehydration energy consumption, and a preset brittleness retention constraint. An execution adjustment module adjusts the heater, fan, dehumidifier, exhaust valve, and circulating air valve according to the control sequence, and continuously corrects the control sequence based on batch status data obtained in the next acquisition cycle.

[0040] The beneficial effects of this invention are:

[0041] 1. By using thickness distribution, loading density, collection cycle weight, surface image reflective grayscale value, near-infrared moisture characteristics, material surface temperature and exhaust humidity change rate as batch status data, dehydration control can no longer rely solely on fixed time or fixed air temperature, thereby improving adaptability to the moisture status of different batches of carrot slices.

[0042] 2. By generating surface water index, estimated free water, estimated bound water, material temperature hysteresis, internal migration restriction coefficient, and edge-center moisture content difference estimate, and based on these, the surface water peeling stage, internal moisture migration maintenance stage, and balanced final drying stage can be determined. This can reduce excessively dry edges and excessively high moisture content in the center and improve the consistency of finished product moisture content.

[0043] 3. By generating control sequences for wind temperature, wind speed, relative humidity, dehumidification valve opening, and circulating air ratio through a predictive controller, and by using endpoint confidence calibration and rolling correction to control the final drying process, unnecessary high-power dehydration and excessive dehumidification can be reduced, thereby reducing unit dehydration energy consumption and maintaining the crispness of carrot slices. Attached Figure Description

[0044] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0045] Figure 1 This is a flowchart of the energy-saving dehydration method for carrot slices according to the present invention.

[0046] Figure 2 This is a flowchart of the S4 energy-saving predictive control and endpoint confidence calibration of the present invention. Detailed Implementation

[0047] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0048] refer to Figures 1-2A method for energy-saving dehydration of carrot slices includes: S1, acquiring batch status data of the batch to be dehydrated, wherein the batch status data includes carrot slice thickness distribution, loading density, initial weight, current acquisition cycle weight, surface image reflective grayscale value, near-infrared moisture characteristics, material surface temperature, inlet air temperature, return air temperature, exhaust air humidity, relative humidity change rate, and wind speed; S2, generating thickness dispersion coefficient, surface attached water index, total water estimate, free water estimate, bound water estimate, moisture descent rate, surface evaporation intensity, material temperature hysteresis, internal migration restriction coefficient, and edge-center moisture difference estimate based on the batch status data; S3, based on the surface attached water index, the moisture descent rate, and the relative humidity change rate, the method for energy-saving dehydration of carrot slices includes: S1, acquiring batch status data of the batch to be dehydrated, wherein the batch status data includes carrot slice thickness distribution, loading density, initial weight, current acquisition cycle weight, surface image reflective grayscale value, near-infrared moisture characteristics, material surface temperature, inlet air temperature, exhaust air humidity, relative humidity change rate, and wind speed; S3, generating thickness dispersion coefficient, surface attached water index ... The current dehydration stage is determined by the rate of change of relative humidity, the material temperature hysteresis, and the internal migration restriction coefficient. The current dehydration stage is either the surface water peeling stage, the internal moisture migration maintenance stage, or the balanced final drying stage. S4. A control sequence is generated based on the current dehydration stage, the estimated edge-center moisture content difference, the internal migration restriction coefficient, the target endpoint moisture content, the unit dehydration energy consumption, and the preset brittleness retention constraint. The control sequence includes air temperature, air velocity, relative humidity, exhaust valve opening, and circulating air ratio. S5. The heater, fan, dehumidifier, exhaust valve, and circulating air valve are adjusted according to the control sequence, and the control sequence is continuously corrected based on the batch status data obtained in the next acquisition cycle.

[0049] In this specific embodiment, S1 includes:

[0050] The process begins when the carrot slices are loaded onto the tray and the heater has not yet entered the dehydration power output state. A two-dimensional coordinate system is first established with the effective bearing area of ​​the tray as the reference, and the area is divided into three equal segments along the length and width directions. The thickness of the carrot slices is measured at nine intersection points to form the thickness distribution. The thickness distribution is used to characterize the thickness dispersion of the same batch of carrot slices in the tray.

[0051] Meanwhile, the initial weight of the entire batch is read by the weighing unit after the tray is stabilized, and the initial weight of the entire batch is divided by the effective bearing area of ​​the tray to obtain the loading weight corresponding to the unit tray area. Then, at the end of each collection cycle, the weight of the current collection cycle is read. The weight of the current collection cycle corresponds to the real-time weight of the remaining material on the tray at the end of the cycle. The initial weight of the entire batch and the weight of the current collection cycle both use the same zero-point calibration result to avoid weighing drift affecting the subsequent moisture judgment.

[0052] A visible light camera, fixedly mounted above the tray, acquires a visible light image of the entire tray under constant illumination. Based on the outer contour of the tray, the visible light image is divided into an edge region image and a central region image. The edge region image is an image band of a predetermined width extending inward from the outer edge of the tray, and the central region image is the remaining image region after deducting the edge region. After grayscale processing, the reflected grayscale values ​​of the surface image are extracted from the divided images. The edge region image and the central region image are then stored separately for subsequent analysis.

[0053] Near-infrared spectral data are synchronously acquired by a near-infrared probe that maintains a fixed distance from the upper surface of the tray. The probe scans the edge region and the center region once in each acquisition cycle and outputs the corresponding raw spectral vector. The raw spectral vector serves as the input source for near-infrared moisture characteristics.

[0054] The surface temperature of the material is collected by an infrared temperature measurement module arranged above the tray, the inlet air temperature is collected by an inlet air temperature sensor, the return air temperature is collected by a return air temperature sensor, the exhaust air humidity is continuously collected by a humidity sensor on the exhaust duct, and the wind speed is collected by a wind speed sensor inside the drying chamber. The above temperature, humidity and wind speed data are aligned with the image data and weighing data according to the same timestamp.

[0055] Among them, the The exhaust humidity at the end of each collection cycle is recorded as . , No. The exhaust humidity at the end of each collection cycle is recorded as . The collection cycle duration is denoted as The rate of change of relative humidity according to Confirmed, among which The relative humidity change rate is represented by rh, relative humidity is represented by out, and exhaust air is represented by out. Indicates the first Exhaust humidity at the end of each collection cycle Indicates the first Exhaust humidity at the end of each collection cycle Indicates the collection period number. This indicates the duration of the data collection cycle between two consecutive data collections;

[0056] Therefore, the thickness distribution, the loading weight per unit pallet area, the initial weight of the whole batch, the weight of the current acquisition cycle, the reflective gray value of the surface image, the near-infrared spectral data, the material surface temperature, the inlet air temperature, the return air temperature, the exhaust air humidity, the relative humidity change rate, and the wind speed together constitute the batch status data output in step S1, providing a unified input for subsequent moisture content estimation and stage discrimination.

[0057] In this specific embodiment, S2 includes:

[0058] Moisture content estimation was performed based on batch state data, starting with obtaining the maximum thickness from the nine-point thickness measurement results. Minimum thickness and average thickness , and according to Determine the thickness dispersion coefficient ,in This indicates the maximum thickness among the nine thickness measurement points. This indicates the minimum thickness among the nine thickness measurement points. This represents the average thickness measured at nine points. Pick Millimeters to avoid the denominator being zero;

[0059] The reflective grayscale value of the surface image is denoted as The baseline reflective grayscale value after loading is recorded as follows: The absolute difference between the two divided by Surface water index obtained ,in These are baseline values ​​collected under conditions of an empty tray, constant lighting, and fixed camera exposure. The larger the value, the more significant the surface water adhesion;

[0060] Near-infrared moisture characteristics are used The absorbance at six fixed wavelengths (1940nm) and its first derivative form a 12-dimensional input vector. This vector is first standardized using Z-scores based on the mean and standard deviation of the calibrated samples. Then, it is input into a three-output partial least squares regression model with a fixed number of four latent variables. The model is calibrated offline using 60 samples whose true values ​​were measured by the drying method. The output is an estimate of the total water content. Free water estimate and estimated value of bound water ;

[0061] rate of moisture loss according to Determined, in the formula This indicates the weight at the end of the previous data collection cycle. Indicates the weight of the current data collection period. Indicates the duration of the sampling period and the surface evaporation intensity. The rate of moisture loss is combined with the rate of change of relative humidity. Make corrections, specifically take ;

[0062] Material temperature hysteresis From the surface temperature of the material With return air temperature The difference is determined, that is ,in This indicates the surface temperature of the material in the current cycle. Indicates the return air temperature for the current cycle;

[0063] Internal migration constraint coefficient according to Determined, in the formula Indicates the proportion of bound water to total water. Indicates will according to The value after upper bound normalization and truncation to the interval between 0 and 1. Indicates will The value after normalization to the upper limit of 0.40 and truncation to the interval between 0 and 1 0.20 is a fixed weight and the sum of the three is 1;

[0064] Edge region image taken from the outer edge of the tray inward The image band width is defined as the central region, which is the remaining area after deducting the edge regions. The water content estimate at the edges is also included. The water content at the center is estimated by weighting the average color change of the edge region image, the shrinkage rate of the edge region image, the estimated total water content of the edge region, and the estimated free water content of the edge region with fixed weights of 0.25, 0.25, 0.30, and 0.20. The total water estimate, free water estimate, and average thickness corresponding to the central region are calculated as follows: The color change is obtained by weighting with a fixed weight of 0.20, where the color change is determined by the difference between the average HSV luminance component of the edge region in the current period and the luminance component of the tray reference, and the shrinkage rate is determined by the reduction ratio of the current projected area of ​​the edge region relative to the initial projected area of ​​the tray.

[0065] Will and After converting to a uniform percentage of mass moisture content, the calculation is performed. The edge-center water cut difference was estimated. And according to the preset mapping table Mapped to uniformity constraint coefficients , among which when hour Take 0.10, when hour Take 0.30, when hour Take 0.60, when hour Taking 1.00, the result is... and Together they constitute the water content state parameters output in step S2.

[0066] In this specific embodiment, S3 includes:

[0067] A threshold discriminator consisting of three sequential comparison units is used to identify the stages of water content parameters. The discrimination process is performed once after each acquisition cycle, and the stages are judged sequentially in the order of surface water peeling, internal moisture migration and maintenance, and equilibrium final drying. This is to avoid inconsistent outputs when multiple stage conditions are met simultaneously within the same cycle. The current dehydration stage number is denoted as […]. ,in This indicates the stage of surface water peeling. This indicates the stage of internal moisture migration and maintenance. The equilibrium final stage is represented by the specific discrimination rule written as follows: correspond and correspond and correspond and ,in Pick , Pick Pick , Pick , Pick , Pick , Pick , Pick , Pick , Pick ,in The surface water index. The relative humidity change rate The rate of water loss. This is the material temperature lag. For the estimate of free water, This is the internal migration constraint coefficient. This is an estimate of the total water volume. This is an estimate of the edge-center water cut difference, and The temperature difference between the material surface temperature and the return air temperature obtained in step S2 is used to reflect whether heat has been transferred into the material and to perform thermal hysteresis verification at the stage boundary. If none of the three conditions are met during the judgment, the temperature determined in the previous collection cycle is used. This serves as the current dehydration stage, and the results will be reassessed based on the latest parameters in the next collection cycle.

[0068] In this specific embodiment, S4 includes:

[0069] A rolling time-domain model is used to predict the controller and generate the control sequence. The sampling period of the controller is the same as the acquisition period in step S1. Consistent, prediction time domain Fixed at 2, controlling the time domain Fixed at 2, state vector Based on the current total water estimate Marginal-central water cut difference estimate and internal migration constraint coefficient Composition, input vector By wind temperature set value Wind speed setpoint Relative humidity setpoint Dehumidification valve opening and recirculated air ratio Composition, in which Indicates the first Hot air temperature setpoint for each control cycle Indicates the first Hot air velocity setpoint for each control cycle Indicates the first The relative humidity setpoint of the drying chamber for each control cycle. Indicates the first The percentage of the dehumidification valve opening in each control cycle. Indicates the first The circulating wind ratio for each control cycle is determined by a first-order discrete affine prediction model internally used in the controller. The model state is determined by... and The model is composed in a fixed order, and the parameters are obtained offline from 60 batches of historical carrot slice drying data using the least squares method. The parameters are fixed after equipment installation and are not re-estimated online. The model is then used based on the current dehydration stage obtained in step S3. Call the corresponding reference trajectory table to make The focus is on removing surface water. At that time, the focus was on stabilizing internal moisture migration. The focus is on ensuring the final drying process is not too brittle; the optimization objective of the controller is written as:

[0070] ;

[0071] in Describe the objective function. This indicates the prediction step size, and takes values ​​of 1 and 2. Indicates the first Predicted water content for one future control cycle, Indicates the target endpoint moisture content. Indicates the first Internal migration constraint coefficient for each control cycle Indicates the first Predicted unit dehydration energy consumption for each future control cycle. The value is obtained by dividing the sum of the predicted power consumption of the heater, fan, and dehumidifier by the predicted mass of moisture removed in that cycle. Indicates the first Uniformity constraint coefficient for each control cycle Indicates the first Predicted edge-center water cut difference for each future control cycle. This represents the control increment vector. Indicates the first The control quantity vector for each control cycle Indicates the first The control quantity vector for each control cycle Represents the L2 norm, Pick Pick Pick Take 0.10;

[0072] When solving the problem, apply variation constraints between adjacent control cycles, ensuring that the wind temperature variation does not exceed [a certain limit]. The wind speed change does not exceed The change in the opening of the exhaust valve should not exceed 5 percentage points, the change in the recirculated air ratio should not exceed 0.05, and the relative humidity setpoint should be within acceptable limits. Limited to to Between, wind temperature setpoint Not exceeding the upper limit of wind temperature given by the brittle retention constraint Furthermore, the control sequence covers at least two subsequent control cycles;

[0073] After obtaining the predictive control sequence, the endpoint confidence calibration unit receives the near-infrared predicted moisture content obtained from the near-infrared moisture characteristics and the total water estimate. Moisture content can be estimated by weighing the initial weight and the weight of the current collection cycle. The cumulative moisture content of exhaust air is obtained by accumulating the exhaust air humidity and wind speed. Marginal water content estimates Central water cut estimate Marginal-central water cut difference estimate and the prediction bias of the most recent collection period ,in The cumulative humidity is calculated by integrating the data over the collection period. The conversion factor is fixed after the equipment is calibrated and used as a residual sample consistency check quantity to eliminate the interference of short-term humidity pulses on the quantile coefficient.

[0074] The endpoint confidence calibration unit uses the most recent 12 acquisition cycles. A residual sample set was constructed and calculated together with 60 batches of offline calibration samples. quantile coefficient and standard deviation Thus, the upper confidence boundary of the central moisture content and the lower confidence boundary of the peripheral moisture content are obtained, written as:

[0075] ;

[0076] in This represents the upper confidence boundary for the central moisture content. This represents the confidence boundary at the marginal moisture content. express quantiles corresponding to confidence levels This represents the sample standard deviation of the central residual. This represents the standard deviation of the marginal residual samples during the first run. The data was taken from 60 batches of offline calibration samples and updated on a rolling basis according to the most recent 12 collection cycles after the operation stabilized.

[0077] when and At that time, a command to stop high-power dehydration will be written into the control sequence, where Pick Pick This instruction serves as the termination marker for the control sequence. The execution adjustment module switches the heater output state in the next cycle while maintaining the current closed-loop correction result unchanged.

[0078] In this specific embodiment, S5 includes:

[0079] After each acquisition cycle, a rolling correction is performed. The adjustment module first reacquires the batch status data and then updates the surface water index using the same fixed model as in steps S2 and S4. Free water estimate Bound water estimate Marginal-central water cut difference estimate Upper confidence boundary of central moisture content and confidence boundary at marginal moisture content subscript Indicates the first For each acquisition cycle, the endpoint confidence calibration unit still uses the residual samples from the most recent 12 acquisition cycles as a window. These residual samples are the difference between the near-infrared predicted moisture content and the moisture content estimated by weighing, and are calculated according to... quantiles corresponding to confidence levels The above two types of confidence boundaries are corrected by rolling adjustments; the control sequence for the current control cycle is denoted as:

[0080] ;

[0081] in Set the air temperature value. Set the wind speed value. Set the relative humidity value. For the opening degree of the dehumidification valve, The recirculation air ratio is adjusted by the control module, which converts it into the target output power of the heater, the fan frequency, the dehumidifier operating level, the exhaust valve opening, and the recirculation air valve opening, respectively.

[0082] in Indicates the first The current dehydration stage number is obtained from the cycle determination, and when and When entering low-temperature pulse dehumidification control, the aforementioned The sixth preset threshold and take Pulse period Take 120 seconds, and within this cycle, limit the wind temperature to no higher than [the specified value]. The wind speed will be limited to 1.2 to... Between, set the opening of the dehumidification valve to Set the recirculated air ratio to 0.35 until... Then switch back to the regular rolling correction sequence;

[0083] After the update At that time, among them Use the preset uniformity threshold from step S3 and take First The reduction will be carried out in 5-percentage-point increments until... Then Increase in increments of 0.05 until... ;

[0084] After the update At that time, among them The upper limit of the target endpoint moisture content and take The control sequence corresponding to the internal moisture migration and maintenance stage is extended by 2 control cycles. During the extension period, the temperature, wind speed, relative humidity, exhaust valve opening and circulating air ratio output in step S4 are kept unchanged.

[0085] After the update At that time, among them To preset the lower limit of edge moisture content in the brittleness retention constraint and take First according to Step size reduced until Then according to Step size reduced until Furthermore, when multiple correction conditions are met simultaneously within the same cycle, they are determined sequentially in the order of low-temperature pulse dehumidification control, edge moisture content lower limit protection, center moisture content upper limit extension, and uniformity correction. Once a previous condition is triggered, the next condition will not be executed.

[0086] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0087] This invention enables the dehydration process of carrot slices to be controlled based on changes in surface-attached water, free water, bound water, and edge-center water content difference through a continuous processing chain between batch state data, water content parameters, and stage discrimination results. This allows energy-saving regulation to correspond to the actual water migration state.

[0088] This invention sets up edge-center moisture content difference compensation, dual-channel stage discrimination of free water and bound water, and endpoint confidence calibration, so that the control sequence not only considers the target endpoint moisture content, but also the risk of excessive dryness at the edge, the hysteresis moisture content at the center, and the brittleness retention constraint, thereby better achieving the consistency of finished product moisture and energy-saving effect.

Claims

1. An energy-saving dehydration method for carrot slices, characterized in that, include: S1. Obtain batch status data for the batch to be dehydrated, including carrot slice thickness distribution, loading density, initial weight, current acquisition cycle weight, surface image reflective grayscale value, near-infrared moisture characteristics, material surface temperature, inlet air temperature, return air temperature, exhaust air humidity, relative humidity change rate, and wind speed; S2. Generate thickness dispersion coefficient, surface water index, total water estimate, free water estimate, bound water estimate, moisture descent rate, surface evaporation intensity, material temperature hysteresis, internal migration restriction coefficient, and edge-center moisture difference estimate based on the batch status data; S3. Based on the surface water index, moisture descent rate, and relative humidity change rate, generate the following data: The current dehydration stage is determined by the humidity change rate, material temperature lag, and internal migration restriction coefficient. The current dehydration stage is either the surface water peeling stage, the internal moisture migration maintenance stage, or the equilibrium final drying stage. S4. A control sequence is generated based on the current dehydration stage, the estimated edge-center moisture content difference, the internal migration restriction coefficient, the target endpoint moisture content, the unit dehydration energy consumption, and the preset brittleness retention constraint. The control sequence includes air temperature, air velocity, relative humidity, exhaust valve opening, and recirculation air ratio. S5. The heater, fan, dehumidifier, exhaust valve, and recirculation air valve are adjusted according to the control sequence, and the control sequence is continuously corrected based on the batch status data obtained in the next acquisition cycle.

2. The energy-saving dehydration method for carrot slices according to claim 1, characterized in that, S1 includes: After the carrot slices are loaded onto the tray and before the heater enters the dehydration power output, the thickness measurement value at different positions of the tray, the loading weight corresponding to the unit tray area, the initial weight of the whole batch, the weight of each collection cycle, the material surface temperature, visible light images and near-infrared spectral data are collected; the visible light images are divided into edge region images and center region images according to the tray position, and the ratio of the exhaust humidity difference between two consecutive collection cycles to the collection cycle duration is used as the relative humidity change rate.

3. The energy-saving dehydration method for carrot slices according to claim 1, characterized in that, S2 includes: The thickness dispersion coefficient is determined based on the maximum, minimum, and average thickness of the thickness distribution; the surface adhering water index is determined based on the grayscale difference between the reflected grayscale value of the surface image and the reference reflected grayscale value after loading; the estimated total water, the estimated free water, and the estimated bound water are determined based on the near-infrared moisture characteristics; the moisture descent rate and the surface evaporation intensity are determined based on the initial weight, the weight of the current acquisition cycle, and the relative humidity change rate; the material temperature hysteresis is determined based on the difference between the material surface temperature and the return air temperature; the internal migration restriction coefficient is a parameter obtained by weighting the estimated bound water value to the estimated total water value, the material temperature hysteresis, and the thickness dispersion coefficient with preset weights.

4. The energy-saving dehydration method for carrot slices according to claim 3, characterized in that, Step S2 further includes: determining the edge water content estimate based on the color change of the edge region image, the shrinkage rate of the edge region image, the estimated near-infrared total water content and the estimated free water content corresponding to the edge region; determining the center water content estimate based on the estimated near-infrared total water content, the estimated free water content and the thickness measurement value corresponding to the center region; converting the edge water content estimate and the center water content estimate to the same water content unit, calculating the difference between the center water content estimate and the edge water content estimate to obtain the edge-center water content difference estimate, and mapping the edge-center water content difference estimate to a uniformity constraint coefficient according to a preset mapping table.

5. The energy-saving dehydration method for carrot slices according to claim 3, characterized in that, S3 includes: when the surface water index is greater than a first preset threshold, the relative humidity change rate is greater than a second preset threshold, and the moisture loss rate is greater than an initial water loss rate threshold, the current dehydration stage is determined to be a surface water stripping stage; when the estimated free water value is not greater than a third preset threshold, the internal migration restriction coefficient is greater than a fourth preset threshold, and the moisture loss rate is not greater than a migration maintenance rate threshold, the current dehydration stage is determined to be an internal moisture migration maintenance stage; when the estimated free water value is less than a fifth preset threshold, the estimated total water value is not greater than the upper limit of the target endpoint moisture content, and the estimated edge-center moisture content difference is not greater than a preset uniformity threshold, the current dehydration stage is determined to be a balanced final drying stage.

6. The energy-saving dehydration method for carrot slices according to claim 4, characterized in that, S4 includes: inputting the current dehydration stage, the uniformity constraint coefficient, the internal migration restriction coefficient, the unit dehydration energy consumption, the target endpoint moisture content, and the preset brittleness retention constraint into the predictive controller; the preset brittleness retention constraint includes the upper limit of wind temperature, the single-cycle wind temperature change boundary, and the lower limit of edge moisture content corresponding to the brittleness evaluation index of the target product; the predictive controller outputs the control sequence covering at least two subsequent control cycles, with the constraints that the changes in wind temperature, wind speed, exhaust valve opening, and circulating air ratio between adjacent control cycles do not exceed the corresponding preset change boundaries.

7. The energy-saving dehydration method for carrot slices according to claim 6, characterized in that, Step S4 further includes: inputting the near-infrared predicted moisture content obtained from the near-infrared moisture characteristics and the total water estimate, the weighing-estimated moisture content obtained from the initial weight and the current acquisition cycle weight, the cumulative exhaust moisture content obtained from the exhaust humidity, the wind speed, and the cumulative moisture content over the acquisition cycle, the edge moisture content estimate, the center moisture content estimate, the edge-center moisture content difference estimate, and the prediction deviation of the most recent acquisition cycle into the endpoint confidence calibration unit, wherein the prediction deviation of the most recent acquisition cycle is the difference between the near-infrared predicted moisture content and the weighing-estimated moisture content. The difference between the rates; the endpoint confidence calibration unit generates an upper confidence boundary for the central moisture content based on the estimated central moisture content and the deviation quantile corresponding to the preset confidence level, and generates a lower confidence boundary for the edge moisture content based on the estimated edge moisture content, the estimated edge-central moisture content difference, and the deviation quantile; when the upper confidence boundary for the central moisture content is not greater than the upper limit of the target endpoint moisture content and the lower confidence boundary for the edge moisture content is not less than the lower limit of the edge moisture content in the preset brittleness retention constraint, a high-power dehydration stop command will be written into the control sequence.

8. The energy-saving dehydration method for carrot slices according to claim 7, characterized in that, S5 includes: After each acquisition cycle, the batch status data is reacquired and the surface water index, free water estimate, bound water estimate, edge-center moisture content difference estimate, upper confidence boundary of center moisture content, and lower confidence boundary of edge moisture content are updated. When the updated edge-center moisture content difference estimate is greater than the preset uniformity threshold, the opening of the exhaust valve is reduced in subsequent control cycles, and the proportion of circulating air is increased after the exhaust valve opening reaches the preset lower limit. When the updated upper confidence boundary of center moisture content is greater than the upper limit of the target endpoint moisture content, the control sequence corresponding to the internal moisture migration maintenance stage is extended. When the updated lower confidence boundary of edge moisture content is less than the lower limit of edge moisture content in the preset brittleness maintenance constraint, the wind temperature or wind speed is reduced in subsequent control cycles.

9. The energy-saving dehydration method for carrot slices according to claim 8, characterized in that, Step S5 further includes: when the current dehydration stage is the surface water peeling stage and the surface water index is greater than the sixth preset threshold, low-temperature pulse dehumidification control is executed; the low-temperature pulse dehumidification control includes limiting the air temperature to within the upper limit of the initial air temperature, limiting the air speed to within the initial air speed setting range, and setting the dehumidification valve opening to the initial dehumidification opening within the preset pulse cycle, until the surface water index is not greater than the sixth preset threshold and then switching to the control sequence.

10. An energy-saving dehydration system for carrot slices, used to perform the energy-saving dehydration method for carrot slices according to any one of claims 1 to 9, characterized in that, include: The batch status acquisition module is used to acquire batch status data of the batches to be dehydrated; The moisture content estimation module is used to generate, based on the batch state data, thickness dispersion coefficient, surface attached water index, total water estimate, free water estimate, bound water estimate, moisture decline rate, surface evaporation intensity, material temperature hysteresis, internal migration restriction coefficient, and edge-center moisture content difference estimate, wherein the internal migration restriction coefficient is a parameter obtained by weighting the bound water estimate to the total water estimate, the material temperature hysteresis, and the thickness dispersion coefficient with preset weights. The stage discrimination module is used to determine the current dehydration stage based on the surface water index, the moisture reduction rate, the relative humidity change rate, the material temperature hysteresis, and the internal migration restriction coefficient; the energy-saving prediction and control module is used to generate a control sequence based on the current dehydration stage, the estimated edge-center moisture content difference, the internal migration restriction coefficient, the target endpoint moisture content, the unit dehydration energy consumption, and the preset brittleness retention constraint; the execution adjustment module is used to adjust the heater, fan, dehumidifier, exhaust valve, and circulating air valve according to the control sequence, and to continuously correct the control sequence based on the batch status data obtained in the next acquisition cycle.

Images