Bean food curing and frying co-processing technology with automatic temperature control function

Abstract

The invention relates to the technical field of bean food processing, and discloses an automatic temperature control bean food curing and frying co-processing technology which comprises the following steps: step 1, obtaining raw material characteristic information of a to-be-processed bean material; step 2, performing curing treatment on the bean materials to be treated; step 3, generating gelatinization characteristic information representing the curing treatment effect according to the monitoring result of the curing treatment process; and step 4, frying the cured beans, and carrying out cooperative adjustment based on the gelatinization characteristic information and the real-time monitoring result of the multi-mode information of the bean state in the frying process. According to the method, the output power of the heating system in the curing treatment process is monitored in real time, the gelatinization end point is judged according to the power curve, judgment can be directly carried out according to the actual physical and chemical reaction state of bean starch gelatinization, and compared with fixed duration and temperature control, the method can more accurately control the gelatinization degree of beans.

Description

Technical Field

[0001] This invention relates to the field of soybean food processing technology, specifically to an automated temperature-controlled co-processing technology for cooking and frying soybean foods. Background Technology

[0002] Legumes play an important role in the food industry due to their rich nutrition and unique flavor. Their processing typically includes key steps such as cooking and frying. These steps aim to induce a series of complex physicochemical changes in the legume raw materials, including starch gelatinization, protein denaturation, moisture evaporation, and Maillard reactions, thereby giving the product the desired golden appearance, crispy texture, and specific flavor.

[0003] In existing practices of soybean food production, ensuring the stability and consistency of the final product quality remains a core technical challenge. This challenge mainly stems from the relatively rigid and rudimentary methods used by existing processes to control various variables during processing.

[0004] On the one hand, as a natural agricultural product, beans inevitably exhibit fluctuations in their initial physicochemical properties (such as moisture content, protein and fat content) between different batches, and even within the same batch. However, traditional production lines often use a fixed set of process parameters (such as fixed cooking temperature and time, fixed frying temperature and time) to process all incoming materials. This processing method cannot adaptively adjust to the actual characteristics of each batch of raw materials. When the properties of the raw materials deviate, the fixed process parameters are no longer the optimal solution, directly leading to significant quality fluctuations in the final product in terms of color, taste, and other aspects.

[0005] On the other hand, in existing processes, each unit operation typically operates as an independent step, lacking effective data transfer between them. For example, the gelatinization process, as a pre-frying step, directly impacts the texture formation and browning rate of the product during subsequent frying. However, in traditional processes, the final state of the gelatinization process is not quantified and transmitted to the frying step. The frying step still starts according to its preset parameters, unrelated to the preceding steps, and cannot compensate for or adjust deviations that occur during the gelatinization process.

[0006] Furthermore, within specific processing stages (such as cooking and frying), control methods are relatively indirect and open-loop. For example, the control of the degree of cooking often relies on a fixed time set empirically, rather than directly monitoring the actual progress of the core reaction of starch gelatinization, leading to the risk of insufficient or excessive gelatinization. Similarly, during frying, the control system typically only maintains a constant oil temperature, lacking real-time, online monitoring and feedback adjustment of key quality indicators of the product itself (such as the rate of internal moisture removal and the rate of surface color formation), thus making it difficult to achieve simultaneous and precise control of the two core quality dimensions of internal crispness and external color. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides an automated temperature-controlled co-processing technology for cooking and frying soybean products. This technology solves the problems of unstable final product quality caused by fixed process parameters, inability to adaptively adjust according to raw material characteristics, lack of data coordination between processes, and lack of real-time closed-loop control in key areas.

[0008] To achieve the above objectives, the present invention provides the following technical solution: an automated temperature-controlled co-processing technology for cooking and frying soybean products, comprising the following steps:

[0009] Step 1: Obtain the raw material characteristic information of the soybeans to be processed.

[0010] Before the soybeans enter the processing flow, a near-infrared spectroscopy analyzer is used to perform a non-contact scan of the batch of soybeans to be processed, acquiring raw spectral data reflecting their internal chemical composition. By processing and analyzing the spectral data, quantified raw material characteristic information is extracted. This raw material characteristic information includes quantified indicators of at least one of the following: initial moisture content, protein content, and fat content of the soybeans to be processed. This step provides initial data for subsequent adaptive processing.

[0011] Step 2: Perform a cooking process on the soybeans to be processed.

[0012] The soybeans, for which raw material characteristic information has been obtained, are fed into a gelatinizer for maturation. Throughout the maturation process, the output power of the heating system is monitored in real time using a power sensor. The maturation process control here no longer relies on fixed time parameters, but is based on the real-time monitoring results of the heating system's output power.

[0013] Specifically, the criterion for determining the end of the maturation process is: after the characteristic peak formed by the endothermic gelatinization of soybean starch is experienced, the output power of the heating system drops back from the peak and stabilizes within the preset baseline power threshold range. This determination method can accurately reflect the actual gelatinization state inside the soybean.

[0014] Step 3: Generate gelatinization feature information that characterizes the effect of this maturation process.

[0015] After the maturation process is completed, the control system generates a set of digitized gelatinization characteristic information based on the complete heating system output power curve monitored and recorded in step 2. This information is used to quantitatively describe the actual effect of this maturation process and is transmitted as information to the subsequent frying process.

[0016] The gelatinization feature information specifically includes the following parameters:

[0017] Total energy absorbed during gelatinization E abs The calculation formula is as follows:

[0018]

[0019] In the formula, P(t) is the instantaneous value of the output power of the heating system, P base The baseline power before the gelatinization reaction occurs, t start and t end These represent the start and end times of the gelatinization process, respectively.

[0020] Peak power P of the heating system output power peak The calculation formula is as follows:

[0021] P peak =max(P(t));

[0022] The duration Δt of the ripening treatment G The calculation formula is:

[0023] Δt G =t end -t start ;

[0024] Step 4: Deep-fry the cooked soybeans.

[0025] The soybeans carrying the gelatinization characteristic information are fed into a continuous drum fryer for frying. The frying process control in this step is a coordinated adjustment process based on prior information and real-time feedback.

[0026] The coordinated adjustment process is as follows: First, the control system sets the initial parameters for frying (such as the initial frying temperature and the conveyor belt reference speed) based on the gelatinization characteristic information transmitted from step 3. Subsequently, during the frying process, the system monitors the multimodal information of the soybean state in real time through sensors, and dynamically adjusts the frying temperature and frying time based on the real-time monitoring results.

[0027] The multimodal information includes acoustic information and optical information:

[0028] Acoustic information is collected by acoustic sensors installed on the frying equipment; it consists of acoustic signals generated by the evaporation of moisture from the soybeans during the frying process. The control process includes adjusting the frying temperature and frying time based on the changing trend of the acoustic signals (such as the rate of signal energy decay) to control the rate of dehydration of the soybeans.

[0029] Optical information is acquired through optical sensors installed on the frying equipment, and consists of optical data characterizing the surface color of the soybeans. The control process further includes adjusting the frying temperature and frying time based on the difference between the optical information and a preset target color range, in order to control the final color of the soybeans.

[0030] The frying process is terminated based on a combination of conditions. These conditions are: a certain characteristic value of the acoustic signal (such as signal energy) reaches a first preset threshold, and the color represented by the optical information enters a second preset target range. This combination of conditions ensures that the finished product meets predetermined standards in both internal crispness and external color.

[0031] This invention provides an automated temperature-controlled co-processing technology for cooking and frying soybean products. It offers the following advantages:

[0032] 1. This invention monitors the output power of the heating system during the maturation process in real time and dynamically determines the gelatinization endpoint based on the characteristic peak shape of the power curve. Therefore, it can directly judge the actual physicochemical reaction state of soybean starch gelatinization. Compared with traditional fixed time and temperature control, this invention can more accurately control the degree of soybean gelatinization.

[0033] 2. This invention generates digitized gelatinization characteristic information after the cooking process is completed and transmits it as an input parameter to the frying process. Therefore, it can establish a data link between the independent cooking and frying processes, so that the initial parameter setting of the frying process can be adjusted based on the actual processing effect of the previous process. This achieves coordinated optimization of the upstream and downstream of the process chain and improves the final product quality.

[0034] 3. This invention integrates acoustic and optical sensors to monitor multimodal information during the frying process, and uses this information to implement closed-loop feedback control of frying temperature and time. Therefore, it can simultaneously and directly monitor the dehydration rate related to the crispness of the product's interior and the surface color related to its appearance quality, thus achieving precise control of the key quality indicators of the final product and ensuring the high quality of the finished product.

[0035] 4. By pre-acquiring the raw material characteristic information of the soybeans to be processed, this invention provides a basis for automatically setting the temperature parameters of the subsequent cooking and frying processes. This makes the temperature control of the entire process no longer dependent on fixed manual settings, but can be automatically adjusted and optimized according to the actual characteristics of each batch of raw materials, thereby achieving true automated temperature control. This effectively overcomes the problem of unsuitable processing temperature caused by raw material fluctuations and ensures product quality. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the process flow of the present invention. Detailed Implementation

[0037] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] To better understand the present invention, the above content will be described in detail below with reference to specific embodiments.

[0039] Please see the appendix Figure 1 , Figure 1 This is a schematic diagram of a process flow diagram for an automated temperature-controlled co-processing technology for cooking and frying soybean products, according to an embodiment of the present invention. The process may include the following steps:

[0040] Step 1: Obtain the raw material characteristic information of the soybeans to be processed.

[0041] Step 2: Cook the soybeans to be processed.

[0042] Step 3: Generate gelatinization feature information that characterizes the effect of this maturation process.

[0043] Step 4: Deep-fry the cooked soybeans.

[0044] In step 1, the soybeans to be processed are first scanned using a near-infrared spectroscopy analyzer to obtain their raw spectral data. A pre-defined algorithm model is then used to extract quantified raw material characteristic information from this data. This raw material characteristic information forms a feature vector V. raw It is used to characterize the initial physicochemical state of this batch of soybeans.

[0045] In step 2, the soybeans, for which raw material characteristic information has been obtained, are conveyed to the gelatinizer. The heating system then heats the soybeans. During this process, the output power of the heating system is monitored in real time by a power sensor connected to the system, and the gelatinization process is controlled based on this real-time monitoring result, particularly for determining the end point of the gelatinization process.

[0046] In step 3, after the cooking process is complete, digitized gelatinization characteristic information is generated based on the heating system output power data recorded in step 2, which covers the entire cooking process. This gelatinization characteristic information quantitatively describes the actual processing effect of this cooking process and is used to guide subsequent frying steps. The gelatinization characteristic information can form a feature vector F. G Specifically, it includes, but is not limited to, the following parameters:

[0047] Total energy absorbed during gelatinization E abs The calculation formula is as follows:

[0048]

[0049] In the formula, P(t) is the instantaneous value of the output power of the heating system, P base The baseline power before the gelatinization reaction occurs, t start and t end These represent the start and end times of the gelatinization process, respectively.

[0050] Peak power P of the heating system output power peak The calculation formula is as follows:

[0051] P peak =max(P(t));

[0052] Duration of ripening treatment Δt G The calculation formula is:

[0053] Δt G =t end -t start ;

[0054] In step 4, the cooked soybeans are conveyed to a continuous drum fryer. Subsequently, based on the gelatinization characteristic information obtained in step 3, and the multimodal information of the soybean state acquired in real time from multiple sensors built into the continuous drum fryer, the frying temperature and frying time of the continuous drum fryer are coordinated and controlled. In this way, the adaptive response of the subsequent process to the processing results of the previous process is realized, as well as closed-loop feedback control in its own processing process.

[0055] In a specific embodiment:

[0056] In step 1, before the raw materials enter the gelatinizer, they are conveyed by a conveyor belt, and a near-infrared spectrometer is used, which is positioned above the conveyor belt, so that its optical lens can perform non-contact continuous scanning of the soybeans to be processed spread on the conveyor belt.

[0057] During the scanning process, the near-infrared spectrometer emits near-infrared light within a specific wavelength range (e.g., 900-1700 nm) into the soybean and receives its reflectance spectrum, subsequently acquiring the raw spectral data S from the near-infrared spectrometer. This raw spectral data S is a set containing absorbance or reflectance values ​​at multiple wavelength points. To eliminate noise interference caused by physical factors such as sample surface scattering and optical path variations, the raw spectral data S is first preprocessed. Preprocessing algorithms include, but are not limited to, multivariate scattering correction (MSC) or standard normal variable transformation (SNV).

[0058] After preprocessing, a pre-constructed chemometric correction model is used to extract quantified raw material characteristic information from the spectral data. In this embodiment, the correction model is a partial least squares (PLS) regression model. The preprocessed spectral data is then used as input, and the PLS model is used for calculation, outputting a set of quantified index values. These index values ​​constitute the raw material characteristic information characterizing the intrinsic properties of this batch of soybeans. This raw material characteristic information is organized into a feature vector V. raw Its specific form is:

[0059] V raw = [w, p, f];

[0060] Where w is the initial moisture content of the soybeans to be treated, p is the protein content, and f is the fat content. These values ​​are calculated using a calibration model calibrated with standard chemical measurement methods (such as oven drying and Kjeldahl nitrogen determination).

[0061] In another extended embodiment, after obtaining raw material characteristic information V raw Next, an adaptive process model selection step is performed. In this embodiment, a process model library is pre-established. This library contains multiple adaptive process models {M1, M2, ..., M...}. n}, each model M i Each corresponds to a set of processing parameters for a specific range of raw material properties. This parameter set includes the target power curve morphology parameters for the cooking process, the initial temperature and initial frying time for the frying process, and thresholds for various feedback controls.

[0062] Based on the obtained raw material characteristic information V raw The numerical values ​​are matched from the process model library, and an optimal adaptive process model M is loaded. bestThe matching logic can be based on a minimum distance criterion. For example, each model has a central feature vector V associated with it. center,i Calculate the eigenvector V of the current raw material. raw With each model's central feature vector V in the library center,i Euclidean distance D i And select the model corresponding to the one with the smallest distance:

[0063]

[0064] M best =M k where D k =min(D i );

[0065] In the formula, w i p i f i Model M i The central eigenvector V center,i The components of M; best M is the optimal adaptive process model selected from the process model library that best matches the raw material characteristics of the soybeans to be processed. k This refers to the k-th adaptive process model in the process model library, where the subscript k indicates the model that makes distance D... k The index of the model that reaches the minimum value.

[0066] min(.) is a function that takes the minimum value from the set of all calculated distance values ​​{D1, D2, ..., D...}. n Find the smallest value in}.

[0067] D i V is the raw material feature vector of the soybeans to be processed. raw Compared with the i-th adaptive process model M in the process model library i The associated central eigenvector V center,i The Euclidean distance between them; D k It is all the calculated distances D i The minimum value in.

[0068] If selected M best All the parameters contained within it will be loaded to guide the subsequent maturation process in step 2 and the frying process in step 4.

[0069] After step 1, once the raw material characteristic information is obtained, the soybeans are fed into a gelatinizer for maturation, and the heating system of the gelatinizer is set to a target operating temperature.

[0070] Throughout the maturation process, a power sensor connected to the heating system of the gelatinizer continuously monitors and records the instantaneous output power P(t) of the heating system at a preset sampling frequency (e.g., once per second). These data points constitute a time series, i.e., a power curve. The process control of the maturation process, especially the determination of its end point, is based entirely on the real-time analysis of this power curve, rather than relying on a fixed processing time.

[0071] Specifically, the power curve P(t) exhibits a typical shape. In the initial heating phase, the power P(t) remains at a low baseline level. base This power is primarily used to maintain the stability of the system temperature. When the starch in the soybeans begins to absorb a large amount of water and heat and undergoes a gelatinization reaction, because this is an endothermic process, the output power of the heating system will rise sharply to maintain the set target temperature, thus forming a significant characteristic peak on the power curve. As the gelatinization reaction of the soybeans nears completion, the heat absorption rate decreases, and the output power of the heating system subsequently drops from the peak value P. peak It has fallen back.

[0072] In this embodiment, the criterion for determining the end of the aging process is precisely defined as: after experiencing a characteristic peak, the output power P(t) of the heating system decreases from the peak value P. peak It falls back and enters a preset baseline power threshold range [P] base -δ,P base Within +δ], and within this range, it remains stable for more than a preset time length τ. stable Among them, P base τ represents the baseline power before the gelatinization reaction occurs, δ represents the allowable power fluctuation, and τ represents the baseline power. stable The time required to determine stability. Once this composite condition is satisfied, the ripening process is determined to have completed at the current time t. end End the process and immediately stop heating.

[0073] Immediately after the maturation process is completed, step 3 is executed, which involves generating a set of gelatinization characteristic information to quantitatively characterize the effect of this maturation process based on the recorded complete power curve. This information is organized into a feature vector F. G This is used to transmit information to subsequent frying processes. In this embodiment, the gelatinization feature information F G Specifically, it includes the following three parameters:

[0074] Total energy absorbed during gelatinization E abs This parameter reflects the total heat absorbed by the soybeans from the outside environment during the entire gelatinization process, and its calculation formula is:

[0075]

[0076] In the formula, P(t) is the instantaneous value of the output power of the heating system, P base The baseline power before the gelatinization reaction occurs, t start and t end These represent the start and end times of the gelatinization process, respectively.

[0077] Peak power P of the heating system output power peak This parameter reflects the maximum endothermic rate when the gelatinization reaction is most vigorous, and its calculation formula is:

[0078] P peak =max(P(t));

[0079] Duration of ripening treatment Δt G This parameter reflects the actual time span of the gelatinization reaction, and the calculation formula is:

[0080] Δt G =t end -t start ;

[0081] When the cooked soybeans enter the continuous drum fryer, the initial frying process parameters are first set. This step constitutes the feedforward control link for the subsequent frying process. The core of this feedforward control is that the initial frying process parameters are not fixed values, but are based on the gelatinization characteristic information F generated in step 3 above, which characterizes the actual cooking effect of this batch of soybeans. G It is calculated dynamically.

[0082] Specifically, the initial temperature T of frying fry,init and the initial preset frying time t fry,init Calculations are performed using a pre-defined function model.

[0083] In this embodiment, the function model is a multiple linear regression model. The initial frying temperature T... fry,init and the initial preset frying time t fry,init The calculation formulas include:

[0084] T fry,init =k1·E abs +k2·P peak +k3·Δt G +C T ;

[0085] t fry,init =w1·E abs +w2·P peak +w3·Δt G +C t ;

[0086] In the formula, T fry,initThe initial frying temperature is set; t fry,init The initial preset frying time is set, and this time serves as the reference duration for subsequent closed-loop feedback control; E abs P peak , Δt G These represent the total energy absorbed during the gelatinization process, the peak power of the heating system output, and the duration of the maturation treatment, respectively, in the gelatinization feature information generated in the preceding steps; k1, k2, and k3 are preset weighting coefficients used to calculate the initial temperature; C T w1, w2, and w3 are preset bias constants used to calculate the initial temperature; w1, w2, and w3 are preset weighting coefficients used to calculate the initial time; C t This is a preset bias constant used to calculate the initial time.

[0087] Weighting coefficients k1, k2, k3, w1, w2, w3 and bias constant C T C t These coefficients were pre-calibrated and stored using statistical regression analysis. They establish a direct mathematical relationship between the quantitative results of the gelatinization process and the initial conditions of the frying process.

[0088] By performing this calculation, a set of perfectly matched initial frying parameters can be set for each batch of soybeans that has undergone a specific cooking process. Once set, the heating system of the continuous drum fryer adjusts the oil temperature to the calculated T. fry,init and with t fry,init The initial control target duration is used to initiate subsequent real-time feedback control of the frying process.

[0089] After setting the initial frying process parameters based on the gelatinization characteristics, the frying process enters the real-time feedback control stage. In this stage, the frying temperature is dynamically adjusted by online monitoring of acoustic information directly related to the internal dehydration rate of the soybeans, in order to control the formation of crispness inside the product.

[0090] In this embodiment, multiple high-sensitivity acoustic sensors (e.g., high-temperature resistant capacitive microphones) are arranged above the exhaust port or cavity of the drum-type continuous fryer. The function of these sensors is to collect the sound signals generated during the frying process when the internal moisture of the soybeans evaporates violently and escapes from the surface.

[0091] An acoustic sensor continuously acquires sound pressure signals at a preset sampling rate. The acquired raw time-domain signal is processed to extract an index characterizing the sound energy intensity. In this embodiment, the real-time acoustic intensity index I is obtained by calculating the root mean square (RMS) value of the signal within a specific time window. a (t). This indicator I aThe value of (t) is directly related to the evaporation rate of moisture inside the soybeans, and it decreases as the frying time increases.

[0092] In one embodiment, the adaptive process model M selected in step 1 best The middle part includes a preset target acoustic intensity attenuation curve I. a,target (t). This curve describes the expected trajectory of acoustic intensity over time under ideal frying conditions. This curve can be defined as a function, such as an exponential decay function:

[0093] I a,target (t)=A·exp(-λ·t / t fry,init )+C noise ;

[0094] In the formula, A is the initial sound intensity amplitude, λ is the attenuation coefficient, and C... noise The background noise baseline of the equipment is given, and t is the current frying time. fry,init The initial preset frying time set during the feedforward control phase. Parameters A, λ, and C noise All are derived from the selected process model M best supply.

[0095] At any time t during the frying process, calculate the acoustic intensity index I measured in real time. a (t) and the expected value I on the target acoustic intensity attenuation curve a,target The deviation e between (t) a (t):

[0096] e a (t)=I a,target (t)-I a (t);

[0097] Based on this deviation e a (t), and a proportional-integral-derivative (PID) control algorithm is used to calculate an adjustment amount ΔT for the frying temperature. a (t):

[0098]

[0099] In the formula, K p,a K i,a K d,a These are the proportional, integral, and derivative gain coefficients for acoustic feedback control, respectively, and their values ​​are also determined by the selected adaptive process model M. best Preset; For deviation e a (t) Rate of change over time.

[0100] Calculated temperature adjustment ΔTa (t) is used to update the temperature setpoint of the drum-type continuous fryer. The new temperature setpoint T fry,set Calculate as follows:

[0101] T fry,set (t)=T fry,prev-set +ΔT a (t);

[0102] In the formula, T fry,prev-set This is the temperature setpoint from the previous control cycle. In this way, when the actual dehydration rate (determined by I)... a (t) reflects the deviation e when the rate is lower than the expected rate. a When (t) is positive, the temperature setpoint is increased to accelerate dehydration; conversely, it is decreased, thus achieving closed-loop control over the formation process of the internal structure of soybeans.

[0103] In another embodiment, parallel to the acoustic-based temperature feedback control is an optical-based feedback control process. This process dynamically adjusts the total frying time by monitoring changes in the surface color of the soybeans online, thereby achieving precise control over the final product's appearance quality.

[0104] In this embodiment, a machine vision system is deployed above the drum-type continuous fryer. The system includes a digital industrial camera and a standardized light source for providing uniform and stable illumination, and is configured to view and continuously photograph the soybeans being processed in hot oil.

[0105] During the frying process, the machine vision system acquires a digital image of the soybeans at preset time intervals (e.g., every 2 seconds). For each acquired image frame, the soybean region in the image is first identified using an image segmentation algorithm, and the average RGB color value of all pixels within that region is calculated. Subsequently, this average RGB value is converted to a device-independent, standardized CIELAB color space, thereby obtaining a set of real-time, quantitative chromaticity coordinates (L). * (t),a * (t), b*(t)). Where, L * (t) represents the brightness of the soybean at time t. Its value is directly related to the degree of Maillard reaction and caramelization reaction on the soybean surface and is a key indicator for characterizing the color of the product.

[0106] In this embodiment, the optical feedback control adjusts the running speed v of the conveyor belt in the continuous fryer. c This is used to change the total residence time of the soybeans in hot oil, i.e., the total frying time. The control objective is to increase the real-time brightness value L of the soybeans. * (t) can accurately follow the preset target brightness change curve The target curve is stored in the adaptive process model M selected in step 1. best The text describes the preset ideal target final brightness value. The trajectory of change.

[0107] At any time t during the frying process, calculate the real-time measured brightness value L. * (t) and the expected value on the target brightness change curve The deviation between e c (t):

[0108]

[0109] Based on this deviation e c (t), the adjustment amount Δv for the conveyor belt speed is calculated by a proportional-integral-derivative (PID) control algorithm. c (t):

[0110]

[0111] In the formula, K p,c K i,c K d,c These are the proportional, integral, and derivative gain coefficients for optical feedback control, respectively, whose values ​​are determined by the selected adaptive process model M. best Preset; For deviation e c (t) rate of change over time;

[0112] Calculated speed adjustment Δv c (t) is used to update the conveyor belt speed setting. The new speed setting value is Δv. c,new (t) is calculated as follows:

[0113] Δv c,new (t)=Δv c,prev (t)-Δv c (t);

[0114] In the formula, Δv c,prev (t) is the speed setpoint from the previous control cycle. In this way, when the actual browning rate of the soybeans is slower than the desired rate (i.e., L...),... * (t) is higher than ), deviation e c When (t) is positive, the conveyor belt speed is reduced, thereby extending the residence time of the soybeans in the hot oil to accelerate browning; conversely, when (t) is negative, the speed is increased to shorten the time, thus achieving closed-loop control of the color of the final product.

[0115] In another extended embodiment, the end of the frying process is not determined by a single fixed duration or a single sensor index, but rather by a composite judgment of information from multiple dimensions to determine the final frying endpoint. This method ensures that the finished product meets preset quality standards in both appearance and internal texture.

[0116] The determination of the frying endpoint is based on the following three conditions monitored in parallel:

[0117] The first condition is the optical endpoint condition. This condition is directly related to the final appearance and color of the product. During the frying process, the brightness value L of the soybean surface is monitored in real time. * (t). When the real-time brightness value L * (t) This condition is met when the target final brightness range is reached or entered.

[0118] Specifically, this condition can be expressed as:

[0119]

[0120] In the formula, L * (t) represents the surface brightness value of the soybeans measured at time t; The final brightness value of the preset ideal target; ∈ L This represents a very small brightness tolerance range. Parameters and ∈ L All are derived from the adaptive process model M selected in step 1. best supply.

[0121] The second condition is the acoustic endpoint condition. This condition is directly related to the product's internal crispness and final moisture content. During frying, the acoustic intensity index I generated by moisture evaporation is monitored in real time. a (t). When the acoustic intensity decays to below a preset endpoint threshold, it indicates that internal dehydration is essentially complete, and the expected crispness has been achieved; at this point, the condition is satisfied. Specifically, this condition can be expressed as:

[0122] I a (t)≤I a,end (t);

[0123] In the formula, I a (t) represents the acoustic intensity index measured at time t; I a,end (t) represents the preset acoustic endpoint threshold. This threshold is also determined by the selected adaptive process model M. best supply.

[0124] The third condition is the time limit condition. This condition serves as a protective measure to handle situations such as sensor malfunctions or extreme deviations in raw material characteristics. At the start of frying, the initial preset frying time t is calculated based on the feedforward control stage.fry,init Set a maximum allowable frying time t max This condition is met when the total frying time reaches this upper limit. Specifically, this condition can be expressed as:

[0125] t total ≥t max ;

[0126] In the formula, t total t represents the total frying time already elapsed; max The calculation method is as follows:

[0127] t max =α·t fry,init ;

[0128] In the formula, α is a preset safety factor.

[0129] In this embodiment, the termination command for frying is triggered only when the following logical relationship is met: ((the first condition is met) and (the second condition is met)) or (the third condition is met). That is, under normal circumstances, both the optical endpoint condition and the acoustic endpoint condition must be met simultaneously for frying to be considered complete; while the time limit condition serves as a mandatory termination condition independent of the former two. Once the termination command is triggered, the drum-type continuous fryer immediately discharges the finished product from the hot oil, completing the entire frying process.

[0130] In one extended embodiment, the process of establishing the process model library includes an experimental data acquisition phase and a model building phase.

[0131] Specifically:

[0132] During the data acquisition phase, multiple batches of soybean samples covering a wide range of physicochemical properties were prepared. Standard chemical measurement methods, such as oven drying, Kjeldahl nitrogen determination, and Soxhlet extraction, were used to accurately determine the initial moisture content (w), protein content (p), and fat content (f) of each batch of soybean samples, constructing their raw material characteristic vector (V). raw =[w,p,f].

[0133] Subsequently, these soybean samples with precisely calibrated attributes were processed one by one using the cooking-frying synergistic processing technology of this invention. During the processing, for each batch of samples, the process parameters used were systematically changed and recorded, including but not limited to: the target working temperature for cooking, and the initial temperature T for frying. fry,init Initial preset frying time t fry,init PID gain coefficient (K) in acoustic feedback control p,a ,K i,a ,K d,a ), PID gain coefficient (K) of optical feedback controlp,c ,K i,c ,K d,c ), and various thresholds for determining the frying endpoint.

[0134] After each experiment, the finished product undergoes an objective and quantitative quality assessment. Key quality indicators are measured using instruments (e.g., colorimeter, texture analyzer), such as the CIELAB value of surface color, the maximum breaking force characterizing crispness, or the acoustic fracture signal. Based on these objective indicators, the corresponding quality grade of the finished product for each set of process parameters is determined.

[0135] In the model building phase, the first step is to process all the collected raw material feature vectors V. raw The dataset is processed. An unsupervised machine learning algorithm, such as k-means clustering, is used to divide the multidimensional raw material feature space into N non-overlapping regions or clusters. Each cluster represents a class of raw materials with similar intrinsic properties. The centroid of each cluster is calculated, and this centroid serves as the representative central feature vector V of that class of raw materials. center,i .

[0136] Next, for each identified raw material cluster, the combination of process parameters that can stably produce the highest quality finished product is selected from all the experimental data corresponding to that cluster. This combination is considered the optimal process setting for that type of raw material.

[0137] Finally, the central feature vector V of each raw material cluster is... center,i It is correlated with its corresponding optimal combination of process parameters to form an independent adaptive process model M. i A complete adaptive process model M i The following data is explicitly stored in it:

[0138] The central feature vector corresponding to this model:

[0139] V center,i =[w i ,p i ,f i ].

[0140] The feedforward control model coefficients (k1, k2, k3, C) used to calculate the initial parameters of frying are... T ) and (w1,w2,w3,C t ).

[0141] Target curve parameters used for feedback control, such as the target acoustic intensity attenuation curve I a,target The coefficients (A,λ) of (t) and the target brightness change curve

[0142] PID gain coefficient (K) used for feedback control p,a ,K i,a ,K d,a ) and (K p,c ,K i,c ,K d,c ).

[0143] Thresholds for various dimensions used to determine the frying endpoint, including the final target brightness value. Brightness tolerance ∈ L Acoustic endpoint threshold I a,end (t) and the time limit safety factor α.

[0144] All n established adaptive process models {M1, M2, ..., M n These elements, when combined, form a complete process model library. This library is stored for use in actual production. By matching the distance between the real-time detected raw material feature vectors and the central feature vectors of each model in the library, a set of optimal process parameters suitable for any batch of soybeans to be processed can be loaded.

[0145] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An automated temperature-controlled co-processing technology for cooking and frying soybean products, characterized in that, Includes the following steps: Step 1: Obtain the raw material characteristic information of the soybeans to be processed; Step 2: The soybeans to be processed are subjected to a cooking process, wherein the cooking process is controlled based on the real-time monitoring results of the output power of the heating system during the cooking process. Step 3: After the maturation process is completed, based on the monitoring results of the maturation process, generate gelatinization characteristic information that characterizes the effect of this maturation process; Step 4: Deep-fry the cooked soybeans. The deep-frying process is controlled by adjusting the gelatinization characteristic information and the real-time monitoring results of the multimodal information of the soybean state during the deep-frying process.

2. The automated temperature-controlled co-processing technology for cooking and frying soybean products according to claim 1, characterized in that, In step 1, obtaining the raw material characteristic information of the soybeans to be processed specifically includes: The soybean material to be treated was scanned using a near-infrared spectroscopy analyzer to obtain spectral data, and the raw material characteristic information was extracted from the spectral data.

3. The automated temperature-controlled co-processing technology for cooking and frying soybean products according to claim 2, characterized in that, The raw material characteristic information includes at least one of the following: initial moisture content, protein content, and fat content of the soybeans to be processed.

4. The automated temperature-controlled co-processing technology for cooking and frying soybean products according to claim 1, characterized in that, In step 2, the conditions for determining the end of the ripening process include: After experiencing a characteristic peak formed by the endothermic reaction of soybean gelatinization, the output power of the heating system drops back from the peak and stabilizes within the preset baseline power threshold range.

5. The automated temperature-controlled co-processing technology for cooking and frying soybean products according to claim 1, characterized in that, In step 3, the generated gelatinization feature information includes: The total energy absorbed during the gelatinization process, the peak power of the heating system output, and the duration of the maturation treatment are calculated based on the output power curve of the heating system.

6. The automated temperature-controlled co-processing technology for cooking and frying soybean products according to claim 1, characterized in that, In step 4, the process control of the frying treatment specifically includes: Based on the gelatinization characteristic information, the initial parameters of the frying process are set; During the frying process, the frying temperature and frying time are dynamically adjusted based on the real-time monitoring results of the multimodal information.

7. The automated temperature-controlled co-processing technology for cooking and frying soybean products according to claim 6, characterized in that, The multimodal information includes acoustic signals collected by acoustic sensors that are generated due to the evaporation of moisture from soybeans; The process control of the frying treatment includes: adjusting the frying temperature and frying time based on the changing trend of the acoustic signal to control the dehydration rate of the soybeans.

8. The automated temperature-controlled co-processing technology for cooking and frying soybean products according to claim 7, characterized in that, The multimodal information also includes optical information that characterizes the surface color of soybeans, collected by optical sensors; The process control of the frying treatment further includes: adjusting the frying temperature and frying time based on the difference between the optical information and the target color range to control the color of the soybeans.

9. The automated temperature-controlled co-processing technology for cooking and frying soybean products according to claim 8, characterized in that, The conditions for determining the end of the frying process include: The characteristic value of the acoustic signal reaches a first preset threshold, and the color represented by the optical information enters a second preset target range.

10. The automated temperature-controlled co-processing technology for cooking and frying soybean products according to claim 1, characterized in that, The maturation process is carried out in a gelatinizer; the frying process is carried out in a continuous drum fryer.

Images