Method for dynamically adjusting the control of the fluid extraction of a product
The method dynamically adjusts drying and draining conditions using real-time modeling and automatic parameter adjustments, addressing the variability in manual processes and reducing energy consumption in the food industry.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- CLAUGER
- Filing Date
- 2025-11-10
- Publication Date
- 2026-05-21
Smart Images
Figure EP2025082510_21052026_PF_FP_ABST
Abstract
Description
[0001] DESCRIPTION
[0002] Method for dynamically adjusting the control of fluid extraction from a product - TECHNICAL FIELD
[0003] The present invention relates to a system and a method for dynamically adjusting the control of fluid extraction, such as drying or draining, of products.
[0004] The invention finds a particularly advantageous application for the drying or draining of food or non-food products, such as meat, dairy, or vegetable products, such as cured meats, cheeses, fruits, vegetables, wood, etc.
[0005] STATE OF THE ART
[0006] In the food industry, drying products such as cured meats or cheese, or draining dairy products, are complex and delicate processes.
[0007] For example, in the case of drying in the processed meat sector, drying must be carried out under precise temperature and humidity conditions. Specifically, industrial drying of processed meats is performed using dedicated equipment such as a drying cabinet or chamber, in which the climatic conditions are controlled throughout the drying process.
[0008] In practice, the drying chamber, where the products to be dried are stored, is equipped with devices for controlling and regulating the chamber's climatic conditions throughout the various drying cycles. These drying conditions are predefined by the operator according to the type and characteristics of the product to be dried.
[0009] The drying chamber is typically equipped with sensors to monitor and control drying parameters in real time, such as chamber temperature, humidity, ventilation, etc., based on instructions entered by the operator. Some drying chambers may have systems that detect deviations in drying parameters, alerting the operator via a visual or audible signal so that the operator can manually adjust the drying parameters according to the observed deviations.
[0010] Traditionally, operators must monitor the drying process and manually adjust the drying parameters to achieve a product that meets specifications. The quality of the final product therefore depends primarily on the operators' expertise and manual adjustments based on visual and olfactory observation. This practice leads to variations in the final product's quality and significant energy consumption.
[0011] In the case of dairy products (especially cheeses), draining essentially consists of separating the curd from the whey. The draining phase is a crucial step as it determines the texture, structure, and, to some extent, the flavor of the final product. Generally, draining instructions vary depending on the type (fresh cheese, soft cheese, hard cheese, etc.) and the desired characteristics (creamy, firm, etc.). The draining phase is usually carried out in draining rooms that also provide a controlled environment in terms of temperature and humidity. Furthermore, as with drying, the draining instructions are generally adjusted manually by the operator during the draining process, so the final quality of the product will depend primarily on the operator's expertise and manual adjustments based on visual and olfactory observations.Generally, the operator manually and occasionally measures the pH of the product, visually assesses the whey, etc. This practice therefore also leads to variations in the quality of the final product as well as significant energy consumption.
[0012] For hard cheeses, the draining stage can be followed by a drying period to reduce their water content and prepare the rind before ripening. This drying can also be carried out in a temperature- and humidity-controlled environment. Thus, in currently implemented processes, draining products like cheese relies on traditional practices combining monitoring of environmental parameters (temperature, relative humidity, ventilation), manual and spot checks (pH, visual and olfactory observations), and the use of standardized curves to model the draining or acidification processes. However, these approaches have limitations because fixed models do not account for real-world variations such as differences in weight loss or pH.Furthermore, standardized curves do not incorporate human observations, such as serum color or clarity, into digital systems. Finally, they result in high energy consumption, particularly due to ventilation not adjusted to the actual serum flow, which generates additional costs.
[0013] The present invention addresses this need for more adaptive and efficient solutions.
[0014] PRESENTATION OF THE INVENTION
[0015] The present invention relates to improving the process of controlling the drying or draining of a product by providing a solution for automating the regulation of drying or draining conditions. In particular, the invention offers a solution for automatically and dynamically adjusting the drying or draining conditions throughout the various drying or draining cycles.
[0016] Subsequently, expressions such as "fluid weight reduction", "extraction", "fluid extraction", "fluid reduction",
[0017] "Dehydration" will refer indifferently to drying or draining, the fluid being water, whey, etc.
[0018] To achieve this, the invention implements real-time modeling of the fluid extraction profile (drying or draining, for example), with automatic adjustment of the parameters related to fluid extraction throughout the extraction cycle in order to achieve product quality and optimize the energy efficiency of the extraction cycles. The invention thus relates to a method for dynamically adjusting the control parameters for fluid extraction in products, such as those based on plants, meat, or milk, stored in a chamber (for drying or draining, for example), throughout the various fluid extraction cycles.
[0019] The invention also relates to a system configured to implement the dynamic adjustment process, and a drying control method based on the dynamic adjustment process.
[0020] According to the invention, the method for dynamically adjusting the control parameters of the extraction of products stored in an extraction chamber, throughout the different extraction cycles, the extraction including drying or draining, comprises:
[0021] - an initial phase of:
[0022] . determination of a target weight loss profile based on at least the type of product, the initial weight of the product, the duration of the extraction cycle, and the target weight loss of the product at the end of the cycle, the target weight loss profile defining the theoretical evolution of the weight loss of the products during the extraction cycle, . determination of the climatic extraction conditions to be applied in the chamber throughout the extraction cycle responding to the theoretical evolution of the weight loss and the product quality;
[0023] - an extraction phase following the target weight loss profile determined in the initial phase, during which the following actions are carried out:
[0024] a / application in the chamber of climatic conditions corresponding to the target weight loss profile;
[0025] b / real-time measurement throughout the extraction cycle of quantitative data relating to the extraction and including at least the actual weight loss, the temperature and humidity level of the chamber, and the ventilation speed;
[0026] c / real-time establishment of the actual weight loss profile;
[0027] d / real-time establishment of a predictive weight loss profile based on current quantitative data;
[0028] e / automatic adjustment of the climatic conditions to be applied in the chamber if a predefined deviation between the target profile and the actual profile is observed, or if a predefined deviation between the target profile and the predictive profile is observed; and application of these adjusted climatic conditions for the duration of the remaining cycle;
[0029] f / implementation of steps b / to e / until the end of the extraction cycle.
[0030] The automatic adjustment step e includes the application of one or more corrective actions including the adjustment of device settings impacting the temperature in the chamber, the ventilation speed, the humidity in the chamber, the cycle duration, etc.
[0031] In other words, the dynamic adjustment process includes:
[0032] An initial phase of determining a target weight loss profile and determining the climatic conditions to be applied in the chamber throughout the extraction period that meet the target weight loss profile.
[0033] Determining the target weight loss profile takes into account, in particular:
[0034] - the type of product (to be dried or drained), and possibly the fluid extraction characteristics (drying or draining) induced by the composition of the product; - the initial weight of the product;
[0035] - the duration of the extraction cycle; and
[0036] - the target weight loss of the product cumulative at the end of the extraction cycle.
[0037] The target weight loss profile defines the theoretical evolution of the (fluid) weight loss of the products during the extraction cycle. The climatic conditions to be applied (temperature, humidity, ventilation speed, etc.) in the chamber throughout the extraction period correspond to this theoretical weight loss profile.
[0038] Following this initial phase, an extraction phase, based on the target weight loss profile determined in the initial phase, is implemented during which the following actions are carried out:
[0039] - application in the chamber of climatic conditions corresponding to the target weight loss profile;
[0040] - real-time measurement throughout the extraction cycle of quantitative data relating to fluid extraction, and real-time establishment of the actual weight loss profile; - determination of a predictive weight loss profile for the remaining cycle duration if the climatic conditions corresponding to the target weight loss profile are maintained; - automatic adjustment of the climatic conditions to be applied in the chamber if a predefined deviation between the target profile and the actual profile is observed, or if a predefined deviation between the target profile and the predictive profile is observed; and
[0041] - application of these adjusted climatic conditions.
[0042] Quantitative data relating to the ongoing fluid extraction include, for example: - the actual weight loss of the product;
[0043] - the room temperature;
[0044] - the relative humidity of the room; and / or
[0045] - ventilation speed, etc.
[0046] The model representing the target weight loss profile over time can be defined by the following general equation:
[0047] I 1 — e T
[0048] P(t) = P cy
[0049]
[0050] \1 — e~~
[0051] Or
[0052] P(t) is the weight loss at time t;
[0053] P cy is the weight loss of the product at the end of the extraction cycle;
[0054] t C y is the extraction cycle time; and
[0055] T is a time constant related to the properties of the product and the extraction conditions.
[0056] The model representing the target weight loss profile over time can also be defined by a polynomial, logarithmic, multi-exponential type model.
[0057] In one embodiment, the extraction can be draining, and the quantitative data further include the acidity of the product, via continuous pH measurement within the product, for example, directly in the core of the dairy product or in a representative solution. Thus, in the case of a draining cycle, for example of dairy products:
[0058] - the initial phase also includes the determination of a target acidity profile, with the climatic conditions to be applied determined to meet both the target weight loss profile and the target acidity profile;
[0059] - the extraction phase further includes real-time pH measurement and real-time establishment of the actual acidity profile, as well as the establishment of the predictive acidity profile if the applied climatic conditions are maintained; - automatic adjustment of the climatic conditions is carried out if a predefined deviation between the target acidity profile and the actual acidity profile is observed, or if a predefined deviation between the target acidity profile and the predictive acidity profile is observed, the adjustment of the climatic conditions responding to both the target weight loss profile and the target acidity profile; and
[0060] - application of these adjusted climatic conditions.
[0061] The target acidity profile defines the theoretical evolution of the product's pH level during the extraction cycle.
[0062] The asymptotic decay model related to the target acidification profile can be defined by:
[0063] p
[0064]
[0065] HtargetCt) pHinitial ^P^max X — 6
[0066] Or
[0067] pH c ibi e (t) is the pH at time t;
[0068] pHinitiai is the initial pH value at the beginning of the draining process;
[0069] ΔpH max is the maximum expected variation of pH during a theoretical cycle of infinite duration (t → +∞);
[0070] T P H is a time constant, influencing the rate of acidification, and is relative to the properties of the product to be drained and the draining conditions.
[0071] Adjusting the climatic conditions may involve implementing one or more corrective actions, such as adjusting the parameters or settings of devices that impact the climatic conditions in the chamber and thus the evolution of quantitative data, for example, chamber temperature, ventilation speed, chamber humidity, cycle duration, etc. Advantageously, the automatic climatic conditions adjustment step can also take into account qualitative data relating to the operator's observations regarding the products during fluidic extraction. For example, in the case of sausage, qualitative data may include crust formation, firmness, color, etc.For example, in the case of dairy products, qualitative data may include the visual appearance of the whey (color, clarity, visible deposits), the condition of the product (texture, firmness, or other sensory indicator), the olfactory perception, etc.
[0072] Alternatively, depending on the available sensors or analyzers, qualitative data can be included within the quantitative data. For example, the visual appearance of whey can be determined using a suitable optical sensor, the condition of the dairy product can be determined using a suitable analyzer, and the olfactory perception can be determined using sensors such as olfactometers, electronic noses, or any other suitable analyzer.
[0073] Thus, in practice, the dynamic adjustment process may also include: - during the initial phase, a weighting is assigned to each of the qualitative data, during the initial phase, according to the quality of the product to be obtained; and - during the extraction phase, all or part of the weightings are adjusted throughout the extraction cycle according to the observed difference between the actual weight loss and the target weight loss, and / or indications relating to the qualitative data entered by the operator corresponding to the operator's perception of the product for which fluid extraction is in progress;
[0074] - during the extraction phase, the dynamic adjustment of the climatic conditions to be applied throughout the cycle also takes into account the evolution of qualitative data.
[0075] The level or weighting reflects the importance or criticality of the qualitative data on the quality of the final product (dried or drained) or on the extraction process.
[0076] In one variant, the initial phase may also include:
[0077] - the identification of qualitative data relating to the type of products and corresponding to the operator's feelings or observations on the products for which fluid extraction is underway; and
[0078] - the assignment of a level or weighting to each of the qualitative data, depending on the quality of the product to be obtained.
[0079] The extraction phase may also include the adjustment of all or part of the weightings, throughout the extraction cycle, based on the observed difference between the actual weight loss and the target weight loss, and / or indications relating to qualitative data entered by the operator corresponding to the operator's perception of the product for which fluid extraction is in progress.
[0080] The information entered by the operator can take the form of an evaluation or assessment level of the product by the operator for each of the qualitative data points, for example, as weights, scores, ratings, indices, or the positioning of a slider on a rating scale. The system may include an interface configured to allow the operator to enter this data.
[0081] Advantageously, the predictive profile is determined by taking qualitative data into account.
[0082] Thus, depending on the indications (and therefore the feelings) entered by the operator, the system can adjust all or part of the weightings assigned to the qualitative data.
[0083] In addition, when one or more assigned weights are changed, the system implements one or more predefined corrective actions associated with the qualitative data for which the weights are changed.
[0084] Corrective actions may include adjusting humidity, ventilation, chamber temperature, etc. The weighting of each qualitative data point can thus be dynamically adjusted based on the observed weight loss deviation and operator feedback. In practice, the dynamic adjustment of climatic conditions to be applied throughout the cycle therefore takes into account the evolution of both quantitative and qualitative data. Advantageously, the determination of the target weight loss profile and / or the target acidity profile during the initial phase is based on the weight loss and / or target acidity profiles of previous cycles.
[0085] Advantageously, the extraction phase can include a holding period during which no deviation calculation is performed, and the measured quantitative data are recorded but not compared to the target curves.
[0086] Advantageously, the dynamic adjustment process can also include the optimization of energy consumption related to the extraction cycle, consisting of: - measuring energy consumption in real time, including both electrical and thermal consumption;
[0087] - calculate the energy efficiency defined by the ratio of weight loss per kWh consumed; - depending on the difference between the calculated energy efficiency and the predefined minimum acceptable energy efficiency ratio, apply measures aimed at optimizing energy efficiency, such measures including at least one of the following actions: automatic shutdown of the extraction cycle, adjustment of ventilation.
[0088] BRIEF DESCRIPTION OF THE DRAWINGS
[0089] The present invention and its advantages will become more apparent from the following description of several embodiments given by way of non-limiting examples, with reference to the accompanying drawings.
[0090] Figure 1 illustrates a target weight loss profile, a profile of the actual weight loss trend, and a dynamic weight loss prediction profile, specifically for sausage drying and according to one embodiment. Figure 2 illustrates a target weight loss profile and a profile of the actual weight loss trend, specifically for sausage drying and according to one embodiment.
[0091] [Fig 3] is a graph illustrating the evolution of the measurement of the weight of condensed water hour by hour during a sausage drying cycle and according to one embodiment.
[0092] [Fig 4] is a graph illustrating the upper and lower tolerances around the target weight loss profile, specifically for sausage drying and according to one embodiment. [Fig 5] is an example of a user interface adapted to the processed meat sector, according to one embodiment.
[0093] [Fig 6] is a simplified flowchart illustrating the major steps of the dynamic adjustment process according to one embodiment.
[0094] DETAILED DESCRIPTION
[0095] Variant: *** DRYING ***
[0096] The different steps implemented by the system according to an embodiment are detailed below, in the particular case of sausage drying, it being understood that the general principle applies in the case of drying all meat, dairy or vegetable products.
[0097] Step 1: Phase of determining the target weight loss kinetics
[0098] Before the start of each drying cycle (initial phase 100 in Figure 6), the system determines a theoretical or predicted target weight loss profile (step 101), for example, in the form of a predicted target weight loss curve. This target profile is established based on the specific characteristics of the product to be dried. For example, in the case of sausage, the target profile can be established by taking into account:
[0099] - the type of product (e.g., sausage) which determines the drying behavior depending on the composition of the product to be dried (moisture, fat);
[0100] - the initial weight of the product to be dried;
[0101] - the duration of a drying cycle (for example, generally 10 to 30 days for sausages); and
[0102] - the weight loss target, for example in the form of a percentage relative to the initial weight (e.g. 30% weight loss).
[0103] The system establishes, in particular, a target curve of the theoretical or predicted evolution of weight loss during the entire duration of the drying cycle.
[0104] The system can also incorporate historical data from previous drying cycles to refine the target weight loss curve. In other words, before each drying cycle begins, a target weight loss curve is generated based on the characteristics of the product to be dried (initial weight, composition, initial moisture content, fat content, etc.), the theoretical drying time, and the target weight loss (e.g., 30% in 20 days). This information allows the system to define the evolution of the drying climatic conditions to be implemented (such as temperature and humidity) throughout the drying cycle. This target weight loss curve is theoretical or predictive.
[0105] In practice, the model representing weight loss over time can be defined by the following general equation:
[0106] 1 — e T \
[0107]
[0108] 1 — e~ T y
[0109] Or
[0110] P(t) is the weight loss at time t;
[0111] P cy is the weight loss of the product at the end of the extraction cycle;
[0112] tcy is the duration of the extraction cycle; and
[0113] T is a time constant related to the properties of the product and the extraction conditions.
[0114] This weight loss model can be applied to any product to be dried or drained.
[0115] This model describes how weight loss changes over time depending on the product's characteristics. The exponential law reflects the fact that water loss is rapid at first, then slows down over time. The time constant T depends on the characteristics of the product being dried and influences the rate of this change: the lower it is, the faster the weight loss. Thus, the time constant T can be adjusted according to the product's specific characteristics. For a product that loses moisture more slowly, the value of T can be higher. In practice, the constant T is initially estimated and then learned, for a given product, from previous drying cycles.
[0116] The model defined above assumes that the weight loss of the product being dried is constant. This is particularly true for sausages, whose weight loss generally follows a decreasing exponential trend, typical of drying processes. Initially, the loss is faster because free water is easily removed. Over time, the rate of loss slows down because removing the bound water from the product becomes more difficult. This phenomenon is well illustrated in documents relating to sausage drying, where the initial drying phases remove a large amount of water, and the final phase focuses on eliminating residual moisture.
[0117] In figures 1, 2 and 4, the Co curve is an example of a theoretical target weight loss profile for sausage drying.
[0118] In the case of uneven weight loss over time, other more complex models (polynomial, logarithmic, multi-exponential) can be used to adjust the shape of the curve. The following second-order polynomial model is given as an example:
[0119] VP cv tt
[0120]
[0121] With:
[0122] P(t) is the weight loss at time t;
[0123] P cy is the weight loss of the product at the end of the extraction cycle;
[0124] tcy is the duration of the extraction cycle; and
[0125] x (%) is the ratio between the water loss at the end of the cycle (kg / h) and the water loss at the beginning of the cycle (kg / h). This ratio is relative to the product properties and the extraction conditions.
[0126] The system determines the evolution of the drying climatic conditions to be implemented during the drying cycle to achieve the target profile (step 102), as well as the corresponding instructions or controls for the devices or equipment in or connected to the chamber that impact these climatic conditions. These climatic conditions include, for example: temperature, humidity, ventilation speed, etc.
[0127] Step 2: Real-time monitoring of quantitative data and continuous adjustment of climatic conditions. Once the theoretical target weight loss profile is determined, an extraction phase 200 (Figure 6) is activated: drying under the climatic drying conditions corresponding to the theoretical target profile is implemented (step 201), and the system continuously measures and collects quantitative drying data throughout the drying cycle (step 202). Quantitative data can include any data measurable without operator intervention, using a sensor, analyzer, or any suitable measurement module.
[0128] These quantitative data include, in particular:
[0129] - the actual weight loss of products stored in the drying chamber (via sensors measuring condensed water for example);
[0130] - the room temperature;
[0131] - the humidity level in the room, etc.
[0132] For example, the system can be coupled with suitable sensors, devices or modules configured to continuously measure throughout the entire drying cycle:
[0133] - the weight loss of stored products, for example via the measurement of water condensation present in the storage room; and
[0134] - so-called "climatic" parameters (or climatic variables), that is to say those relating to the environment in the drying chamber which have an impact on the drying of the stored products, such as the temperature of the chamber, the humidity of the chamber, the ventilation speed, etc.
[0135] From this quantitative data, the system establishes the actual weight loss profile in real time (step 203). In Figures 1 and 2, the curve Ci illustrates an example of the actual evolution of weight loss of stored products during the first days of drying.
[0136] The evolution of water loss via the measurement of condensed water is illustrated in Figure 3: the ordinate axis is relative to the volume of condensed water V (Liter), and the abscissa axis is relative to time t.
[0137] These quantitative data are compared to the predicted target profile. In particular, the actual weight loss can be compared to the predicted target profile. The system can also establish a predictive weight loss profile (step 204), for the remaining drying cycle time, based on the current quantitative data. The detection of a predefined discrepancy between the actual (current) or predicted (end-of-cycle) weight loss and the theoretical curve triggers the adjustment of all or part of the climatic parameters (step 205 with a return to step 201).
[0138] In practice, as illustrated in Figure 4, high tolerance curves (TH) and low tolerance curves (TB) are defined. When the deviation falls outside these tolerances, the system adjusts the climatic parameters.
[0139] For example, if a drying delay compared to the theoretical drying time is identified, or a drying advance compared to the theoretical drying time, the system is configured to automatically adjust the climatic parameters (ventilation, humidity, temperature) in order to maintain the actual weight loss profile close to the theoretical weight loss profile. This automatic adjustment results in the implementation of one or more corrective actions.
[0140] In practice, the predictive control law applied by the system to adjust climatic variables when a deviation in weight loss is detected can be given by:
[0141] AP(t) Preel ( - Pcible (
[0142] ç t
[0143] u(t) = Kp.àPttl + Ki AP (r)dT
[0144]
[0145] Jo
[0146] Or
[0147] Pcibie(t) is the projected target weight loss;
[0148] Préei(t) is the actual weight loss;
[0149] AP(t) is the difference between the actual weight loss P r ee(t) and the expected weight P c ibie(t); u(t) is the corrective action applied by the system;
[0150] K p is a proportional gain, predefined to adjust for the instantaneous gap; and
[0151] Ki is an integral gain, predefined to correct for the accumulation of the gap over time.
[0152] This approach ensures that the system remains close to the target trajectory by reacting quickly and appropriately to detected variations. Based on the detected AP(t) deviation, the system adjusts its drying profile prediction for the remainder of the cycle.
[0153] In practice, these adjustments can also be influenced by the operator's observations of the product's evolution (color, firmness, crusting), which enrich the product modeling. Thus, the adjustment of climatic parameters or conditions to be applied can take into account the operator's perception.
[0154] Step 3: Integrating the operator's feedback
[0155] The system can include a user interface that allows the operator to adjust sensitivity sliders to refine the system's prediction. These sliders allow for the integration of qualitative data related to the product during the drying process.
[0156] For example, for sausage drying, such quantitative data might include:
[0157] - crusting: if the surface of the product dries out too quickly, the humidity or ventilation should be adjusted;
[0158] - Firmness: allows monitoring of the evolution of the product's internal texture. A texture that is too firm indicates drying too quickly, implying the need to adjust the humidity and temperature of the chamber.
[0159] In practice, the weight loss profile taking into account qualitative data can be defined as follows:
[0160] ^
[0161]
[0162] adjusted (0 ^theoretical (0 d” f felt (Xi)
[0163] Or
[0164] Paadjusté(t) is the adjusted prediction (or profile) of weight loss at time t, taking into account qualitative data;
[0165] Ptheoretical(t) is the initial prediction (or profile) of weight loss; and
[0166] Fressentie(ri) is a corrective function dependent on the perception n relative to a qualitative data i. This solution allows the visual and olfactory observations of the operator to be integrated into predictive adjustments, thus increasing the reliability and accuracy of the system to maintain product quality.
[0167] In practice, a weighting is assigned to each qualitative data point, this weighting being dynamically adjusted during drying, and contributes to the adjustment of climatic conditions.
[0168] An example of weighting rules to incorporate the operator's feelings is described below, in the case of sausage drying.
[0169] Integrating operator feedback into the control system allows qualitative data (such as firmness, crusting or color) to be combined with quantitative data (temperature, humidity, weight loss), in order to maximize product quality and allow for automatic correction in case of deviations.
[0170] To ensure optimal corrective actions, weighting rules are defined for the different sensory factors.
[0171] A sensory factor corresponds to a qualitative piece of data relating to the type of products to be dried. The value of the sensory factor represents the operator's perception or observation of the products during the drying process, after visual and / or olfactory evaluation.
[0172] Sensory factors or qualitative data that may be taken into account:
[0173] The main sensory or perceptual factors that the operator can adjust include, for example:
[0174] - Crusting: is an indicator of the rate of formation of the dry outer layer on the product.
[0175] - Firmness: is an indicator of the internal texture of the product (soft or firm).
[0176] - Colour: is a visual indicator of the proper progress of maturation.
[0177] - Casing adhesion: is an assessment of the connection between the skin and the meat of the product.
[0178] - Odor: indicates a good fermentation process or contamination. - Product shape and uniformity: indicates any deformation of the product during drying.
[0179] Rules for weighing feelings:
[0180] Each sensory factor (or perception) has a different influence on the quality of the product, and its weighting is adjusted according to its relative importance to the proper progress of the drying process.
[0181] As an example, weightings initially defined for each of the
[0182] Sensory factors can include:
[0183] - Crusting: 30%
[0184] - Firmness: 25%
[0185] - Color: 20%
[0186] - Tire adhesion: 15%
[0187] - Odor: 5%
[0188] - Shape and uniformity: 5%
[0189] These weightings are defined in particular according to their direct impact on product quality. For example, poor crusting can quickly affect the entire drying process by preventing uniform dehydration of the product, which justifies its high weighting.
[0190] Adaptive weighting rule based on deviations and previous cycles:
[0191] The goal is to allow the system to adjust these weightings during the drying cycle based on the observed differences between the actual weight loss curve and the target curve. This adaptive mechanism allows for greater emphasis to be placed on certain factors at critical points in the cycle.
[0192] The weighting evolves dynamically based on observed differences and operator feedback.
[0193] Adaptive weighting algorithm:
[0194] Wt (t) = w,-o + a ■ AP(t) + P ■ r t Or
[0195] Wi(t) is the weighting of factor i (crusting, etc.) at time t;
[0196] Wio is the initial weighting of factor i (e.g., 30% for crusting);
[0197] AP(t) is the difference between the actual weight loss and the target weight loss at time tn, and is the operator's perception regarding factor i.
[0198] a and P are weighting coefficients adjusting the influence of discrepancies and operator perceptions.
[0199] Example of a weighted adjustment:
[0200] If, after 10 days of drying, the difference in weight loss is significant (e.g. AP(10) > 5%), the system can increase the weighting of feelings related to crusting and firmness, as these factors are critical to understanding the causes of delay or advancement in the process.
[0201] The new weighting of crusting could increase to 40% if the operator indicates that the product is too dry on the surface, and that of firmness to 30%.
[0202] Step 4: Implementation of automatic corrective actions
[0203] Once the weightings are adjusted, the system proposes or implements automatic corrective actions based on the most influential feedback and measured discrepancies. These actions are tailored to correct the problems identified by the operator.
[0204] Rule for activating corrective actions:
[0205] For each feeling n, a corrective action is triggered if:
[0206] if Wi(t) > threshold
[0207] Actiorii
[0208]
[0209] if W(t) > threshold
[0210] Or
[0211] seuili is a predefined weighting threshold for each factor i.
[0212] Thus, if the adjusted weighting Wi(t) exceeds this threshold, the corrective action associated with this feeling is automatically triggered.
[0213] Examples of automatic corrective actions:
[0214] - Crusting too quickly: the weighting of crusting is adjusted to 40% (w cr (outage = 40%). Corrective action: increase humidity by 5% and reduce ventilation by 15% to slow down crust formation.
[0215] - Insufficient firmness: the firmness weighting is adjusted to 30% (wf e (rmete=30%). Corrective action: increase ventilation by 10% to accelerate internal drying.
[0216] - Color too pale: the relative color weighting is adjusted to 25% (w CO (light=25%). Corrective action: increase the temperature by 2°C to promote more even ripening and accelerate color development.
[0217] Concrete examples of adjustments based on observed differences and feelings, in the case of sausage drying, will be described below.
[0218] Case 1: Significant delay in the actual weight loss profile - Pessimistic prediction Scenario:
[0219] The drying cycle of a 10 kg sausage begins.
[0220] The target curve predicts a 20% weight loss after 10 days.
[0221] However, after 5 days (t = 5), the actual weight loss is only 7%, whereas the theoretical curve predicted 10%. This represents a delay in weight loss. The system can signal this delay to the operator.
[0222] The operator indicates that the product is still too soft and the color remains too pale. The system translates the operator's feedback into parameters to be adjusted, for example, adjusting the ventilation because the product is too soft to make up for the weight loss deficit.
[0223] System action or digital twin:
[0224] The system detects a delay in weight loss.
[0225] The system incorporates the operator's perception, adjusts the prediction to anticipate a possible persistent delay.
[0226] The system proposes to increase ventilation and reduce humidity to accelerate evaporation.
[0227] - Discrepancy detected in weight loss: AP(5)= -0.3 kg - Adjusted prediction: The model now predicts a final weight loss of 28% instead of 30%, at the end of the drying cycle, unless adjustments are made.
[0228] - Correction proposed or implemented by the system: increase ventilation by 15% and reduce humidity from 85% to 80%.
[0229] Expected result:
[0230] This adjustment allows us to gradually catch up.
[0231] After 10 days, the actual weight loss returned in line with the adjusted prediction, and the system continues to refine its calculations based on changes in measurements and operator feedback.
[0232] Case 2: Observed advance of the actual weight loss profile - Optimistic prediction
[0233] Scenario:
[0234] Another cycle starts with a target of 30% weight loss for a similar product.
[0235] After 10 days (t = 10), the system detects that the weight loss is 22%, whereas the target curve predicted 20%.
[0236] The operator confirms that the firmness is good and that the product has taken on an appropriate color.
[0237] System action or digital twin:
[0238] The system detects a lead over the target curve and readjusts the prediction to forecast potentially faster weight loss than expected.
[0239] By incorporating positive feedback from the operator (satisfactory firmness and color), the system proposes to maintain the current pace with a slight reduction in ventilation to avoid excessive drying.
[0240] - Discrepancy detected in weight loss: AP(10) = +0.2 kg
[0241] - Adjusted prediction: the final weight loss is readjusted to be achieved in 18 days instead of 20 days
[0242] - Correction proposed or implemented by the system: maintain ventilation but reduce the temperature by 2°C to slightly slow down the process and stabilize quality. Expected result:
[0243] The product maintains good quality, and the weight loss curve follows the readjusted model.
[0244] The objective has been achieved with a confirmed lead time, allowing the cycle to be completed two days earlier, with energy savings.
[0245] Case 3: Advance observed, then accentuated - Adjusted prediction
[0246] Scenario:
[0247] For a 15 kg product with a target weight loss of 35%, the system predicts a loss of 10% after 5 days.
[0248] However, the actual weight loss already reaches 12%, and the operator notes that the crusting is too rapid and the product seems too firm.
[0249] System action or digital twin:
[0250] The system detects a significant advance in weight loss.
[0251] Taking into account the operator's perception (rapid crusting, excessive firmness), the system adjusts its prediction to anticipate an imbalance in product quality if the drying process continues at this rate.
[0252] The system proposes or implements an increase in humidity and a reduction in ventilation to prevent excessive surface drying.
[0253] - Initial deviation detected: AP(5)=+0.3 kg
[0254] - Adjusted prediction: the model predicts final weight loss achieved in just 16 days.
[0255] - Proposed correction: increase the room humidity from 80% to 85% and reduce ventilation by 20%.
[0256] Expected result:
[0257] By adjusting environmental or climatic parameters, the acceleration of the process is tempered, making it possible to avoid excessive drying and to maintain the expected quality of the product.
[0258] The cycle concludes according to the revised forecasts, ensuring a well-balanced product. These three cases illustrate how the adjustment system reacts dynamically to observed discrepancies between the theoretical curve and actual data, while incorporating operator feedback to refine its predictions and adjustments. This system prevents excessive delays or advances in drying, while guaranteeing that the final product meets quality criteria.
[0259] Step 5: Integrated Energy Optimization
[0260] In addition to the dynamic adjustment of drying parameters, the process may also include the optimization of energy consumption related to drying.
[0261] Energy consumption is measured in real time for each phase of the cycle, including both electrical (ventilation, compressors) and thermal (heating and cooling) consumption.
[0262] The system calculates the energy efficiency ratio (kg of weight loss per kWh consumed) and automatically stops the drying cycle when this ratio becomes too low.
[0263] Energy efficiency law:
[0264] The energy efficiency R(t) is defined by:
[0265] D I -.\ _ Pperte^fy PperteÇ^ At)~
[0266]
[0267] Etotal(t) ~ E tota i(t — ht)
[0268] Or
[0269] R(t) is the moving average energy efficiency at time t,
[0270] Pperte(t) is the cumulative weight loss of the product at time t,
[0271] Etotai(t) is the total cumulative energy consumption (electrical + thermal) at time t.
[0272] At is the predefined sliding time interval over which R(t) is averaged (example 1 day).
[0273] Automatic cycle stop condition:
[0274] The system can interrupt the drying cycle when:
[0275] R(t) < Threshold
[0276] Rseuii represents a minimum acceptable energy efficiency ratio. This model guarantees efficient energy management, allowing for cost optimization while maintaining product quality.
[0277] Step 6: Learning and complete automation
[0278] The system can use historical data to improve its predictions over time. Through machine learning, future drying cycles are optimized by dynamically adjusting dryer parameters. Automation evolves toward autonomous management where the operator intervenes only in case of anomalies.
[0279] As the system or digital twin collects data on feelings and corrective actions taken, a machine learning module can dynamically adjust the weightings and activation thresholds based on the results of previous cycles. The system optimizes its performance over time through a learning loop.
[0280] In addition, the system can automatically adjust the weighting coefficients a and P to modulate the influence of deviations and feelings, depending on the successes or failures of previous corrective actions.
[0281] Thanks to this learning loop, the system becomes able to predict more accurately which correction will be most effective depending on the specific conditions of the product and the cycle.
[0282] The adaptive sensory weighting system prioritizes critical sensory factors while automatically adjusting the drying process to maintain product quality. Integrating these sensory feedback, combined with measured deviations from the target curve, enables personalized control and automatic activation of corrective actions. Through machine learning, the system continuously refines the weightings and corrections, making the drying process increasingly efficient with each cycle.
[0283] Example of an operator interface. An example of a user interface 1 adapted to the delicatessen sector is illustrated in Figure 5. This user interface 1 allows the operator to enter parameters enabling the system to determine the theoretical or predicted target weight loss profile 10. This interface allows the operator, in particular, to define:
[0284] - the maximum capacity of chamber 11 (in kg);
[0285] - the type of product 12 (e.g., sausage) which determines the drying behavior according to the composition of the product to be dried (moisture, fat);
[0286] - the duration of cycle 13 (in hours);
[0287] - the chamber load at the start of cycle 14 (in kg);
[0288] - target weight loss 15 (in %);
[0289] - the high and low tolerances 16 (in %);
[0290] - the threshold of the energy ratio 17 (in Wh / kg).
[0291] Of course, the system can automatically determine the values of certain parameters based on previous drying cycles, while still allowing the operator to make adjustments. For example, when the operator enters
[0292] "Sausage" as product type 12, the system can offer the operator values concerning the target weight loss 15, the tolerances 16, the energy ratio threshold 17.
[0293] User interface 1 can also display:
[0294] - time-related information about drying such as start of cycle 18, end of cycle 19, time elapsed 20, time remaining 21, etc.
[0295] - information relating to the drying process such as current load rate 22, current weight loss 23, current difference 24 between target weight loss and actual weight loss, current energy ratio 25, predicted weight loss at the end of the cycle 26, prediction of the difference at the end of the cycle 27, prediction of the ratio at the end of the cycle 28.
[0296] The user interface 1 can also allow the operator to evaluate qualitative data, or to indicate their feelings 30 concerning, for example, homogeneity 31, blooming 32, crusting 33, firmness 34, color 35, adherence to the casing 36, odor 37, or variability in weight loss 38. On the interface shown in Figure 5, the feelings are captured by adjusting the sliders, the system translating the position of each slider into a weighting value.
[0297] Variant: *** DRAINAGE ***
[0298] The process of dynamically adjusting the control parameters for drying can be transposed to the draining process, for example to the draining of dairy products in which the weight loss results from the operation of separating the curd from the whey, for example by pressing.
[0299] Thus, in addition to the climatic data mentioned above for the drying process, such as the ambient temperature of the dripping chamber, the ambient relative humidity, and the temperature of the cold loop, the adjustment process for dripping may take into account additional data such as:
[0300] - measurements relating to whey flow, namely hourly flow (volume or flow rate per hour), and / or cumulative whey volume.
[0301] - measurements relating to the pH of products, determined continuously, directly on a product (for example, at the core of the dairy product or in a representative solution).
[0302] Thus, following the principle of the adjustment process applied to drying described above, in the case of a draining cycle, the dynamic adjustment process for the control parameters of the draining of products stored in a draining chamber may include, in particular:
[0303] During the initial phase 100 (figure 6):
[0304] . the determination of the target weight loss profile (step 101) for the product to be drained; . the determination of the target acidity profile;
[0305] . the determination of the climatic drainage conditions (step 102) throughout the drainage cycle meeting the target weight loss profile and the target acidity profile.
[0306] The target acidity profile defines the theoretical evolution of the product's pH level during the draining cycle, with pH being an indicator of acidification and product maturation.
[0307] During the 200 draining phase and throughout the draining cycle:
[0308] . the application of the initial climatic conditions corrected for any adjustments determined in step 205 (step 201).
[0309] . real-time measurement of quantitative data (step 202) including pH measurement, and real-time establishment of the actual weight loss profile (step 203) and the actual acidity profile; and
[0310] . the determination of a predictive weight loss profile (step 204) and a predictive acidity profile for the remaining cycle duration if the initial climatic conditions are maintained;
[0311] - the automatic determination of adjustments to be applied to the chamber's climatic conditions (step 205) if a predefined deviation between the target profiles and the actual profiles is observed, or if a predefined deviation between the target profiles and the predictive profiles is observed; and
[0312] - the application of these adjusted climatic conditions (return to step 201).
[0313] The automatic adjustment of the climatic conditions to be applied in the chamber can also take into account the qualitative data entered by the operator.
[0314] Determination of the target weight loss kinetics
[0315] Before each draining cycle begins, the system determines a theoretical or predicted target weight loss profile, for example, in the form of a predicted target weight loss curve. This target profile is established based on the specific characteristics of the product to be drained. For example, in the case of dairy products to be drained, the target profile can be established by taking into account:
[0316] - the type of product to be made and the associated recipe;
[0317] - of the initial weight of the product to be drained;
[0318] - the duration of a draining cycle;
[0319] - the weight loss target, for example in the form of a percentage relative to the initial weight.
[0320] The weight loss profile in the case of draining corresponds to an hourly flow profile (cumulative volume or hourly flow rate) of whey. The same model as that used for drying can be applied.
[0321] P(t) = P cy
[0322]
[0323] \the T
[0324] Or
[0325] P(t) is the weight loss at time t, this weight loss is relative to the cumulative flow rate or volume of whey;
[0326] P cy is the weight loss of the product at the end of the draining cycle;
[0327] t C y is the duration of the draining; and
[0328] T is a time constant related to the properties of the product and the extraction conditions.
[0329] Determination of acidity kinetics
[0330] A model for monitoring acidification could be:
[0331] p
[0332]
[0333] target_pH(t) = initial_pH − ΔpH_max × (1 − e^(−t / τ_pH))
[0334] Or
[0335] pH c ibie(t) is the pH at time t;
[0336] pHinitiai is the initial pH value at the start of the dripping process (e.g. 6.5); ApHmax is the maximum expected pH variation during the theoretical infinite duration cycle (t — +°°);
[0337] T P H is a time constant (e.g. 10), influencing the rate of acidification, and is relative to the properties of the product to be drained (e.g. initial moisture, fat) and the draining conditions.
[0338] At the beginning of the process, the pH decreases rapidly, reaching approximately 60% of its maximum variation after a time equivalent to TP H.
[0339] Determination of climatic drainage conditions
[0340] The system determines the evolution of the dewatering climatic conditions to be implemented during the dewatering cycle to achieve the target weight loss and acidity profiles, as well as the corresponding instructions or controls for the devices or equipment in or connected to the chamber that impact these climatic conditions. These climatic conditions include, for example: temperature, humidity, ventilation speed, etc.
[0341] Real-time measurement of quantitative data and continuous adjustment of climatic conditions
[0342] Once the theoretical target profiles for weight loss and acidity are determined, drainage under the climatic conditions corresponding to the theoretical target profiles is implemented, and the system continuously measures and collects quantitative data related to the drainage throughout the drainage cycle.
[0343] Quantitative data can include any data that can be measured without operator intervention, using a sensor, analyzer, or any suitable measurement module.
[0344] These quantitative data include, in particular:
[0345] - the actual weight loss of the products stored in the chamber, via the measurement of the cumulative whey flow volume or the whey flow rate; - the chamber climatic conditions, such as temperature, humidity level in the chamber, etc.
[0346] - the pH of the products, via for example a fixed pH sensor installed in the chamber.
[0347] Based on these measured data, an actual profile and a predictive profile for weight loss and acidity are estimated. For example:
[0348] - if ventilation is increased, the model projects an acceleration of weight loss in the next few hours.
[0349] - if humidity is reduced, the model anticipates a stabilization or a slowing down of acidification.
[0350] Example of a predictive profile of weight loss adjusted for current climatic parameters:
[0351] Projected(.t) Prerequisite "h ^P adjustment^)
[0352] ^
[0353]
[0354] P adjustment^) = k ± x ventilation + k2x humidity + k3x temperature
[0355] Or
[0356] Projected(t) is the predicted weight loss if the predicted climatic conditions are maintained;
[0357] Préei is the actual weight loss measured at time t;
[0358] ventilation is the measured airflow resulting from the ventilation of the room; humidity is the humidity level measured in the room;
[0359] temperature is the measured temperature of the room;
[0360] ki, k2 and ks are coefficients that can be adjusted according to the characteristics of the product and historical responses.
[0361] Example of a predictive profile of acidity adjusted by environmental parameters:
[0362] projected pH(t) actual pH "h pH ajus t emen t(t)
[0363] PH a precisely (t) = k4x temperature + k5x humidity
[0364] Or
[0365] pHprojected(t) is the predicted pH if the predicted climatic conditions are maintained;
[0366] pHrei is the actual pH measured at time t;
[0367] temperature is the measured temperature of the room;
[0368] humidity is the humidity level measured in the room;
[0369] K4, and ks are coefficients that can be adjusted according to the characteristics of the product and historical responses.
[0370] Actual or predicted profiles are compared to target profiles. Specifically, the detection of a predefined discrepancy between actual or predicted weight loss and target weight loss triggers the adjustment of all or some of the climatic parameters. Similarly, the detection of a predefined discrepancy between actual or predicted pH and target pH triggers the adjustment of all or some of the climatic parameters.
[0371] In practice, for each target profile, upper and lower tolerance curves are defined. When the deviation falls outside these tolerances, the system adjusts the climatic parameters.
[0372] As an example, threshold levels can be defined as follows:
[0373] Thresholds for flow (weight loss): - Lag: Negative deviation > 5%.
[0374] - Very late: Negative gap > 10%.
[0375] - Ahead: Positive difference > 5%.
[0376] - Very far ahead: Positive gap > 10%.
[0377] pH thresholds:
[0378] - Late: Deviation of -0.2 units from the target curve.
[0379] - Very late: Difference of -0.4 units.
[0380] - Ahead: Difference of +0.2 units.
[0381] - Very far ahead: Difference of +0.4 units.
[0382] In practice, these thresholds are adjustable by the operator. For example, for a more delicate cheese, the thresholds can be reduced to 3% and 7% for runoff. These thresholds can also be adjusted dynamically based on a history of actual deviations observed during previous cycles.
[0383] Adjusting climate parameters includes identifying one or more corrective actions to be implemented manually by the operator or automatically by the system.
[0384] Corrective actions
[0385] In practice, each deviation is linked to probable causes. For example, excessively rapid weight loss could be due to over-ventilation, while a pH that is too low could indicate an unsuitable temperature. The corrective actions proposed by the system must therefore be consistent, with a clear indication of priorities.
[0386] For example, when a discrepancy between target and actual profiles is detected, the system can generate an alert message for the operator, suggesting manual corrective actions to be taken. For example:
[0387] - for weight loss:
[0388] When a deviation corresponding to a delay in weight loss is detected, the system can inform the operator with the following message: "Delay detected in serum flow. Check the drainage channels and consider slightly increasing ventilation." When a deviation corresponding to an advance in weight loss is detected, the system can inform the operator with the following message: "Serum loss too rapid. Reduce aeration to slow the flow and prevent excessive dripping."
[0389] - for pH:
[0390] When a deviation corresponding to a pH that is too low is detected, the system can inform the operator via the following message: "pH below the target curve. Check the temperature and humidity to stabilize the fermentation."
[0391] When a deviation corresponding to a pH that is too high is detected, the system can inform the operator via the following message: "pH above target. Ensure that environmental conditions favor acidification."
[0392] In practice, weight loss and pH are interdependent parameters. Too rapid a flow can influence the pH (slowing acidification), and a delay in flow can be accompanied by excessive acidification.
[0393] Therefore, the system of the invention can identify priorities in corrective actions. For example:
[0394] - If the flow is delayed and the pH is too low: the system may suggest first correcting the flow to avoid prolonged stagnation, but maintaining controlled humidity so as not to worsen acidification.
[0395] - If the flow is too fast and the pH too high: the system may suggest prioritizing the reduction of ventilation to slow the flow, while slightly increasing the temperature to stimulate acidification.
[0396] These corrective actions can be performed manually by the operator, but the system can be configured to perform these corrective actions automatically.
[0397] Another example of actions to take when a discrepancy is detected:
[0398] Example 1: Simple case (flow or pH drift)
[0399] Detection: Delay in serum loss detected.
[0400] Suggested action: Slightly increase ventilation and check the condition of the drainage channels. The pH is within target range; no action is required.
[0401] Example 2: Complex case, two drifts detected (flow and pH aligned). Detection 1: Excessive serum loss. Proposed action: Reduce ventilation and maintain humidity to slow the flow.
[0402] Detection 2: pH above target.
[0403] Proposed action: Slightly increase the temperature to stimulate acidification, and verify that the adjustment does not disrupt the balance of the process.
[0404] Example 3: Contradictory case (divergent flow and pH)
[0405] Detection: Two deviations detected with possible conflicting causes:
[0406] Delayed serum loss: the system proposes increasing ventilation and maintaining humidity to accelerate flow;
[0407] . pH too high: the system suggests slightly increasing the temperature to stabilize the acidification.
[0408] The system recommends a priority in the corrective actions to be taken: First correct the flow if the product appears visually too wet.
[0409] Other examples in which prioritization takes the threshold level into account:
[0410] Example 1: A single (moderate) drift
[0411] Message: "Serum loss is delayed (-6%). Slightly increase ventilation to compensate for the loss."
[0412] Example 2: Several (serious) abuses
[0413] Message: "Two deviations detected:
[0414] Very delayed serum loss (-12%): High priority. Significantly increase ventilation and check drainage channels.
[0415] pH significantly elevated (+0.5 units): Reduce humidity to stabilize acidification. Example 3: Conflict between drifts
[0416] Message: "Two critical deviations with conflicting actions:
[0417] Very delayed serum loss (-10%): Increase ventilation.
[0418] . pH very low (-0.4 unit): Reduce ventilation and humidity.
[0419] Recommended priority: Stabilize the pH first if the visual condition of the product appears correct.
[0420] Integration of operator feedback: The system can integrate a user interface that allows the operator to enter qualitative data relating to the product being drained to refine the system's prediction. This qualitative data can relate to the visual appearance (clarity or opacity of the serum), the texture of the product (too soft, too firm), or even the color (an indicator of maturation or quality).
[0421] For example, in the case of dairy product draining, such qualitative data may relate to the product being drained, for example:
[0422] - Crusting:
[0423] Visual and tactile observation to detect a crust that is too rapid or absent.
[0424] Impact: Adjustments to humidity or ventilation.
[0425] - Firmness:
[0426] . Tactile feel (texture too soft or too firm).
[0427] Impact: Adjustments to balance internal drying.
[0428] - Color:
[0429] Visual observation to identify delayed or accelerated maturation.
[0430] Impact: Changes in temperature or humidity.
[0431] For example:
[0432] Pale color: Indicates a delay in ripening or excessive humidity. / Possible actions: Slightly increase the temperature (e.g., by 2°C) to accelerate ripening. Reduce the humidity (e.g., by 3%) to limit internal condensation. Normal color (expected shade): Process is functioning correctly. / Action: No adjustments necessary.
[0433] Dark color: May indicate excessive humidity or uneven drying. / Possible actions: Check ventilation. Slightly reduce the temperature (e.g., by 1°C) to slow down excessive ripening.
[0434] Qualitative data can also relate to the whey flowing from the product. For example, qualitative data can include the visual appearance and color of the whey.
[0435] Visual aspect:
[0436] - Clarity:
[0437] Clear: Indicates normal and efficient flow. / Action: No intervention required. Cloudy: May indicate an imbalance or partial blockage in the system. / Possible actions: Check filters and flow channels. Slightly increase ventilation to accelerate flow.
[0438] Color:
[0439] - Light yellow (normal): Balanced drainage process. / Action: No intervention required.
[0440] - Dark yellow: Indicates possible serum stagnation or poor filtration. / Possible actions: Reduce humidity to limit filter saturation. Check flow in drainage channels.
[0441] - Pale to dark green: Chemical imbalance (e.g., excess acidity or contamination). / Possible actions: Check temperature and humidity conditions. Check pH for abnormal acidification.
[0442] Qualitative data can also include spot pH measurements taken by the operator during manual product sampling at key times. These spot measurements complement continuous data by providing human validation, especially when there is doubt about the sensors or for specific batches.
[0443] This qualitative data (or operator perceptions) is used to estimate predictive profiles. In particular, these perceptions directly influence the projected curves by adjusting the coefficients related to environmental conditions. The predictive model that takes these perceptions into account can be:
[0444] Projected_p(t) = Actual_p + ΔAdjustment_p(t) + perceived_f(r_i)
[0445] Or
[0446] fresenti is the correction factor specific to a qualitative data point n
[0447] As with drying, weighting rules can be defined for the different perceived effects. The climatic conditions to be applied are also adapted based on qualitative data. For example, adjustments to environmental parameters (temperature, humidity) and proposed changes in the next cycle.
[0448] Adjustment of certain constants at the next cycle To ensure continuous improvement, the system can at the end of each cycle adjust some of the coefficients used in the predictive models.
[0449] Adjustment coefficient ki
[0450] The coefficients ki, k2, ki, k4 and ks can be adjusted after each draining cycle: Zz — the cumulative
[0451] k_new = k_old
[0452]
[0453] target
[0454] Or
[0455] knouveau is the adjusted coefficient;
[0456] kancien is the coefficient of the previous cycle;
[0457] Acumui is the difference between the predicted weight loss at the end of the cycle and the actual weight loss at the end of the cycle.
[0458] The target is the target weight loss
[0459] At the end of each cycle, the system compares the projections to the actual results to refine the coefficients (k1, k2, k3, k4, and k5). For example, if the whey is cloudy
[0460] When ventilation speed is systematically increased, the ventilation-related coefficient ki in the predictive model is automatically adjusted to predict this correction. For example, if a 10% increase in ventilation results in a weight loss of +1.5% instead of the projected +2%, the ki coefficient is adjusted accordingly.
[0461] Constant adjustment T P H
[0462] The constant T P H can also be adjusted.
[0463] τ_pH^new = τ_pH^old × (1 + ΔpH_cumul / ΔpH_max)
[0464] τ_pH^ancien
[0465] pH 1 +
[0466] pH m ax
[0467] Or
[0468] τ_pH^new is the adjusted constant
[0469] τ_pH^ancien is the constant of the previous cycle
[0470]
[0471] pH CU mui is the difference between the pH at the end of the target kinetic cycle and the actual pH at the end of the cycle.
[0472] ΔpH_max is the maximum expected change in pH during a theoretical cycle of infinite duration (t → +∞); For example:
[0473] For the first cycle (cycle 1), T P H = 10, and the final pH at the end of cycle 1 is ahead by +0.4 units.
[0474] The adjustment for the next cycle (cycle 2): T P H = 9.6.
[0475] For cycle 2: T PH = 9.6, and the final pH at the end of cycle 2 is lagged by -0.2 units. The adjustment for the next cycle (cycle 3): T P H = 9.8.
[0476] Time delay integration
[0477] The system can incorporate a time-delay period. During this period, no deviation calculations or alerts are generated or triggered. The measured values (weight loss, pH) are recorded but are not compared to the target curves. The duration of this time-delay period can be, for example, 60 minutes from the start of the cycle. The operator may be able to adjust this duration.
[0478] This delay is particularly useful to avoid generating alerts when the system takes time to reach its optimal conditions after launch (e.g., stabilized humidity, regulated ventilation).
[0479] Determining a confidence index
[0480] A confidence index based on historical cycles helps to improve the reliability of the system's recommendations. This can help operators prioritize adjustments that are most likely to resolve discrepancies without creating conflicts.
[0481] The confidence index is a score (e.g., from 0 to 100%) assigned to each proposed adjustment, based on:
[0482] - Historical effectiveness: Frequency of similar adjustments that have effectively corrected the gaps.
[0483] - Consistency: No significant conflicts were observed with other parameters during adjustment.
[0484] The confidence index can be calculated using the following formula:
[0485] (successful cases)
[0486] confidence index = - x 100
[0487]
[0488] Total cases For example: Over 10 cycles, a temperature increase of 2°C corrected a pH deviation in 8 cases without causing a conflict with weight loss.
[0489] 8
[0490] Confidence index = (8 / 10) × 100 = 80%
[0491]
[0492] Integrated energy optimization
[0493] In addition to dynamically adjusting the draining parameters, the process can also include optimizing energy consumption related to draining. Energy consumption is measured in real time for each phase of the cycle, including both the electrical consumption of the ventilation and the thermal consumption (heating and cooling) of the draining cell. The energy efficiency ratio (energy consumed per volume of whey drained) is also calculated, and the system can take corrective action, or even automatically stop the draining cycle, when energy efficiency becomes too low.
[0494] As an example, the energy consumed by ventilation (E ven (tiiation) can be determined using the following formula:
[0495] E_ventilation = Q_air × C_energy
[0496] With
[0497] Qair, the ventilated air flow rate (corresponds to the volume of air circulated per hour (m³) 3 / h));
[0498] Energy: kWh consumed per 1m³ 3 ventilated air.
[0499] Energy efficiency law:
[0500] The energy efficiency R(t) is defined by:
[0501] R(t) = (V_flow(t) - V_flow(t-Δt)) / (E_total(t) - E_total(t-Δt))
[0502]
[0503] E_total(t) - E_total(t-Δt)
[0504] Or
[0505] R(t) is the moving average energy efficiency at time t,
[0506] Vécouié(t) is the cumulative volume of whey discharged at time t, and corresponds to the cumulative weight loss of the product at time t.
[0507] Etotai(t) is the total cumulative energy consumption (electrical + thermal) at time t. At is the predefined sliding time interval over which R(t) is averaged (example 1 day).
[0508] Automatic cycle stop condition:
[0509] The system can interrupt the drying cycle when:
[0510] R(t) < Threshold
[0511] Rseuii represents a minimum acceptable energy efficiency ratio.
[0512] This model guarantees efficient energy management, allowing for cost optimization while maintaining product quality.
Claims
1. CLAIMS 1. A method for dynamically adjusting the control parameters of the extraction of products stored in an extraction chamber, throughout the different extraction cycles, the extraction including drying or draining, the dynamic adjustment method comprising: 3.- an initial phase (100) during which the following actions are carried out:
4. Determination of a target weight loss profile (101) based on at least the product type, the initial product weight, the duration of the extraction cycle, and the target product weight loss at the end of the cycle, the target weight loss profile defining the theoretical evolution of product weight loss during the extraction cycle; determination of the extraction climatic conditions (102) to be applied in the chamber throughout the extraction cycle, corresponding to the theoretical evolution of weight loss; - an extraction phase (200) following the target weight loss profile determined in the initial phase, during which the following actions are carried out: 5.a / application in the chamber of climatic conditions (201) corresponding to the target weight loss profile; 6.b / real-time measurement throughout the extraction cycle of quantitative data (202) relating to the extraction and including at least the actual weight loss, the temperature and humidity level of the chamber, the ventilation speed; 7.c / real-time establishment of the actual weight loss profile (203); 8.d / real-time establishment of a predictive weight loss profile (204) based on current quantitative data; 9.e / automatic adjustment of climatic conditions (205) to be applied in the chamber if a predefined deviation between the target profile and the actual profile is observed, or if a predefined deviation between the target profile and the predictive profile is observed, and application of these adjusted climatic conditions for the duration of the remaining cycle; 10.f / implementation of steps b / to e / until the end of the extraction cycle.
2. Dynamic adjustment method according to claim 1, wherein the automatic adjustment step e / comprises the application of one or more corrective actions including the adjustment of device setpoints impacting the temperature in the chamber, the ventilation speed, the humidity in the chamber, the cycle duration, etc.
3. A dynamic adjustment method according to claim 1 or 2, wherein the model representing the weight loss over time is defined by the following general equation: 12.1 — e T 13.tëj 15.
16. 1 — e~~ 17. Where 18.P(t) is the weight loss at time t; 19.P cy is the weight loss of the product at the end of the extraction cycle; 20.tcy is the extraction cycle time; and 21.T is a time constant related to the properties of the product and the extraction conditions.
4. Dynamic adjustment method according to any one of claims 1 to 3, wherein the model representing the target weight loss profile over time is defined by a polynomial, logarithmic, or multi-exponential type model.
5. Dynamic adjustment method according to any one of claims 1 to 4, wherein the quantitative data further include the acidity of the product, via continuous measurement of the pH in the product.
6. A dynamic adjustment method according to claim 5, wherein the extraction cycle is a draining cycle, and: 25.- the initial phase further includes the determination of a target acidity profile, the climatic conditions to be applied being determined to meet both the target weight loss profile and the target acidity profile; 26. The extraction phase further includes real-time pH measurement and real-time establishment of the actual acidity profile, as well as the establishment of a predictive acidity profile if the applied climatic conditions are maintained; automatic adjustment of climatic conditions is carried out if a predefined deviation between the target acidity profile and the actual acidity profile is observed, or if a predefined deviation between the target acidity profile and the predictive acidity profile is observed, the adjustment of climatic conditions responding to both the target weight loss profile and the target acidity profile; and 27.- application of these adjusted climatic conditions.
7. A dynamic adjustment method according to claim 1, wherein the logarithmic model for tracking acidification is defined by: 29.p 30.
31. target_pH(t) = initial_pH − ΔpH_max × (1 − e^(−t / τ_pH)) 32.where 33.pH c ibi e (t) is the pH at time t; 34.pHinitiai is the initial pH value at the start of the draining process; 35.ΔpH max is the maximum expected variation of pH during a theoretical cycle of infinite duration (t → +∞); 36.T P H is a time constant, influencing the rate of acidification, and is relative to the properties of the product to be drained and the draining conditions.
8. Dynamic adjustment method according to any one of claims 1 to 7, wherein the step e / of automatic adjustment of climatic conditions also takes into account qualitative data relating to the operator's feelings concerning the products during fluidic extraction.
9. A dynamic adjustment method according to claim 8, wherein:
39. During the initial phase, a weighting is assigned to each of the qualitative data points, based on the quality of the product to be obtained; and 40.- during the extraction phase, all or part of the weightings are adjusted throughout the extraction cycle according to the observed difference between the actual weight loss and the target weight loss, and / or indications relating to the qualitative data entered by the operator corresponding to the operator's perception of the product for which fluid extraction is in progress; 41.- During the extraction phase, the dynamic adjustment of the climatic conditions to be applied throughout the cycle also takes into account the evolution of qualitative data.
10. Dynamic adjustment method according to claim 9, wherein the predictive profile is determined taking into account qualitative data.
11. Dynamic adjustment method according to any one of claims 1 to 10, wherein the determination of the target weight loss profile and / or the target acidity profile during the initial phase is based on weight loss profiles and / or the target acidity profile of previous cycles.
12. A dynamic adjustment method according to any one of claims 1 to 11, comprising a time-delay period during which no deviation calculation is performed, and the quantitative data measured are recorded but not compared to the target curves.
13. A dynamic adjustment method according to any one of claims 1 to 12, further comprising the optimization of energy consumption related to the extraction cycle, consisting of: 45.- to measure energy consumption in real time, including both electrical and thermal consumption; 46.- calculate the energy efficiency defined by the ratio of weight loss per kWh consumed; - depending on the difference between the calculated energy efficiency and the predefined minimum acceptable energy efficiency ratio, apply measures aimed at optimizing energy efficiency, such measures including at least one of the following actions: automatic shutdown of the extraction cycle, adjustment of ventilation.