TWI853535B - Beverage quality prediction system during blending process - Google Patents

Abstract

A beverage mixing quality prediction system is provided. The beverage mixing quality prediction system includes a container, a stirring control module, a first monitoring module, a second monitoring module and a control unit. The container is used to contain a mixed material. The stirring control module includes a stirring element, which extends into the container to stir the mixed material. The first monitoring module is arranged in a circulation pipeline outside the container to monitor the quality of the mixed material in real time. At least a part of the second monitoring module is arranged in the container to monitor the quality change of the mixed material during the mixing process. The control unit is electrically connected to the stirring control module, the first monitoring module and the second monitoring module, and is used to control the stirring control module and receive quality monitoring data from the first monitoring module and the second monitoring module, and establish at least one material mixing quality prediction model based on the quality monitoring data.

Description

Beverage Mixing Quality Prediction System

The present invention relates to a quality prediction system, and more particularly, to a beverage preparation mixing quality prediction system.

In the food industry, material mixing is an indispensable process step. The mixing types can include dispersing and dissolving solid powders or mixing immiscible liquids.

At present, food factories mostly use manual visual sampling or empirical rules to judge the end point of mixing for material mixing processes. And if the barrel is made of stainless steel, it is impossible to use the naked eye to observe the current state of the material, such as powder agglomeration or foaming degree. Therefore, the existing sampling and judgment methods are not only labor-intensive, but also may lead to misjudgment of the mixing end point due to different operators, batch operations, sampling point errors or different equipment. If the mixing endpoint is misjudged or the mixing status cannot be immediately known, the following situations may occur: (1) over-mixing, resulting in deterioration of the quality of the mixed material, which makes the material unusable and wasted; (2) incorrect ingredient ratio or unstable filler quality, resulting in poor uniformity and stability of the final product; and (3) pipeline blockage, resulting in damage to the mixing equipment.

In some embodiments, a beverage preparation mixing quality prediction system includes a container, a stirring control module, a first monitoring module, a second monitoring module and a control unit. The container is used to contain a mixed material. The stirring control module includes a stirring member, which extends into the container to stir the mixed material. The first monitoring module is arranged outside the container to monitor the quality of the mixed material in real time. At least a part of the second monitoring module is arranged in the container to monitor the quality change of the mixed material during the mixing process. The control unit is electrically connected to the stirring control module, the first monitoring module and the second monitoring module, and is used to control the stirring control module and receive quality monitoring data from the first monitoring module and the second monitoring module, and establish at least one material mixing quality prediction model based on the quality monitoring data.

In some embodiments, a beverage preparation and mixing quality prediction system includes a container, a stirring and controlling module, a quality analysis device, and a control unit. The container is used to contain a mixed material. The stirring and controlling module includes a stirring member, which extends into the container to stir the mixed material. The quality analysis device is disposed on one side of the container to analyze the quality of the mixed material taken out of the container. The control unit is electrically connected to the stirring and controlling module and the quality analysis device to control the stirring and controlling module and receive quality analysis data from the quality analysis device, and establish at least one material mixing quality prediction model based on the quality analysis data.

Referring to FIG. 1 , it is a schematic diagram showing a beverage preparation and mixing quality prediction system 1 according to some embodiments of the present invention. The beverage preparation and mixing quality prediction system 1 of the present invention comprises a container 100, a stirring control module 200, a circulation pump 300, a circulation pipeline 400, a quality analysis device 500, a first monitoring module 600, a second monitoring module 700, a material preparation container 810, a material feeding control module A, an input/output module B, and a control unit C.

The container 100 is used to contain a mixed material M. In some embodiments, the container 100 may be a vertical cylindrical container, and it is not limited to a closed container, and may also include an open container. The container 100 may be a transparent material or an opaque material. In some embodiments, as shown in FIG. 1 , the container 100 may include an upper portion 101, a lower portion 102, and a discharge valve 103. The lower portion 102 is opposite to the upper portion 101. The discharge valve 103 is disposed at the lower portion 102 to discharge the mixed material M.

The stirring control module 200 may include a stirring member 230, a driving motor 210 and a torque sensing element 220. The stirring member 230 may extend into the container 100 to stir the mixed material M. In some embodiments, the stirring member 230 may be replaced as needed, and may be visually material or product characteristics and is not limited to a single type. For example, in some embodiments, the stirring member 230 may be a blade type, including but not limited to a propeller blade, a paddle blade or a turbine blade. In some embodiments, the stirring member 230 may be a crushing shear type, including but not limited to a toothed, grooved or round hole type tool. The driving motor 210 is used to drive the stirring member 230. In some embodiments, as shown in FIG. 1 , the drive motor 210 may be connected to the stirring member 230 via the torque sensing element 220 . The torque sensing element 220 may be used to monitor the torque change during the stirring process, and then convert the viscosity value of the mixed material M. In some embodiments, the drive motor 210 may be equipped with a frequency converter to adjust the rotation speed of the drive motor 210 .

The circulation pipeline 400 can be connected to the upper part 101 and the lower part 102 of the container 100 to guide the mixed material M in and out of the container 100. The circulation pump 300 can be arranged on the circulation pipeline 400 to circulate and transport the mixed material M in and out of the container 100. In some embodiments, the circulation pipeline 400 can be provided with a monitoring device for monitoring operation or a shearing device for homogenization operation. In some embodiments, the circulation pipeline 400 can be made of stainless steel or plastic. In some embodiments, the circulation pump 300 can include but is not limited to a diaphragm pump, a centrifugal pump, a rotor pump, a peristaltic pump, an electromagnetic pump or a shear pump. Among them, the shear pump can provide shear force to assist in the homogenization operation of the mixed material M.

The quality analysis device 500 may be disposed on one side of the container 100 to analyze the quality of the mixed material M taken out of the container 100. In some embodiments, the quality analysis device 500 may be an off-line (non-online) analysis device that detects quality by sampling, such as a laboratory analysis device. The quality analysis device 500 may not be limited to a specific device, and any device that can characterize the quality of the mixed material M may be used. In some embodiments, the quality analysis device 500 may include but is not limited to one or more of the following: a fluid rheology analysis device 510, a particle size analysis device 520, and an electrokinetic potential analysis device 530.

The first monitoring module 600 can be disposed on the circulation pipeline outside the container 100 to monitor the quality of the mixed material M in real time. In other words, the first monitoring module 600 is an online monitoring module. In some embodiments, as shown in FIG1 , the first monitoring module 600 can be disposed on the circulation pipeline 400. The first monitoring module 600 can select an appropriate monitoring element according to different mixing modes of the mixed material M (including, for example, dissolution, dispersion or emulsification and homogenization), and the number of monitoring elements used can be one or more. In some embodiments, the first monitoring module 600 may include but is not limited to one or more of the following: a viscosity sensor 610, a particle size sensor 620, a total suspended matter sensor 631, a near-infrared photometer 632, a turbidity meter 633, a soluble solid sensor 634, a conductivity sensor 640, a flow sensor 650 and a dispersion stability analysis device 660. In some embodiments, as shown in FIG. 1 , the setting position of the dispersion stability analysis device 660 may be different from the setting positions of other monitoring elements (including, for example, the viscosity sensing element 610, the particle size sensing element 620, the total suspended matter sensing element 631, the near-infrared photometer 632, the turbidity meter 633, the soluble solids sensing element 634, the conductivity sensing element 640, and the flow sensing element 650).

At least a portion of the second monitoring module 700 may be disposed in the container 100 to monitor the quality change of the mixed material M during the mixing process. In other words, the second monitoring module 700 is an online monitoring module. The second monitoring module 700 may select suitable monitoring elements according to different purposes or different characteristics of the mixed material M, and the number of monitoring elements used may be one or more. In some embodiments, the second monitoring module 700 may include but is not limited to one or more of the following: a specific gravity sensor 710, an ion sensor 720, a flow rate sensor 730, a temperature sensor 740, a liquid level sensor 750, and a load cell 760. In some embodiments, as shown in FIG. 1 , the placement position of the load cell 760 may be different from the placement positions of other monitoring elements (including, for example, the specific gravity sensing element 710 , the ion sensing element 720 , the flow rate sensing element 730 , the temperature sensing element 740 , and the liquid level sensing element 750 ).

The control unit C can be electrically connected to the stirring control module 200, the quality analysis device 500, the first monitoring module 600 and the second monitoring module 700 to control the stirring control module 200, receive the quality analysis data of the quality analysis device 500 and receive the quality monitoring data of the first monitoring module 600 and the second monitoring module 700, and establish at least one material mixing quality prediction model based on the quality analysis data and the quality monitoring data. In some embodiments, the control unit C may include a computer, a programmable logic controller or a combination thereof.

In some embodiments, the at least one material mixture quality prediction model can be established by deep learning the quality analysis data and the quality monitoring data using an artificial neural network (ANN). In some embodiments, the quality analysis data of the quality analysis device 500 can be used as the output value of the deep learning prediction model, and the quality monitoring data of the first monitoring module 600 and the second monitoring module 700 can be used as the input value of the deep learning prediction model. The artificial neural network model is trained and modified (including adjusting the weight of the neuron) by a backpropagation algorithm to establish the at least one material mixture quality prediction model. In some embodiments, the input value of the at least one material mixing quality prediction model may also be the material type, formula and/or operating parameters of the process equipment. In some embodiments, the artificial neural network model may be established by the software in the control unit C and may be used to: a) obtain each process parameter and/or material formula information, input to obtain the prediction value and compare and analyze with the set value; b) obtain the monitoring data of each monitoring module, input to obtain the prediction value and compare and analyze with the set value; c) calculate the mixing quality adjustment parameters at the end of the process based on the analysis results.

In some embodiments, the adaptability of the at least one material mixing quality prediction model is evaluated using the root mean square error (RMSE) and/or the correlation coefficient (R ² ), and the prediction model is further validated. When the predicted value does not match the corresponding actual value, the prediction model parameters can be adjusted according to the result to improve the prediction model correlation coefficient.

The feed control module A is electrically connected to the control unit C (including, for example, a computer and a programmable logic controller), and performs feed control actions according to an instruction of the control unit C. In some embodiments, when the computer receives the data values of each sensing element, it calculates through its internal software to assist in transmitting information such as the feed amount and feed sequence of the raw materials to the programmable logic controller, and issues instructions to the feed control module A through the programmable logic controller to perform feed control actions. In some embodiments, as shown in FIG1 , the feed control module A can cooperate with the material preparation container 810 to perform feed control actions.

The input/output module B is electrically connected to the first monitoring module 600, the second monitoring module 700 and the control unit C, and is used to collect the quality monitoring data at one time and output the quality monitoring data to the control unit C for data analysis or display.

From the above description, it can be known that the functions of the control unit C include receiving the monitoring data and aggregating the data of the above monitoring modules. The computer is equipped with analysis software to perform data analysis, model prediction and subsequent parameter adjustment design. The at least one material mixing quality prediction model is established with ANN, and the model is verified by adjusting the neural weights, and the trained model is further used to predict the quality of the mixed material. The corresponding correction design can also be further calculated for the material that does not meet the target setting value, and fed back to the stirring control module 200 or the feeding control module A.

The beverage preparation and mixing quality prediction system 1 of the present invention obtains the quality monitoring data and quality analysis data of the mixed material M during the mixing process through an online monitoring module (including, for example, the first monitoring module 600 and the second monitoring module 700) and an offline analysis device (including, for example, the quality analysis device 500), and establishes at least one material mixing quality prediction model based on the quality monitoring data and the quality analysis data combined with an artificial neural network (ANN) and a back propagation algorithm. Through the at least one material mixing quality prediction model, the quality of the mixed material M under different material formulas or process equipment parameters can be predicted. And through online real-time monitoring, material deterioration, agglomeration and/or denaturation caused by unnecessary over-mixing can be reduced, thereby improving the quality and preparation efficiency of the mixed material M, shortening the product development schedule and assisting factories in establishing standardized preparation procedures, thereby significantly improving process quality.

Refer to FIG. 2, which is a diagram showing the architecture of a beverage preparation mixed quality prediction system 1a according to some embodiments of the present invention. FIG. 2 is one of the application scenarios of the prediction model established in FIG. The beverage preparation mixed quality prediction system 1a in FIG. 2 has a similar architecture to the beverage preparation mixed quality prediction system 1 in FIG. 1, and the only difference is that the beverage preparation mixed quality prediction system 1a in FIG. 2 omits the quality analysis device 500 in FIG. 1.

In some embodiments, as shown in FIG2 , the control unit C can indirectly obtain the mixing quality similar to that measured by the instrument through the established prediction model based only on the input of the quality monitoring data of the first monitoring module 600 and the second monitoring module 700, so as to understand the stirring state of the mixed material at the current stage. In some embodiments, the material formula and/or process parameter conditions can also be brought into the established prediction model to predict the target quality of the mixing.

Refer to FIG3, which is a block diagram of a beverage preparation mixed quality prediction system 1b according to some embodiments of the present invention. FIG3 is one of the application scenarios of the prediction model established in FIG1. The beverage preparation mixed quality prediction system 1b of FIG3 has a similar structure to the beverage preparation mixed quality prediction system 1 of FIG1, and the only difference is that the beverage preparation mixed quality prediction system 1b of FIG3 omits the first monitoring module 600, the second monitoring module 700 and the input/output module B of FIG1, and the control unit C of FIG3 may include a computer C1 and a programmable logic controller C2.

In some embodiments, as shown in FIG. 3 , the material formula and/or process parameter conditions can be input into the established prediction model to predict the mixing quality as the target setting value. In the mixing process, the software in the computer C1 can only analyze and compare the quality analysis data of the quality analysis device 500 with the target setting value, and calculate the correction parameters based on the difference, which can be the adjustment of the rotation speed, formula or operating temperature, etc., and further control the stirring control module 200 through the programmable logic controller C2 to achieve the set quality target.

Refer to FIG. 4, which is a diagram showing the architecture of a beverage preparation mixed quality prediction system 1c according to some embodiments of the present invention. FIG. 4 is one of the application scenarios of the prediction model established in FIG. The beverage preparation mixed quality prediction system 1c of FIG. 4 has a similar architecture to the beverage preparation mixed quality prediction system 1 of FIG. 1, and the only difference is that the beverage preparation mixed quality prediction system 1c of FIG. 4 omits the quality analysis device 500 of FIG. 1, and the control unit C of FIG. 4 may include a computer C1 and a programmable logic controller C2.

In some embodiments, as shown in FIG. 4 , the control unit C may only predict the quality based on the material formula and/or process parameter condition input as the setting value of the mixing target. During the mixing process, the quality monitoring data of the first monitoring module 600 and the second monitoring module 700 are input into the established model, and the mixed quality similar to that measured by the instrument is obtained through calculation. The mixed quality is analyzed and compared with the target setting value through the software in the computer C1, and the correction parameters are calculated based on the difference, which may be the adjustment of the rotation speed, formula or operating temperature, etc., and the stirring control module 200 is further controlled through the programmable logic controller C2 to achieve the set quality target.

The present invention is described in detail with the following examples, but it is not intended that the present invention is limited to the contents disclosed in these examples.

The mixing and blending modes applicable to the present invention may include but are not limited to solid-liquid powder dispersion, solid material dissolution, compatibilizing liquid-solution mixing, and two-phase immiscible liquid emulsification, etc. In addition, the system provided by the present invention may include but is not limited to solid powder, liquid, and gel, etc., and is not limited to a single material, but may also be a combination of multiple materials.

[ Invention Example 1]

Taking the preparation of soy protein powder solution as an example, Pro-Fam ^® 646 soy protein powder is used as the raw material and water is used as the solvent. Protein solutions of different concentrations (5-10%) are prepared at different preparation temperatures (including, for example, 30°C, 50°C, 70°C and 90°C) and pH values (pH3-7).

Referring to FIG. 1 , in the first embodiment of the invention, the stirring control module 200 may include the stirring element 230, the driving motor 210 and the torque sensing element 220. The stirring element 230 is a propulsion stirring blade, and the driving motor 210 is controlled to drive the stirring element 230 to perform a dispersion and dissolution process of the powder material.

In the invention example 1, the quality analysis device 500 can be the fluid rheology analysis device 510, which measures the viscosity of the soy protein solution prepared under different conditions by sampling, and uses it as the output value for establishing the prediction model.

In Invention Example 1, the first monitoring module 600 may include the soluble solid sensing element 634 and the conductivity sensing element 640. The soluble solid sensing element 634 and the conductivity sensing element 640 are placed in the circulation pipeline 400 to monitor the soluble solid and conductivity changes of the materials in the stirring and mixing process respectively. In addition, the first monitoring module 600 may further include the dispersion stability analysis device 660, which can be connected to the circulation pipeline 400, and through the circulation pump 300 (invention example 1 uses a peristaltic pump) at a low speed, stable and turbulent state, the mixed material in the container 100 is transported to the sample tank in the analysis device, and the mixed material in the sample tank is kept full of the tube, and then the material mixing characteristics, such as volume fraction, are analyzed by optical principles.

In Invention Example 1, the second monitoring module 700 may include the ion sensor 720 and the flow rate sensor 730, which are inserted into the container 100, and the ion sensor 720 is used to measure the circulation condition of the dead corner (poor fluidity area) in the mixing container 100, and the flow rate sensor 730 is used to measure the flow rate of the wall of the container 100 during the mixing process.

In Invention Example 1, the first monitoring module 600 may further include the flow sensor 650, and the second monitoring module 700 may further include the temperature sensor 740, for monitoring changes in the circulation and mixing process environment to ensure the accuracy of the measurement data and optimize the mixing or cleaning and sterilization process.

In Invention Example 1, since the formula contains a variety of raw material combinations, the liquid level sensor 750 or further includes the monitoring value of the load cell 760 to assist in controlling the input amount of each raw material and confirm whether the formula capacity and/or total weight setting value has been reached, so that the mixed material in the container 100 can maintain the correct ratio.

Afterwards, the first monitoring module 600 and the second monitoring module 700 are connected to the input/output module B to aggregate the quality monitoring data, and further transmit the quality monitoring data to the control unit C as an input value for establishing a prediction model.

In Invention Example 1, an ANN prediction model is established with reference to FIG. 1. The establishment of the ANN prediction model may include: (1) using different materials and/or process condition information as input values for establishing the prediction model, and using the measured value of the fluid rheology analysis device 510 as the output value for establishing the prediction model, so as to establish a target quality index that can be predicted under different formulation processes; and (2) using the signal value of the torque sensing element 220 and/or the flow rate sensing element 730 as the input value for establishing the prediction model, and using the viscosity value sampled and measured by the fluid rheology analysis device 510 as the output value for establishing the prediction model. Through the established prediction model, the signal of the above-mentioned online sensing element can be input during the mixing process to predict the viscosity value of the mixed material during the mixing process, so that the mixing status of the mixed material in the container 100 can be confirmed in real time.

The steps of establishing the above-mentioned ANN prediction model may include: organizing the information or data of different processing groups and inputting them into the software for processing, using the artificial neural network system to train the prediction model by adjusting the number of layers, the number of neurons and/or the excitation function, and using the back propagation algorithm; and inserting the verification data that is not used to train the prediction model into the trained prediction model, comparing the difference between the predicted value and the actual value, and modifying the parameters of the prediction model through the verification data to improve the correlation coefficient of the prediction model (>0.8).

In addition, for the application scenario of Figure 2, the soy protein powder can be subjected to a stirring and blending process, and the real-time monitoring signal value of the torque sensing element 220 and/or the flow rate sensing element 730 is input into the model, and the quality measurement similar to that of the instrument is indirectly obtained. For the scenarios of Figures 3 and 4, the above-mentioned trained prediction model is used to predict the dispersion and dissolution quality and real-time monitoring and calculation of correction parameters. When a 10% neutral soy protein dispersion solution is to be prepared, the prediction model can be used to first predict the target setting value (viscosity value 250.0±23.0 cP) for achieving complete dispersion under specific process conditions (neutral pH 7, blending temperature 50°C), and then the soy protein powder can be subjected to a stirring and blending process. For the application of FIG. 3 , the quality analysis device can be used to measure the current quality. If it meets the predicted results, it means that the dispersion target has been achieved. If the quality does not meet the predicted results (about 100 cP), it can be inferred that complete dispersion and dissolution have not been achieved. At this time, the computer C1 of FIG. 3 can design a set of adjustment parameters based on the prediction model, and send instructions to related equipment through the programmable logic controller C2 of FIG. 3 to perform operations such as increasing the drive shaft speed, adjusting the temperature, and extending the stirring time to achieve the expected target quality. For the application of FIG. 4 , the quality can be indirectly analyzed through the signals of the first monitoring module and the second monitoring module. If the predicted result is met, it means that the homogenization target has been achieved. If the quality of the stirring process does not meet the target setting value (about 100 cP), it can be inferred that complete dispersion and dissolution have not been achieved. At this time, the computer C1 of FIG. 4 can design a set of adjustment parameters according to the prediction model, and send instructions to related equipment through the programmable logic controller C2 of FIG. 4 to perform operations such as increasing the drive shaft speed, adjusting the temperature and extending the stirring time to achieve the desired target quality.

[ Invention Example 2 ]

Taking the homogenization of protein solution and liquid fat as an example, the protein mentioned may include whey protein, casein and combinations thereof of animal proteins, or soy protein, pea protein, sesame protein, chickpea protein and combinations thereof of plant proteins. Invention Example 2 uses SUPRO ^® 120 IP soy protein isolate.

The liquid oil can be a plant material or an animal oil, including, for example, soybean oil, sunflower oil, canola oil, medium-chain fat, fish oil, and combinations thereof. Invention Example 2 uses medium-chain fat.

Inventive Example 2, different oil and protein ratios (0.1-2) were used to prepare oil-containing vegetable protein mixtures at different preparation temperatures (30-80° C.) or drive shaft speeds (2,000-10,000 rpm).

With reference to FIG. 1 , in the second embodiment of the invention, the stirring element 230 may be a toothed rotor-stator cutter, and the driving motor 210 is controlled to drive the stirring element 230 to perform a homogenization and emulsification process of the material.

In Invention Example 2, the quality analysis device 500 may include the particle size analysis device 520 and the electrokinetic potential analysis device 530, which respectively measures the particle size, particle size distribution and delta-electromotive force of the material under different homogenization parameters or formulation conditions by sampling to establish a mixed homogenization quality index.

In Invention Example 2, the first monitoring module 600 may include the turbidity meter 633, the conductivity sensing element 640 and the flow sensing element 650. The turbidity meter 633, the conductivity sensing element 640 and the flow sensing element 650 are placed in the circulation pipeline 400 to indirectly monitor the homogeneity degree and the mixing state of the materials in the container 100 respectively. In addition, the first monitoring module 600 may further include the dispersion stability analysis device 660, which can be connected to the circulation pipeline 400, and through the circulation pump 300 (invention example 2 uses a peristaltic pump) at a low speed, stable and turbulent state, the mixed material in the container 100 is transported to the sample tank in the analysis device, and the mixed material in the sample tank is kept full of the tube, and then the material properties, such as particle size or backscattering rate, are analyzed by optical principles.

In Invention Example 2, the second monitoring module 700 may include the temperature sensing element 740 and the liquid level sensing element 750, which are inserted into the container 100, so that the temperature sensing element 740 monitors the temperature change of the material during the homogenization process, and the liquid level sensing element 750 can be used to instantly determine the material input amount or the degree of bubble formation during the homogenization process, so as to improve the controllability and accuracy of the preparation process.

The data values of the above-mentioned online sensing elements can be directly used as quality indicators for monitoring the mixture, and the relative quality can also be indirectly obtained by establishing a model between the monitored data and the measurement values of the offline device.

In Invention Example 2, an ANN prediction model is established, and its input values may include material formula, process parameter conditions and sensing values of sensing elements (including, for example, turbidity value or light scattering rate), and the output values may include the measured values of the particle size analysis device 520 and/or the dispersion stability analysis device 660. The information such as parameters and data of different process processing groups is aggregated and input to the control unit C, and the prediction model is trained by the artificial neural network system and the back propagation algorithm through its internal software; the verification data not used for training the prediction model is then inserted into the established prediction model, and the difference between the prediction value and the actual value is compared. The parameters of the prediction model are corrected through the verification data to improve the correlation coefficient of the prediction model (>0.8).

Applying the above prediction model to the application scenario of Figure 4, when preparing a 5% soy protein mixture containing 10% medium-chain fat, the above established prediction model can be used to predict that the target particle size range of the emulsion that can be obtained under the set process conditions (mixing temperature 80°C, tool speed 7000 rpm, homogenization time 10 minutes) is about 5~8 μm (set value). Afterwards, during the homogenization process, the real-time monitoring values obtained by the above-mentioned sensing elements can be aggregated through the input/output module B and then transmitted to the control unit C to predict the current quality. If the predicted result falls between 5 and 8 μm, it means that the homogenization target has been achieved. If the set value has not yet been reached, the computer C1 in Figure 4 will design the adjustment parameters based on the weights of each neuron in the prediction model, and send instructions to related equipment through the programmable logic controller C2 in Figure 4 to perform operations such as increasing the drive shaft speed, adjusting the temperature, and extending the stirring time to achieve the desired target quality.

Through the description of the above invention examples 1 and 2, the beverage preparation and mixing quality prediction system of the present invention can ensure that the mixed materials meet the required quality requirements, reduce the quality deterioration caused by human misjudgment or excessive stirring in the process, and can also ensure the quality change of materials/mediums in the mixing process in real time, further improving the consistency of product quality. In addition, the beverage preparation and mixing quality prediction system of the present invention can automatically adjust parameters to achieve the target quality, so as to reduce the errors and cost losses caused by human experience judgment.

Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It is understood that these terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning consistent with the background or context of the relevant technology and this disclosure, and should not be interpreted in an idealized or overly formal manner unless specifically defined herein.

The above embodiments are only for illustrating the principle and effect of the present invention, but not for limiting the present invention. The modification and changes made by the persons skilled in the art to the above embodiments still do not violate the spirit of the present invention. The scope of the rights of the present invention shall be as listed in the scope of the patent application described below.

1: Beverage preparation and mixing quality prediction system 1a: Beverage preparation and mixing quality prediction system 1b: Beverage preparation and mixing quality prediction system 1c: Beverage preparation and mixing quality prediction system 100: Container 101: Upper part 102: Lower part 103: Discharge valve 200: Stirring control module 210: Driving motor 220: Torque sensor 230: Stirring element 300: Circulation pump 400: Circulation pipeline 500: Quality analysis device 510: Fluid rheology analysis device 520: Particle size analysis device 530: Electrokinetic potential analysis device 600: First monitoring module 610: Viscosity sensor 620: Particle size sensor 631: Total suspended matter sensor 632: Near infrared photometer 633: Turbidity meter 634: Soluble solids sensor 640: Conductivity sensor 650: Flow sensor 660: Dispersion stability analysis device 700: Second monitoring module 710: Specific gravity sensor 720: Ion sensor 730: Flow rate sensor 740: Temperature sensor 750: Liquid level sensor 760: Load cell 810: Material preparation container A: Feed control module B: Input/output module C: Control unit C1: Computer C2: Programmable logic controller M: Mixing material

When read in conjunction with the accompanying drawings, it is easy to understand the aspects of some embodiments of the present invention from the following detailed description. It should be noted that various structures may not be drawn to scale, and the size of various structures may be arbitrarily increased or reduced for the sake of clarity of discussion.

FIG. 1 shows an architecture diagram of a beverage preparation mixing quality prediction system according to some embodiments of the present invention.

FIG. 2 shows an architecture diagram of a beverage preparation mixing quality prediction system according to some embodiments of the present invention.

FIG3 shows an architecture diagram of a beverage preparation mixing quality prediction system according to some embodiments of the present invention.

FIG4 shows an architecture diagram of a beverage preparation mixing quality prediction system according to some embodiments of the present invention.

1: Beverage preparation and mixing quality prediction system

100:Container

101: Upper part

102: Lower part

103: Discharge valve

200: Stirring control module

210: Driving motor

220: Torque sensing element

230: Stirring parts

300: Circulation pump

400: Circulation pipeline

500: Quality analysis device

510: Fluid rheology analysis device

520: Particle size analysis device

530: Electropotential analysis device

600: First monitoring module

610: Viscosity sensing element

620: Particle size sensor element

631: Total suspended object sensing element

632:Near infrared photometer

633: Turbidity meter

634: Soluble solids sensor element

640: Conductivity sensor element

650: Flow sensor element

660: Dispersion stability analysis device

700: Second monitoring module

710: Gravity sensing element

720: Ion sensor element

730: Flow rate sensing element

740: Temperature sensing element

750: Liquid level sensing element

760: Load element

810: Material preparation container

A: Feeding control module

B: Input/output module

C: Control unit

M: Mixed materials

Claims (16)

A beverage preparation and mixing quality prediction system comprises: a container for containing a mixed material; a stirring control module comprising a stirring element, the stirring element extending into the container for stirring the mixed material; a first monitoring module arranged outside the container for real-time monitoring of the quality of the mixed material, and the first monitoring module selects an appropriate monitoring element according to different mixing modes of the mixed material; a second monitoring module, at least a part of which is arranged in the container, for monitoring the quality change of the mixed material during the mixing process, and the second monitoring module selects an appropriate monitoring element according to different purposes or different characteristics of the mixed material; and a control unit, which is electrically connected to the stirring control module, the first monitoring module and the second monitoring module. Module, used to control the stirring control module and receive the quality monitoring data of the first monitoring module and the second monitoring module, and establish at least one material mixing quality prediction model according to the quality monitoring data; wherein the at least one material mixing quality prediction model is established after deep learning the quality monitoring data by artificial neural network, and the quality monitoring data of the first monitoring module and the second monitoring module are used as input values of the deep learning prediction model, and the artificial neural network model is trained and corrected by the back propagation algorithm to establish the at least one material mixing quality prediction model; wherein the at least one material mixing quality prediction model calculates the corresponding correction design for the material that does not meet the target setting value, and feeds back to the stirring control module. As in claim 1, the beverage preparation and mixing quality prediction system, wherein the stirring control module further includes a driving motor and a torque sensing element, and the driving motor is connected to the stirring element through the torque sensing element. The beverage preparation and mixing quality prediction system of claim 1 further includes: a circulation pipeline, which is connected to an upper part and a lower part of the container, for guiding the mixed material into and out of the container; and a circulation pump, which is arranged on the circulation pipeline, for circulating and transporting the mixed material into and out of the container. As in claim 3, the beverage preparation and mixing quality prediction system, wherein the first monitoring module is disposed on the circulation pipeline. As in claim 1, the beverage preparation and mixing quality prediction system, wherein the first monitoring module includes one or more of the following: a viscosity sensor, a particle size sensor, a total suspended matter sensor, a near-infrared photometer, a turbidity meter, a soluble solids sensor, a conductivity sensor, a flow sensor, and a dispersion stability analysis device. As in claim 1, the beverage mixing quality prediction system, wherein the second monitoring module includes one or more of the following: a specific gravity sensor, an ion sensor, a flow rate sensor, a temperature sensor, a liquid level sensor, and a load cell. The beverage preparation mixed quality prediction system of claim 1 further comprises: a quality analysis device, which is arranged on one side of the container and is used to analyze the quality of the mixed material taken out from the container. As in claim 7, the beverage preparation and mixing quality prediction system, wherein the quality analysis device includes one or more of the following: a fluid rheology analysis device, a particle size analysis device, and an electrokinetic potential analysis device. As in claim 7, the beverage mixing quality prediction system, wherein the control unit receives the quality analysis data from the quality analysis device and establishes the at least one material mixing quality prediction model based on the quality analysis data. The beverage mixing quality prediction system of claim 1 further includes: a feed control module, which is electrically connected to the control unit and performs feed control actions according to an instruction of the control unit. The beverage mixing quality prediction system of claim 1 further includes: an input/output module electrically connected to the first monitoring module, the second monitoring module and the control unit, for collecting the quality monitoring data at one time and outputting the quality monitoring data to the control unit. As in claim 1, the beverage preparation and mixing quality prediction system, wherein the control unit includes a computer, a programmable logic controller or a combination thereof. A beverage preparation mixed quality prediction system includes: a container for containing a mixed material; a stirring control module including a stirring element, the stirring element extending into the container for stirring the mixed material; a quality analysis device disposed on one side of the container for analyzing the quality of the mixed material taken out of the container; and a control unit electrically connected to the stirring control module and the quality analysis device for controlling the stirring control module and receiving quality analysis data from the quality analysis device, and establishing at least one physical property according to the quality analysis data. The at least one material mixing quality prediction model is established after deep learning the quality analysis data using an artificial neural network, and the quality analysis data of the quality analysis device are used as the output value of the deep learning prediction model, and the artificial neural network model is trained and corrected by a back propagation algorithm to establish the at least one material mixing quality prediction model; wherein the at least one material mixing quality prediction model calculates the corresponding correction design for the material that does not meet the target setting value, and feeds back to the stirring control module. As in claim 13, the beverage preparation and mixing quality prediction system, wherein the stirring control module further includes a driving motor and a torque sensing element, and the driving motor is connected to the stirring element through the torque sensing element. As in claim 13, the beverage preparation and mixing quality prediction system, wherein the quality analysis device includes one or more of the following: a fluid rheology analysis device, a particle size analysis device, and an electrokinetic potential analysis device. The beverage mixing quality prediction system of claim 13 further includes: a feed control module, which is electrically connected to the control unit and performs feed control actions according to an instruction of the control unit.

Images