CN120754612B

CN120754612B - Automatic control system for soybean milk separation

Info

Publication number: CN120754612B
Application number: CN202511286740.0A
Authority: CN
Inventors: 廖述成; 李文涛; 童先定; 邓超华; 蔡江; 彭勇; 沈小林
Original assignee: Wugang Linfeng Bean Products Equipment Co ltd
Current assignee: Wugang Linfeng Bean Products Equipment Co ltd
Priority date: 2025-09-10
Filing date: 2025-09-10
Publication date: 2025-12-09
Anticipated expiration: 2045-09-10
Also published as: CN120754612A

Abstract

This invention provides an automated control system for soybean milk separation, applied to soybean product processing equipment, belonging to the field of soybean product processing technology. The main controller is the core processing unit of the automated control system, with built-in pre-configured multi-stage control logic for soybean milk separation. A sensor group, electrically connected to the input port of the main controller, is used to collect environmental parameters during the soybean milk separation process. A human-machine interface, communicatively connected to the main controller, receives user-set process parameters and displays the real-time operating status of the automated control system. The main controller analyzes the environmental parameters to obtain an initial control command set, adjusts the initial control command set according to the process parameters and operating status, and sends it to the actuator group to execute corresponding actions. This solves the pain points of traditional soybean milk separation methods, such as fragmented parameter control, excessive manual intervention, and lengthy setup times during production changes.

Description

Automatic control system for soybean milk separation

Technical Field

The invention relates to the technical field of bean product processing, in particular to an automatic control system for soybean milk separation.

Background

The soybean milk separation is used as a key link for determining the quality of products, the technical level of the soybean milk separation directly influences the protein extraction rate, the product stability and the production efficiency, but the technology has significant limitations in the aspects of automatic control and parameter coordination. For example, vibration screening equipment is the mainstream choice of medium-and small-sized bean products enterprise, realizes thick liquid sediment separation through the three-dimensional vibration of fixed frequency, but its vibration frequency, core parameters such as amplitude need manual work preset and whole fixed. In actual production, the fixed parameter mode cannot adapt to the characteristic difference of raw materials with different bean varieties and different soaking degrees, when the soybean varieties with high protein content are processed, the screen mesh is blocked due to insufficient vibration intensity, and excessive vibration is easily caused to the raw materials with low water content, so that protein denaturation is caused. In addition, the temperature control of the equipment is mostly dependent on an independent heating device, and is lack of linkage with the screening process, so that when the temperature of the soybean milk deviates from an optimal separation interval, the problems of separation efficiency reduction or exceeding of the water content of the soybean dregs are easy to occur.

The key defects of the prior art are concentrated in three aspects, namely, the fragmentation of parameter control, the independent closed-loop control of key parameters such as temperature, liquid level and vibration intensity are adopted, a cooperative adjustment mechanism is lacked, such as the amplitude adjustment and concentration change of a vibrating screen are not related, the self-adaption capability is insufficient, the process parameters cannot be automatically corrected according to the bean variety difference and batch change, the shutdown and readjustment are needed when the bean variety is changed, the automation degree is limited, most of equipment still depends on the manual judgment and the parameter correction, the labor intensity is high, the fluctuation of the product percent of pass is overlarge due to the human error, namely, the production requirements of high precision, high stability and low manual intervention are difficult to meet along with the promotion of consumers on the quality consistency requirement of soybean milk and the pursuit of industrial production on efficiency in the prior art.

The present invention therefore proposes an automated control system for soymilk separation.

Disclosure of Invention

The invention provides an automatic control system for soybean milk separation, which is applied to soybean product processing equipment and comprises the following components:

the main controller is a core processing unit of the automatic control system and internally provided with a preconfigured soybean milk separation multi-stage control logic, wherein the multi-stage control logic comprises a concentration self-adaptive regulation and control stage and a temperature-liquid level cooperative control stage;

The sensor group is electrically connected with an input port of the main controller and is used for collecting environmental parameters of the soybean milk separation process, and at least comprises a liquid level sensor, a temperature sensor, a viscosity detection unit, a photoelectric turbidity sensor and a photoelectric turbidity sensor, wherein the liquid level sensor is used for detecting the liquid level height of the soybean milk in the separation bin;

the man-machine interaction interface is in communication connection with the main controller and is used for receiving the technological parameters set by a user and displaying the working state of the automatic control system in real time;

The main controller is used for analyzing the environmental parameters to obtain an initial control instruction set, wherein the initial control instruction set comprises an initial motor control instruction for a vibration motor driver, an initial heating control instruction for a heating control circuit and an initial trigger opening instruction or an initial closing instruction for a water inlet electromagnetic valve of the vibration motor driver;

And adjusting an initial control instruction set according to the technological parameters and the working state, and issuing the initial control instruction set to an actuator group to execute corresponding actions, wherein the actuator group at least comprises:

a vibration motor driver for driving and controlling the vibration frequency and amplitude of the separating screen;

The heating control circuit is used for adjusting the power of the heating pipe according to the temperature feedback;

And the water inlet electromagnetic valve is used for controlling the on-off of water replenishing in the separation bin.

Preferably, the main controller is used for receiving the current signal collected by the viscosity detection unit as a first feedback value representing the concentration of the soybean milk and receiving the current signal collected by the photoelectric turbidity sensor as a second feedback value representing the turbidity of the soybean milk, so as to determine the initial vibration frequency of the vibration motor driver;

And (2) and ;

Wherein, the An initial vibration frequency of the vibration motor driver; a reference vibration frequency for the vibration motor driver; is a first feedback value; A reference viscosity current value corresponding to the standard concentration; Is a second feedback value; a reference turbidity current value corresponding to the standard turbidity; the viscosity is used as a weight coefficient; the turbidity is used as a turbidity influence weight coefficient; correcting the coefficients for the separation stage; Correcting the coefficient for bean seeds; is a frequency analysis function;

The main controller is used for performing first comparison on a first feedback value obtained at different moments and a preset concentration-vibration parameter mapping table, and performing second comparison on a second feedback value obtained at different moments and a preset turbidity-vibration parameter mapping table, so as to construct a two-dimensional array;

The main controller is used for carrying out fitting analysis on two line numbers in the two-dimensional array respectively, and carrying out step analysis on a concentration fitting function and a turbidity fitting function to obtain an offset set;

The main controller is used for expanding the range of the initial vibration frequency according to the offset set and screening the intermediate value of the expanded range as a target vibration frequency;

The main controller is used for being based on the target vibration frequency Obtaining a target amplitude of the vibration motor driver;

;

wherein, the An initial amplitude for the vibration motor driver; a reference amplitude for the vibration motor driver; Is the influence coefficient of turbidity on amplitude; is a frequency correction term; The maximum vibration frequency in the two-dimensional array is set; Is the aperture of the screen; The length of the wheel base of the vibration motor is;

And the main controller is used for obtaining an initial motor control instruction for controlling the vibration motor driver to work based on the target vibration frequency and the target amplitude.

Preferably, the main controller is used for calculating the output power duty ratio of the heating control circuit;

;

wherein, the A real-time output power duty cycle for the heating control circuit; Is a proportionality coefficient; is the temperature deviation; Is a temperature piecewise coefficient, and ;Is an integration time constant; Is an integral correction coefficient, and ;Is a differential time constant; Is a feedforward compensation coefficient; The target temperature at the current moment in a preset temperature curve; the real-time temperature of the soybean milk detected by the temperature sensor; The real-time density of the soybean milk at the current temperature; the real-time specific heat capacity of the soybean milk at the current temperature; Standard density of soymilk at 25 ℃; standard specific heat capacity of soybean milk at 25 ℃;

The main controller is also used for controlling the power based on the final power duty cycle An initial heating control instruction for controlling the heating pipe is obtained.

Preferably, the main controller is used for receiving the liquid level of the soybean milk acquired by the liquid level sensor in real time, and generating an initial trigger starting instruction when the liquid level of the soybean milk is lower than the lower limit value of the liquid level safety threshold range;

determining a current opening instruction of the water inlet electromagnetic valve according to the historical actual opening instruction and the historical actual opening of the water inlet electromagnetic valve, and controlling the water inlet electromagnetic valve to open for water supplement;

And when the liquid level data reaches the upper limit of the threshold range, generating an initial closing instruction to control the water inlet electromagnetic valve to be closed.

Preferably, the main controller is used for determining the upper limit value and the lower limit value of the liquid level safety threshold value range required by the current production batch according to the planned output set by the man-machine interaction interface;

,;

wherein, the Is an upper limit value; the planned output is set on the basis of a human-computer interaction interface, and k2 is a second experience coefficient; Is a centrifugal force correction coefficient, and ,In order to separate the real-time rotational speed of the actuator,Maximum rotation speed of the separation executing mechanism; Is the cross-sectional area of the separation bin; Is a feed flow correction factor, and ,The real-time opening degree of the electromagnetic flow valve is set; The maximum opening of the electromagnetic flow valve; The full bin height of the separation bin; for comparing the height values;

;

wherein, the Is a lower limit value; Is a hysteresis interval.

Preferably, the main controller includes:

the computing unit is used for computing a difference function between the concentration fitting function and the turbidity fitting function, and carrying out cluster analysis on the concentration difference values at all acquisition moments according to the difference function to obtain standard deviation, mean value and cluster weight of each cluster analysis result;

The first deviation unit is used for obtaining a first sub-deviation set according to the standard deviation, the mean value and the clustering weight of each clustering analysis result and combining the first constant of the concentration fitting function and the second constant of the turbidity fitting function;

the dividing unit is used for dividing the function curve of the difference function according to preset continuous acquisition time to determine a single function of each dividing curve and obtain a single constant of the single function;

The second deviation unit is used for carrying out discrete analysis on the single constant to judge whether the discrete constant exists or not, and obtaining a second sub-deviation set according to the acquisition time of the discrete constant and the average value of the non-discrete constant;

The time group association unit is used for sorting the concentration of the concentration fitting function and the concentration of the turbidity fitting function at all the collection moments respectively, determining a collection time group with the same concentration sorting sequence number, and analyzing the two-dimensional association relation of the concentration fitting function and the turbidity fitting function according to all the collection time groups;

the expansion unit is used for determining the optimal offset according to the first sub-deviation set and the second sub-deviation set and combining the two-dimensional association relation, and performing range expansion on the initial vibration frequency by combining the current soybean milk liquid level.

Preferably, the main controller further comprises:

A sequence acquisition unit for acquiring a historical actual opening instruction sequence of the water inlet electromagnetic valve Corresponding historical actual opening sequenceAnd performs feature extraction to generate an opening time sequence feature vector Vh, wherein,、The j2 th time of history actual opening instruction and history actual opening degree are respectively,Indicating that the switch is turned on,Mo is the total number of instructions in the sequence;

constructing real-time parameter vectors , wherein,For the real-time level value of the separation bin,For the real-time temperature of the soybean milk,Is the target liquid level value;

A prediction unit for combining Vh with And inputting the current opening instruction into a pre-trained opening prediction model to obtain a current opening instruction of the output water inlet electromagnetic valve.

Preferably, the main controller is further configured to obtain a standard state of the automation system based on the process parameter, and perform differential analysis with the working state of the automation system to obtain a differential vector;

The main controller is further configured to input the difference vector to a pre-trained vector analysis model, obtain a first adjustment factor of each initial motor control instruction and a second adjustment factor based on the initial heating control instruction, and obtain an adjusted motor control instruction and an adjusted heating control instruction, so as to control the corresponding vibration motor driver and the heating control circuit to work.

Compared with the prior art, the application has the following beneficial effects:

the multi-stage control logic with concentration self-adaption and temperature-liquid level synergy is integrated through the main controller, and the parameter fragmentation control, the manual intervention and the pain points with long production changing and debugging in the traditional soybean milk separation are thoroughly solved by combining the synchronous acquisition of multiple sensors and the accurate control of an actuator.

Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description and drawings.

The technical scheme of the invention is further described in detail through the drawings and the embodiments.

Drawings

The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:

Fig. 1 is a block diagram of an automated control system for soymilk separation in accordance with an embodiment of the present invention.

Detailed Description

The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present invention only, and are not intended to limit the present invention.

The invention provides an automatic control system for soybean milk separation, which is applied to soybean product processing equipment, as shown in fig. 1, and comprises the following components:

In the embodiment, the bean product processing equipment is industrial equipment for bean product production, and the scheme focuses on matched equipment of a soybean milk separating section, namely special equipment for separating soybean milk from bean dregs, such as a vibrating screen separator, a horizontal spiral centrifugal separator, a negative pressure filtering separator and the like in a soybean milk production line.

The main controller is a programmable logic controller such as Siemens S7-1200CPU or a microcontroller such as STM32.

The multi-stage control logic for soybean milk separation is an algorithm logic which is built in the main controller and is used for realizing accurate control of soybean milk separation in stages, and the limitation of traditional single parameter control is solved. The concentration self-adaptive regulation and control stage is a stage for automatically regulating vibration parameters according to the real-time concentration of the soybean milk and ensuring the slurry-residue separation efficiency. And the temperature-liquid level cooperative control stage is used for associating the real-time temperature of the soybean milk with the liquid level of the separation bin, so that the efficiency waste or quality problem caused by single regulation and control is avoided.

The liquid level sensor is a throw-in type liquid level transmitter model JYB-KO-Y2, the measuring range is 0-100cm, and 4-20mA analog signals are output. The sensor probe is vertically thrown into a separation bin far away from the stirring or vibrating component, a cable is fixed at the top of the bin, and a signal cable is connected to an analog input terminal of the main controller.

The temperature sensor is PT100 platinum resistance temperature sensor, such as OMRONE-CA 1D, measuring range is 0-150deg.C, the sensor probe is packaged in food grade stainless steel sleeve, inserted into the separation bin, the depth of the separation bin is equal to the middle of soybean milk, the resistance signal is converted into 4-20mA signal by temperature transmitter, and connected to the main controller.

The viscosity detection unit selects an on-line rotary viscometer, such as Brookfield DV2T, the measurement range is 10-10000cP, 4-20mA analog signals are output, a measuring rotor of the viscometer is arranged on a feeding pipe of a separation bin and is fixed through a pipeline flange, a signal cable is connected into a main controller, and the acquisition is triggered synchronously with a turbidity sensor.

The photoelectric turbidity sensor is an online photoelectric turbidity meter, such as a Hash 2100Q, the measuring range is 0-1000NTU, 4-20mA analog signals are output, the transmitting end and the receiving end of the sensor are respectively arranged on two sides of a feeding pipeline, the sensor is fixed through a quick-mounting joint, the sensor and the viscosity detection unit share one acquisition trigger signal, and the signal is connected into the main controller.

Setting a timer in the main controller, outputting a switching value trigger signal at fixed time, accessing external trigger interfaces of the viscometer and the turbidity meter respectively, and setting a synchronous data receiving logic in the main controller program to ensure that data of 2 sensors are stored in the same data frame.

The man-machine interaction interface is an interaction medium between a user and an automatic control system, realizes a bidirectional function of inputting technological parameters and displaying the working state of the system, and is connected with the main controller through an RS485 communication interface, wherein the resolution is 1024 multiplied by 600 if a 7-inch or 10-inch industrial touch screen is selected.

The environmental parameters are data sets which are acquired by a sensor group in real time and reflect the current state of the soybean milk separating process, such as 75cm liquid level, 82 ℃ temperature, 480cP viscosity and 130NTU turbidity.

The initial control instruction set is an instruction set which is preliminarily generated by the main controller and used for controlling each actuator based on environmental parameters and preset basic logic, for example, the viscosity is 480cP, the turbidity is 130NTU, and the initial instructions are that the vibration frequency is 50Hz and the amplitude is 2.0mm.

The technological parameters are parameters set by a user through a human-computer interaction interface according to production requirements, and are the basis for adjusting initial instructions by a main controller, for example, the planned output is 1000L, the target temperature is 85 ℃, the reference vibration frequency of a vibration motor driver is 45Hz, the liquid level safety range is 60-90cm, and the bean seed correction coefficient is 1.1.

The working state is various state data of the system running in real time, including environmental parameters collected by the sensor and the current action state of the actuator, and is used for judging whether an adjustment instruction is needed by the main controller.

The vibration motor driver is a three-phase asynchronous vibration motor driver, such as a platform VFD022EL43A, the driver is arranged in an electric cabinet, the input end of the driver is connected with a 380V three-phase power supply, the output end of the driver is connected with a vibration motor, the control end of the driver is connected with a main controller through a signal wire to receive frequency and amplitude instructions, and the vibration motor is connected with a separating screen through a flange to ensure vibration transmission efficiency.

The heating control circuit consists of 220V alternating current power supply, fuse, solid state relay, heating pipe, thermal relay and zero line, wherein the control end of the solid state relay is connected with the main controller, the circuit is arranged on an insulating mounting plate in the electric cabinet, the heating pipe is arranged at the bottom or the side wall of the separation bin, and the thermal relay is used for setting overload protection current to prevent the heating pipe from being burnt.

The water inlet electromagnetic valve is a two-position two-way food-grade electromagnetic valve, such as SMCVX212,212, and the interface specification DN25. The electromagnetic valve is connected in series in the water supplementing pipeline, wherein the water inlet end is connected with tap water or a purified water pipeline, the water outlet end is connected with the top of the separation bin and is fixed through a pipe clamp, and the coil terminal of the electromagnetic valve is connected with the switching value output terminal of the main controller to ensure firm and looseness-proof wiring.

The technical scheme has the advantages that concentration self-adaption and temperature-liquid level cooperative multi-stage control logic is integrated through the main controller, parameter fragmentation control, more manual intervention and long-term pain point during product replacement and debugging in traditional soybean milk separation are thoroughly solved by combining multi-sensor synchronous acquisition and actuator precise control, soybean milk separation efficiency can be improved by 5% to 8%, product qualification rate fluctuation is reduced from +/-8% to +/-2% in practical application, product replacement and debugging time is shortened to 10 minutes from 1 to 2 hours, and meanwhile, 30% of manual operation quantity is reduced, so that high-efficiency and stable requirements of industrial continuous production of soybean products are met.

The invention provides an automatic control system for soybean milk separation, which is characterized in that a main controller is used for receiving a current signal acquired by a viscosity detection unit as a first feedback value representing the concentration of soybean milk and receiving a current signal acquired by a photoelectric turbidity sensor as a second feedback value representing the turbidity of soybean milk, so as to determine the initial vibration frequency of a vibration motor driver;

And (2) and ;

;

In this embodiment, the laboratory tests the optimal frequencies for different conditions and stores the standard condition frequencies in the master controller parameter table.

In this example, for example, soybean milk with a concentration of 8%,At 10mA, turbidity of 100NTU,8MA; is a coefficient for measuring the influence degree of viscosity on vibration frequency, the larger the value is, the more obvious the influence is, wherein the influence of viscosity on frequency is larger than turbidity, and the method is obtained according to a plurality of experimental fitting Takes a value of 0.8 and+。According to the adjustment coefficients of the separation stage, namely the initial separation stage, the middle separation stage and the later separation stage, the change rules of the bean dreg quantity and the concentration in different stages are different, generally, the initial separation stage is 10 minutes before the separation stage,The value of (2) is 1,2, high frequency is needed to prevent blockage, and the middle separation stage is neededHas a value of 1.0, and is in the later stage of separationThe value of (2) is 0.9.

The correction coefficient of soybean seeds is different according to the different values of soybean, red bean and black bean, for example, soybean1.0, Black bean particle hardness of.1, and red bean particle size of 0.9.

In this embodiment, the concentration-vibration parameter map is a soybean milk concentration-optimal vibration frequency parameter map established in advance through experiments, for example, the viscosity current 10mA corresponds to a concentration of 8%, and at this time, the frequency is 50Hz or the like.

In this embodiment, the turbidity-vibration parameter map is a soybean milk turbidity-optimal vibration frequency parameter map established in advance through experiments, for example, turbidity current 8 is turbidity 100NTU, and the frequency is 50Hz.

The two-dimensional array is a time sequence two-dimensional data structure of concentration comparison results and turbidity comparison results at different moments, for example, the concentration at the moment t1 corresponds to the frequency of 55Hz and the turbidity corresponds to the frequency of 53Hz, and at the moment, the two-dimensional data is:。

in this embodiment, the fitting analysis is to approximately describe the trend of the time-vibration parameters in the two-dimensional array by using a mathematical function, so as to obtain a concentration-turbidity fitting function, and the concentration-turbidity fitting function is a linear function, wherein y=kx+b.

For example, the concentration fitting function is y1=k01t+b1, where k01 is 0.5 Hz/min, b1 is 50Hz, and the turbidity fitting function is y2=k02t+b2, where k02 is 0.3 Hz/min, and b2 is 52Hz.

In this embodiment, the step analysis divides the difference function of the fitting function according to preset continuous acquisition time, and the preset continuous acquisition time may be M0 continuous times, so as to obtain an offset set, thereby realizing range expansion.

In this embodiment, the target vibration frequency is averaged by calculating the upper and lower limits of the expansion range.

In this example, the reference amplitude is the reference amplitude of the vibration motor at standard bean species and standard concentration/turbidity, and the coefficient of influence of turbidity on the amplitude is calculated by fitting a large number of realizations.

In the embodiment, the screen mesh diameter is directly obtained according to the screening model, the wheelbase length of the vibrating motor is determined when the equipment leaves the factory, and the vibrating motor is directly used.

The technical scheme has the beneficial effects that through synchronous feedback of the viscosity and turbidity sensor, formula calculation of multiple correction coefficients such as bean seeds and separation stages and dynamic adjustment of two-dimensional array fitting and step analysis are combined, accurate cooperative control of the frequency and the amplitude of the vibration motor is realized, the slurry and slag separation efficiency is improved, the screen blocking probability is reduced, and the continuity and the stability of the soybean milk separation process are ensured.

The invention provides an automatic control system for soybean milk separation, which is characterized in that a main controller is used for calculating the duty ratio of output power of a heating control circuit;

;

In this embodiment, the preset temperature curve is a preset target temperature track which changes with time, and guides the temperature change rule of heating the soybean milk.

In this embodiment, the feedforward compensation coefficient is used to compensate for the temperature variation trend in advance, for example,1.5 WhenAt =2 ℃ per minute, the feedforward loop output is 1.5×2=3.

The real-time density of the soybean milk is based onFrom a temperature-density look-up table, for example,At 80C, at which point,Is that。

In this embodiment, the master controller is based onThe generated instructions for controlling the on-off of the heating pipe or the power adjustment, for example,And when the PWM signal with the period of 1 second and the high level of 0.6 second is output, the heating pipe is controlled to work at 60% power.

The technical scheme has the beneficial effects that the improved PID control algorithm is adopted, the temperature segmentation coefficient and the integral correction coefficient are combined to adapt to the control characteristics of different temperature segments, the feedforward compensation is adopted to respond to the temperature change trend in advance, and meanwhile, the range constraint is carried out on the power duty ratio, so that the temperature control precision of heating the soybean milk is improved, the heating efficiency is improved, the heating pipe damage or the temperature runaway caused by the power overrun is avoided, and the stability of the soybean milk separation temperature condition is ensured.

The invention provides an automatic control system for soybean milk separation, which is characterized in that a main controller is used for receiving the soybean milk liquid level height acquired by a liquid level sensor in real time, and generating an initial trigger starting instruction when the soybean milk liquid level height is lower than the lower limit value of a liquid level safety threshold range;

,;

;

wherein, the Is a lower limit value; Is a hysteresis interval.

In this embodiment, for example, n=1500r/min,2000R/min, then=1.0375,=50%,100% Is thenFor example, the maximum accommodating height of the separation bin is 100cm.

For the hysteresis interval, the valve is prevented from being frequently started and stopped, for example,=10cm。

Preferably, the main controller further comprises:

In this embodiment, the initial trigger on command is that the fluid level is below a lower limit and the master controller generates an initial signal to initiate replenishment.

The historical actual opening command of the water inlet electromagnetic valve is an opening (1)/closing (0) command sequence sent by the main controller in the past, such as {1,0,1}, which indicates the 1 st opening, the 2 nd closing and the 3 rd opening. The historical actual opening is the actual opening of the solenoid valve corresponding to the opening command one by one, for example, {60%,0%,70% }.

In this embodiment, the current opening command is calculated by the main controller by combining the history command and the opening, and if the opening is 65% when the history liquid level is 45cm, the current command is 65%.

In this embodiment, the second empirical factor is an experimentally determined correction factor, for example, the k2 of soybean milk is 1.2.

In this embodiment, the opening time sequence feature vector Vh is an ordered number value group organized according to a fixed dimension, and converts an unstructured historical sequence into a structured input for a pre-trained opening prediction model to learn a historical rule, so as to predict the opening of the current electromagnetic valve, for example, vh may be expressed as:

。

In this embodiment, the pre-trained opening degree prediction model is a machine learning model trained based on more than 1000 sets of history data, the inputs of which are Vh and And outputting a current opening instruction.

The technical scheme has the beneficial effects that the main controller is used for collecting the liquid level in real time and combining with historical data, dynamically calculating the liquid level safety threshold value, and simultaneously accurately predicting the opening of the electromagnetic valve by using the pre-training model, so that the problems of large fluctuation of the liquid level and poor suitability of the traditional fixed threshold value are solved, the liquid level control accuracy is enabled to be within +/-2 cm according to the self-adaptive adjustment of the yield, the rotating speed, the feeding flow and the like, the water supplementing response speed is increased by 40%, the manual debugging is reduced, the liquid level stability of the soybean milk separation is ensured, and the slurry-residue separation efficiency and the product quality consistency are indirectly improved.

The invention provides an automatic control system for soybean milk separation, which comprises a main controller, wherein the main controller comprises:

In the embodiment, the difference function is obtained by fitting concentration differences based on a concentration fitting function and a turbidity fitting function at the same acquisition time, at this time, cluster analysis is performed on all concentration differences to obtain standard deviation, mean value and cluster weight of each cluster result, wherein the cluster weight is the proportion of data quantity in the cluster result to total data quantity, and the cluster algorithm is implemented by adopting a K-means algorithm, for example, the K-means algorithm clusters the differences into two groups of {0,1} { and {2,3,4}, at this time, the first group of mean values is 0.5, the standard deviation is 0.5, the cluster weight is 0.4, the second group of mean values is 3, the standard deviation is 1 and the cluster weight is 0.6.

In this embodiment, the first constant and the second constant are intercepts of the corresponding functions.

At this time, the deviation under each clustering result is (mean + standard deviation under the corresponding clustering result) ×corresponding weight+ (first constant-second constant), and a first sub-deviation set: { the deviation of each clustering result }, is obtained.

In this embodiment, the preset continuous acquisition time takes 3 units as a section, which is set in advance, and the single function is also a linear function of fitting, and the single constant is the intercept of the corresponding single function.

In this example, the discrete analysis was performed using 3The principle determines whether it is a discrete constant, e.g., the first set of constants is {1,1,5,1}, by 3The principle recognizes that 5 is a discrete constant, the moment of the discrete constant is moment 3, the average value of the non-discrete constant is 1, and a second sub-deviation set {0,0,4,0} obtained by calculating the deviation, wherein the discrete constant deviation is 5-1=4, and the non-discrete constant deviation is 0.

In this embodiment, for example, { the collection times corresponding to the 1 st turbidity and the 1 st concentration are combined into one set }, which is the collection time group with the concentration order number.

For example, the concentration fitting values are 3,5,7,9,11 at time 1,2,3,4,5, at this time, the sequenced serial number 1 corresponds to time 5, the sequenced serial number 2 corresponds to time 4, etc., the turbidity fitting values are 3,4,5,6,7 at time 1,2,3,4,5, etc., the sequenced serial number 1 corresponds to time 5, the sequenced serial number 2 corresponds to time 4, etc., the collection time groups with the same serial number are (5, 5), (4, 4), etc., the two-dimensional association relationship is that the concentration is positively correlated with the turbidity, and the sequenced serial numbers are identical, mainly by calculating the correlation coefficient to determine the two-dimensional association relationship.

In this embodiment, the calculation of the Pearson correlation coefficient r based on the concentration fitting value and the turbidity fitting value in all the acquisition time groups from Pearson correlation coefficients belongs to a known attempt, and if r is greater than 0 and represents positive correlation, if r is less than 0 and represents negative correlation, the closer r is to 1 or-1, and the stronger the correlation.

And extracting the sequencing sequence number of each time group and the corresponding concentration value and turbidity value, respectively drawing a sequence number-concentration curve and a sequence number-turbidity curve, judging that the trends are consistent if the two curves are ascending, descending or stable and synchronous, otherwise, judging that the trends are opposite.

Dividing the concentration value and the turbidity value into a plurality of intervals according to equidistant intervals, and counting the co-occurrence frequency of each concentration interval-turbidity interval:

If the co-occurrence frequency of the high concentration interval and the high turbidity interval is high, the co-occurrence frequency of the low concentration interval and the low turbidity interval is high, which indicates that the distribution matching degree of the two is high;

If the co-occurrence frequency of the high concentration interval and the low turbidity interval is high, the distribution matching degree is low.

For example, the high concentration {10%,12% } and the high turbidity {25,30} co-occur 2 times, which is the main co-occurrence combination, showing that the distribution matching degree is high.

Comprehensively judging the type of the association relation by combining the results of Pearson correlation coefficient, trend consistency and distribution matching degree:

If it is The trend consistency is more than or equal to 90%, and the distribution matching degree is high, and at the moment, the high synchronous correlation is judged;

If it is The trend consistency is less than or equal to 30 percent, and the distribution matching degree is low, and at the moment, the weak association/no association is judged;

if r < -0.8, the trend consistency is more than or equal to 90%, but the trend is opposite, the high concentration corresponds to low turbidity in distribution, the low concentration corresponds to high turbidity, and at the moment, the high reverse correlation is judged;

otherwise, the general association is determined.

In this embodiment, the average values of the first sub-deviation set and the second sub-deviation set are obtained as D1 and D2, respectively, and the deviation integrated influence amount f01=d1×wu1+d2×wu2 is calculated by combining the weights wu1 and wu2 associated with the two dimensions.

The setting of the empirical offset is directly obtained according to industrial control experience, and f02 is +2Hz.

Calculating the expansion base f03= (ideal liquid level-soybean milk liquid level) multiplied by 0.5Hz/cm according to the liquid level deviation;

Extended vibration frequency range: [ ft-f03 ] f01+f02, f01+ of f02 the process of the present invention.

The technical scheme has the beneficial effects that through multi-unit cooperation, difference analysis, clustering, deviation calculation, segmentation fitting and two-dimensional association analysis are carried out on the fitting function of concentration and turbidity, and finally the vibration frequency range is dynamically expanded by combining the liquid level, so that the limitation of poor suitability of vibration parameters in traditional control is broken through, the dynamic changes of the concentration, turbidity and liquid level of soybean milk can be self-adapted, the vibration frequency can be more accurately matched with the pulp-residue separation requirement, and the separation efficiency and the quality stability of soybean milk are improved.

The invention provides an automatic control system for soybean milk separation, which is characterized in that the main controller is also used for acquiring a standard state of the automatic system based on technological parameters and performing difference analysis with the working state of the automatic system to obtain a difference vector;

In the embodiment, when the process parameters are completely met, the ideal operation state of the system is obtained from a preset process parameter-standard state mapping table, for example, the process parameters are 98 ℃ and 45Hz, the heating power duty ratio is 80% in the standard state, the vibration frequency is stable by 45Hz, the liquid level is 60cm, the working state is a real-time state, for example, the actual temperature is 92 ℃, the vibration frequency is 42Hz, the liquid level is 55cm, and at the moment, the difference vector is { -6 ℃ -3Hz, -5cm }.

In this embodiment, the pre-trained vector analysis model is a machine learning model trained for data via a large number of differential vector-adjustment factors, for example, the differential vector is output { -6 ℃, -3Hz, -5cm }, the first adjustment factor is output +5Hz, the second adjustment factor is 1.2 times, and at this time, the coefficient of the adjusted initial heating command is 70% ×1.2, where 70% is the heating power duty ratio corresponding to the initial command.

The technical scheme has the beneficial effects that the soybean milk separating working condition fluctuation is adapted in real time by means of standard-actual state difference analysis and intelligent adjustment by combining a pre-training model, and the separating efficiency and the quality stability are ensured.

It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims

1. An automated control system for soymilk separation, applied to a bean product processing device, comprising:

The water inlet electromagnetic valve is used for controlling the on-off of water replenishing in the separation bin;

The main controller is used for receiving the current signal acquired by the viscosity detection unit as a first feedback value representing the concentration of the soybean milk and receiving the current signal acquired by the photoelectric turbidity sensor as a second feedback value representing the turbidity of the soybean milk, and determining the initial vibration frequency of the vibration motor driver;

And (2) and ;

;

the main controller is used for obtaining an initial motor control instruction for controlling the vibration motor driver to work based on the target vibration frequency and the target amplitude;

Wherein, the main control unit includes:

2. The automated control system for soymilk separation of claim 1, wherein said main controller is configured to calculate an output power duty cycle of said heating control circuit;

;

3. The automated control system for soymilk separation according to claim 2, wherein said main controller is configured to receive in real time the soymilk liquid level collected by said liquid level sensor, and generate an initial trigger on command when the soymilk liquid level is below a lower limit of a liquid level safety threshold range;

4. The automated control system for soymilk separation according to claim 3, wherein said main controller is configured to determine an upper limit value and a lower limit value of a liquid level safety threshold range required for a current production lot according to a planned output set by said human-computer interaction interface;

,;

;

wherein, the Is a lower limit value; Is a hysteresis interval.

5. The automated control system for soymilk separation of claim 4, wherein said master controller further comprises:

6. The automated control system for soymilk separation according to claim 1, wherein the main controller is further configured to obtain a standard state of the automated control system based on process parameters, and perform a difference analysis with an operating state of the automated control system to obtain a difference vector;