Edible mushroom stick inoculation parameter regulation method and system based on spectrophotometer

Abstract

The application discloses a kind of edible mushroom stick inoculation parameter regulation method and system based on spectrophotometer, it is related to measurement and analysis material field, the present application is by constructing "optical detection-data analysis-automatic regulation" closed loop control system carried out in the process of edible mushroom stick inoculation;Through specially designed optical detection unit, the OD value of bacteria liquid is acquired in real time.After that, temperature, humidity and CO2 concentration and other data collected by environmental sensor are used to establish model prediction for laboratory edible mushroom strain activity in combination with deep learning algorithm.Finally, the culture room environment and inoculation machine working parameters are dynamically adjusted using PLC control system;It can predict the trend of strain activity in advance, and then the culture parameters can be adjusted in time, so that the survival rate of mycelium is greatly increased and kept stable;Through intelligent environment regulation system, air conditioner / humidifying equipment power is matched according to real-time demand, and the comprehensive energy consumption of culture room is reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of measurement and analysis of materials, and particularly relates to a method and system for regulating inoculation parameters of edible mushroom sticks based on a spectrophotometer. BACKGROUND

[0002] Edible mushrooms are a kind of food rich in nutrients and unique in flavor, with characteristics of high protein, low fat and low sugar, and rich in 18 kinds of amino acids needed by the human body, among which the content of 7 kinds of essential amino acids for the human body is higher than that in milk and meat, and regular consumption is very beneficial to human health; however, in the process of inoculating edible mushrooms and production, the traditional edible mushroom inoculation process mainly relies on manual experience to judge the quality and inoculation amount of the strain, and the strain inoculation method under this condition makes the existing strain prone to long activation time, high pollution risk and low inoculation efficiency; edible mushroom inoculation refers to the process of transferring the target strain to the culture medium according to the aseptic operation technical requirements, which is a strict aseptic operation process, and the operator must establish a strict aseptic operation consciousness and perform inoculation operation according to the regulations, so this process tests the concentration of manual work;

[0003] Spectrophotometry is a method for qualitative and quantitative analysis of a substance by measuring the light absorption of the substance at a specific wavelength or within a certain wavelength range; it has the advantages of high sensitivity, simple operation and rapidness, and is the most commonly used experimental method in biochemical experiments; many substances are measured by spectrophotometry; in a spectrophotometer, when light of different wavelengths is continuously irradiated to a sample solution of a certain concentration, the absorption intensity corresponding to different wavelengths can be obtained, although it has been maturely applied in the field of microorganisms, but direct transplantation to edible mushroom production faces many challenges; the special properties of edible mushroom liquid bring significant technical obstacles: mycelium is easy to form clusters with a diameter of 200-500 μm, causing serious light scattering effect; insoluble components such as bran in the culture medium and micro-bubbles generated during production all lead to measurement errors of the traditional spectrophotometer often exceeding 15%; the complex environment on the industrial production line further increases the technical difficulty, and mechanical vibration, temperature fluctuation and electromagnetic interference in the production site seriously affect the stability of optical measurement; the existing patent technology attempts to modify the laboratory spectrophotometer for online detection, but the failure rate is high in actual operation, and the average mean time between failures is less than 500 hours, which is difficult to meet the reliability requirements of industrial production;

[0004] The current edible fungus antibacterial test is affected by many factors and the culture environment control system has obvious technical shortcomings; as the standard material for antibacterial test, the activity of the test strain in the antibacterial performance test and the amount of inoculation concentration will directly affect the accuracy of the test results, which is the most difficult to control and unify in the antibacterial test; at the same time, when the actuator starts to control the culture environment, 5-10 minutes of transition time is needed, there is a problem of response lag, and the culture environment cannot be accurately controlled; on the other hand, when the environment is adjusted, each parameter is isolated from each other, and the dynamic relationship between the strain state and the environmental conditions cannot be established.

[0005] Through industry practice, it is shown that the existing edible fungus stick inoculation online monitoring technology method needs to be broken through in many aspects; first, a real-time detection method suitable for high viscosity and multi-impurity bacteria liquid characteristics needs to be developed to solve the application obstacles of traditional optical measurement in edible fungus production; second, a multi-parameter comprehensive evaluation system of strain quality needs to be established to overcome the limitations of single detection index; at the same time, the collaborative adjustment and optimization of inoculation parameters and environmental control need to be realized to break the existing operation mode of each system; finally, the reliability and stability of the system in the industrial environment must be improved to ensure that the equipment can operate stably for a long time; the solution of these technical problems will directly affect the process of the transformation of the edible fungus industry from labor-intensive to technology-intensive. SUMMARY

[0006] In view of the problems in the related art, the present application proposes an edible fungus stick inoculation parameter control method and system based on a spectrophotometer to overcome the above technical problems existing in the prior art.

[0007] To solve the above technical problems, the present application is realized by the following technical scheme:

[0008] The present application is an edible fungus stick inoculation parameter control method based on a spectrophotometer, comprising the following steps:

[0009] S1, construct an anti-interference optical detection sub-module system, and perform mechanical installation, optical calibration, white calibration and standard curve establishment operation on the sub-module system to obtain a calibrated optical detection unit;

[0010] S2, construct and deploy a multi-modal sensor network; the multi-modal sensor network is used to collect environmental data of a culture chamber corresponding to the to-be-tested bacteria liquid to obtain culture chamber environmental parameter data;

[0011] S3, construct a strain quality evaluation model, input the culture chamber environmental parameter data into the evaluation model to predict the strain activity of the to-be-tested bacteria liquid and perform quality determination, and adjust the inoculation process according to the quality determination result to obtain adjusted bacteria;

[0012] S4, using the calibrated optical detection unit to collect the OD value of the adjusted bacterial liquid in real time, and obtaining the real-time OD value of the bacterial liquid; when the real-time OD value of the bacterial liquid deviates from the target value by more than 0.2, starting an emergency mode of quickly adjusting the inoculation amount: closing the inoculation valve and immediately stopping the inoculation process; starting a reverse flushing program to flush the pipeline at twice the current flow rate, and the flushing water volume is twice the pipeline volume; adjusting the subsequent inoculation amount according to the deviation direction of the OD value, i.e., reducing the inoculation amount by 10-15% when the OD value is too high, and increasing the inoculation amount by 10-15% when the OD value is too low; system self-checking and recovery, ultrasonic cleaning of the flow cell, sensor calibration, and starting inoculation; when the real-time OD value of the bacterial liquid deviates from the target value by between 0.1 and 0.2, using a fast adjustment mode: dynamically adjusting the proportional gain (Kp) according to the deviation amplitude, suppressing integral saturation through fuzzy logic, and enhancing differential anti-interference capability; adjusting the inoculation amount in proportion to the deviation, and shortening the inoculation time interval; adjusting the environmental temperature set value according to the deviation of the OD value, i.e., changing the temperature set value by ±0.2°C for every 0.1 change in the OD value; otherwise, enabling PID control.

[0013] Preferably, the sub-module system in S1 adopts a three-beam differential detection architecture; the differential detection architecture includes a main detection beam with a specification of 600 nm, a turbidity compensation beam with a specification of 750 nm, and a reference beam with a specification of 850 nm; the optical unit designed for the main detection beam, the turbidity compensation beam, and the reference beam includes a sapphire window with a transmittance of greater than 98%, an LED light source with a stability of ±0.1% and temperature control, and a low-noise photodiode array with a dark current of less than or equal to 100 pA and a NEP of less than or equal to ;

[0014] The differential detection architecture further includes a flow cell with a Z-shaped flow channel, the inner surface of the flow cell with the Z-shaped flow channel is treated by nanoscale polishing, and is installed at an inclination of 20° and cooperates with an ultrasonic sampling system to generate a micro-vibration effect at a frequency of 40 kHz.

[0015] Preferably, the mechanical installation, optical calibration, white calibration, and standard curve establishment operations of the sub-module system include the following steps:

[0016] S11, connecting the flow cell with the Z-shaped flow channel to the bacterial liquid delivery pipeline by a DN15 sanitary clamp, installing the flow cell at an angle range of 20±1° inclination, and providing a Y-shaped filter to process the inlet and outlet; the flow cell with the Z-shaped flow channel is connected to the vibration motor through a flange, and the vibration amplitude is calibrated to a range of 50±5 μm by a laser displacement sensor;

[0017] S12, turning off the light source, and collecting 1000 samples to obtain an average value as a dark reference;

[0018] S13, sterile physiological saline is introduced, and the gain of the photoelectric amplifier is adjusted so that the output voltage fluctuates up and down by 0.001V around 4.000V;

[0019] S14, a three-order polynomial calibration curve is established using mycelium suspensions with OD values in the range of 0.1-2.5 diluted by gradient dilution.

[0020] Preferably, the deployment in S2 comprises the following steps:

[0021] S21, the multi-modal sensor network comprises three sensor nodes for identifying parameters of the bacterial liquid and monitoring data, monitoring the culture environment, and monitoring the equipment state in real time; the three sensor nodes of the multi-modal sensor network are denoted as a first sensor node, a second sensor node, and a third sensor node, respectively; the first sensor node is installed in a sanitary plug-in manner, a pH electrode in the first sensor node is located at a position 1.5m away from the outlet of the flow cell of the Z-shaped flow channel, and a dissolved oxygen sensor is installed at the highest point of the bacterial liquid delivery pipeline;

[0022] The second sensor node is deployed according to a three-dimensional space grid of the culture chamber, the node height is set in layers, and LoRaWAN networking is adopted.

[0023] Preferably, in S21, the data in the data monitoring include pH, conductivity, and dissolved oxygen; the equipment state includes the pressure of the inoculator and the accuracy of the flow meter; the first sensor node, the second sensor node, and the third sensor node are networked through a TSCH time synchronization protocol, and the network delay is less than 50ms.

[0024] Preferably, the evaluation model in S3 comprises an LSTM neural network model and a method combining convolutional network and multi-sensor; the LSTM neural network mainly consists of an input layer, an LSTM layer and a fully connected layer. The input data of the input layer are time series features (time series change trajectory of OD value of bacterial liquid) and other dynamic parameters (synchronous time series data of environment such as humidity, temperature and CO2 concentration); the LSTM layer consists of two layers, each layer has 64 LSTM units, the LSTM gate unit and the output activation function are Sigmoid and ReLU respectively, and the inter-layer Dropout rate is 0.2; the number of neurons of the fully connected layer is 1 and the activation function is Sigmoid; the loss function is mean square error (MSE). The convolutional neural network adopts 1D-CNN architecture, and the input of the input layer is multi-sensor features (pH, conductivity, dissolved oxygen, equipment parameters, etc.); the convolutional layer has three layers, each layer has a convolution kernel size of 3, a step size of 1 and a ReLU activation function, and the filter numbers of the three layers are 32, 64 and 128 respectively; the number of neurons of the first layer of the fully connected layer is 256, the activation function is ReLU, and the Dropout rate is 0.5; the number of neurons of the second layer is 128 and the activation function is ReLU; the number of neurons of the output layer is 1 and the activation function is sigmoid; the loss function is weighted MSE. The LSTM neural network model is used to analyze the change trajectory of OD value of the strain in the time dimension; the method combining convolutional network and multi-sensor is used to process the topological relationship between data in the spatial dimension; the LSTM neural network model and the convolutional network need to be trained before application, wherein the transfer learning strategy is used to pre-train on a plurality of groups of data collected in a controllable laboratory environment, and the data mainly include bacterial liquid parameters (OD value, pH value, dissolved oxygen, conductivity), environmental parameters (temperature, humidity, CO2 concentration, light intensity), strain activity label and equipment simulation data (inoculation machine parameters, vibration data). Then, the data collected in the actual production environment are used to try fine tuning and verification, and the data mainly include bacterial liquid parameters (real-time OD value, pH value, dissolved oxygen), environmental parameters (humidity, temperature, CO2 concentration), equipment state (inoculation machine pressure, flowmeter accuracy, vibration amplitude) and strain activity label.

[0025] Preferably, the calibrated optical detection unit in S4 is used to collect the OD value of the adjusted bacterial liquid in real time, and the real-time OD value of the bacterial liquid is obtained, including the following process:

[0026] The calibrated optical detection unit completes each detection cycle within 215 milliseconds, wherein three steps of ultrasonic cleaning, data collection and output of correction values need to be completed in one cycle; each detection cycle first starts ultrasonic cleaning for 200 milliseconds, then preheats the LED light source for 5 milliseconds and synchronously collects light intensity data of different wavelengths of 600 nanometers, 750 nanometers and 850 nanometers, and finally performs background correction on the collected data to obtain the real-time OD value of the bacterial solution.

[0027] Preferably, the step of judging whether the OD value deviates from the target value in S4 comprises the following steps:

[0028] S41, a special parameter library is established for different bacteria, and the data in the special parameter library include a target OD value range, a temperature change curve, a viscosity compensation coefficient and the like;

[0029] S42, whether the OD value deviates from the target value is judged according to the special parameter library.

[0030] Preferably, the strategy of the PID control in S4 includes a first-level control strategy and an optimization layer control strategy; the first-level control strategy is to adjust the inoculation amount by using a fuzzy PID algorithm; the optimization layer control strategy is to use model predictive control; wherein the initial parameters of the proportional gain (Kp), the integral time (Ti) and the derivative time (Td) of the PID controller determined by using a dynamic response analysis method are initially adjusted and optimized by using system gain (K), time constant (T), pure lag time (L), critical gain (Ku) and critical period (Tu) and the like; the self-tuning algorithm includes a particle swarm optimization algorithm;

[0031] Wherein, the fuzzy rule base corresponding to the fuzzy PID algorithm used to adjust the inoculation amount in S4 contains 81 empirical rules; the model predictive control solves a multi-objective optimization problem once every 5 minutes, while optimizing the inoculation parameters such as inoculation amount, inoculation pressure, inoculation speed and inoculation time interval, and the environmental setting values such as temperature, humidity, CO2 concentration, illumination and air flow speed; the actuator uses a pneumatic diaphragm pump combined with a high-speed electromagnetic valve.

[0032] The edible mushroom stick inoculation parameter regulation system based on a spectrophotometer includes an optical detection unit construction and deployment calibration module, a sensor network construction and deployment module, a bacterial solution environmental parameter acquisition module, a strain viability prediction module, an inoculation adjustment module, a bacterial solution OD value measurement module and a bacterial solution environmental parameter adjustment module.

[0033] The optical detection unit construction and deployment calibration module constructs an anti-interference optical detection sub-module system, which uses a three-beam differential detection architecture; and performs mechanical installation, optical calibration, white calibration and standard curve establishment operations on the differential detection architecture to obtain a calibrated optical detection unit.

[0034] The sensor network construction and deployment module constructs and deploys a multimodal sensor network; the multimodal sensor network includes three sensing nodes, which are used to identify bacterial liquid parameters and perform data monitoring, monitor the culture environment, and monitor the equipment status in real time.

[0035] The bacterial culture environment parameter acquisition module uses the multimodal sensor network to acquire environmental data of the culture room corresponding to the bacterial culture to be tested, and obtains the culture room environmental parameter data.

[0036] The strain viability prediction module constructs an evaluation model for strain quality, and then inputs the environmental parameter data of the culture room into the evaluation model to predict the strain viability of the test liquid, thereby obtaining edible fungi viability prediction data.

[0037] The inoculation adjustment module performs a quality assessment based on the predicted viability data of the edible fungi, and then adjusts the inoculation process based on the quality assessment results to obtain the adjusted strain.

[0038] The bacterial culture OD value measurement module uses the calibrated optical detection unit to collect the OD value of the bacterial culture corresponding to the adjusted bacterial species in real time, and obtain the real-time OD value of the bacterial culture.

[0039] The bacterial culture environment parameter adjustment module constructs an adaptive control system. In conjunction with the adaptive control system, when the real-time OD value of the bacterial culture deviates from the target value by more than 0.2, an emergency mode for rapid adjustment of the inoculum amount is activated; when the real-time OD value of the bacterial culture deviates from the target value by between 0.1 and 0.2, a rapid adjustment mode is adopted; otherwise, PID control is activated.

[0040] The present invention has the following beneficial effects:

[0041] 1. This invention constructs a closed-loop control system for the inoculation of edible mushroom spawn, integrating "optical detection - data analysis - automatic control." A specially designed optical detection unit acquires the OD value of the spawn solution in real time. Then, data such as temperature, humidity, and CO2 concentration collected by environmental sensors are combined with deep learning algorithms to create a model predicting the viability of laboratory edible mushroom spawn. Finally, a PLC control system dynamically adjusts the environment of the cultivation room and the operating parameters of the inoculation machine. This system can predict the trend of spawn activity changes in advance, allowing for timely adjustment of cultivation parameters, significantly increasing and stabilizing mycelial survival rate. Under an electron microscope, the mycelial cell walls are intact, uniform in thickness, and smooth without damage. The mycelia are cylindrical, covered with a dense nanofiber structure, with uniform branching and transparent growth cones at the tips, indicating vigorous metabolic activity. The intelligent environmental control system matches the power of the air conditioning / humidification equipment according to real-time needs, reducing the overall energy consumption of the cultivation room.

[0042] 2、The present application adjusts the inoculation machine parameters (such as inoculation speed / pressure) through PLC, and cooperates with automatic environmental regulation, so that the single batch inoculation time of the rod is shortened, and the daily average production capacity is improved.

[0043] 3、The present application realizes real-time environmental monitoring and automatic sterilization linkage, and greatly reduces the contamination rate of miscellaneous bacteria compared with the traditional process.

[0044] 4、The present application realizes real-time monitoring of the OD value of the bacterial liquid through the optical detection unit, and combines the deep learning model of multiple environmental parameters (temperature and humidity / CO2), reduces the prediction error, and controls the prediction error within the preset threshold range, which greatly improves the accuracy compared with the traditional manual judgment.

[0045] Of course, implementing any product of the present application does not necessarily need to achieve all the advantages described above. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description only some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0047] Figure 1 The detailed work flow chart of the sensitive optical detection module subsystem of the present application is used to represent how the designed optical unit collects and corrects in real time when detecting the OD value;

[0048] Figure 2 The running flow chart of the time-space feature model of the present application for evaluating the quality of the strain, explains the process of strain activity prediction combined with the design of LSTM neural network and convolution network;

[0049] Figure 3 The adjustment flow chart of the adaptive control system of the present application, which describes in detail the hierarchical control strategy of the control system of the present application in response to environmental factors and mechanism burst situation, ensures the stable operation of the system;

[0050] Figure 4 The logic flow chart of the system abnormal situation response strategy of the present application, which sets forth the detailed response mechanism of the designed control system when encountering bubble interference and abnormal strain activity under real-time monitoring;

[0051] Figure 5 The mycelium state chart under electron microscope in the present application. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments of the application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the protection scope of the application.

[0053] Embodiment one

[0054] The embodiment is a method for regulating edible mushroom stick inoculation parameters based on a spectrophotometer, comprising the following steps:

[0055] 1. Constructing a sub-module system for anti-interference optical detection

[0056] Please refer to Figure 1 This part of the sub-module system ingeniously adopts a three-beam differential detection architecture, which contains a main detection beam with a specification of 600 nm, a turbidity compensation beam of 750 nm and a reference beam of 850 nm. For the above designed beams, the optical unit will be composed of a sapphire window with a transmittance of more than 98%, a LED light source with a stability of ±0.1% and temperature control, and a low-noise photodiode array with a dark current less than or equal to 100 pA and a NEP less than or equal to On the other hand, a Z-shaped flow channel flow cell is designed, the inner surface of the flow cell of the Z-shaped flow channel of the device is treated by nanoscale polishing, and the device is installed at an angle of 20° to effectively prevent mycelium deposition. Most importantly, an ultrasonic sampling system is used as an auxiliary function to generate a micro-vibration effect at a frequency of 40 kHz to ensure that the fungus liquid passes through the detection area uniformly.

[0057] 2. Multi-modal sensor network

[0058] Three sensor nodes are designed to play a role in this sensor network part. The first one is a sensor group that identifies fungus liquid parameters and monitors data, including pH, conductivity and dissolved oxygen, with a sampling frequency of 1 Hz. The second one is the design core, which sets up monitoring nodes for the culture environment, and the fluctuation ranges of the main environmental factors are set as temperature ±0.1℃, relative humidity ±1%, CO2±20ppm, light intensity ±50 lux, etc. The third one is for real-time control of the device state, which monitors the accuracy of the pressure and flow meter of the inoculator in real time by setting sensors. All nodes are networked through the TSCH time synchronization protocol, with a network delay of <50ms.

[0059] 3. Constructing an evaluation model for strain quality

[0060] Please refer to Figure 2, first in the time dimension, through the constructed LSTM neural network model to analyze the change trajectory of the strain OD value. The LSTM neural network mainly consists of an input layer, an LSTM layer and a fully connected layer. The input data of the input layer is time series features (time series change trajectory of the OD value of the bacterial solution) and other dynamic parameters (synchronous time series data of the environment such as humidity, temperature, CO2 concentration, etc.); the LSTM layer is composed of two layers, each layer has 64 LSTM units, the LSTM gate unit and the output activation function are Sigmoid and ReLU respectively, and the inter-layer Dropout rate is 0.2; the number of neurons in the fully connected layer is 1 and the activation function is Sigmoid; the loss function is mean square error (MSE). Then in the spatial dimension, the method of combining convolutional network and multi-sensor is used to process the topological relationship between the data. The convolutional neural network adopts 1D-CNN architecture, the input of the input layer is multi-sensor features (pH, conductivity, dissolved oxygen, equipment parameters, etc.); the convolutional layer has three layers, the convolution kernel size of each layer is 3, the step is 1, and the activation function is ReLU, the filter number of the three layers is 32, 64 and 128 respectively; the first layer of the fully connected layer has 256 neurons, the activation function is ReLU, and the Dropout rate is 0.5; the second layer has 128 neurons and the activation function is ReLU; the number of neurons in the output layer is 1 and the activation function is sigmoid; the loss function is weighted MSE. Before application, model training is required, and the transfer learning strategy is tried, first pre-training on a plurality of data collected in the laboratory controllable environment, the data mainly include bacterial solution parameters (OD value, pH value, dissolved oxygen, conductivity), environmental parameters (temperature, humidity, CO2 concentration, light intensity), strain activity label, equipment simulation data (inoculation machine parameters, vibration data). Then try to fine-tune and verify the data collected in the actual production environment, which mainly includes bacterial solution parameters (real-time OD value, pH value, dissolved oxygen), environmental parameters (humidity, temperature, CO2 concentration), equipment status (inoculation machine pressure, flowmeter accuracy, vibration amplitude), strain activity label.

[0061] 4. Adaptive control system

[0062] The design adopts a two-level control strategy, the first level control strategy adopts a fuzzy PID algorithm to adjust the inoculation amount, and its fuzzy rule base contains 81 empirical rules, such as “if OD is high and activity is low, then reduce the flow rate greatly”; the second level is the optimization layer control strategy, which adopts model predictive control, solves a multi-objective optimization problem once every 5 minutes, and optimizes the inoculation parameters and environmental set values at the same time. The actuator adopts a pneumatic diaphragm pump (flow regulation ratio 100:1) combined with a high-speed electromagnetic valve (response time <10 ms) to ensure that the control accuracy reaches ±1 mL / rod.

[0063] The specific operation mode is as follows:

[0064] 1. Hardware deployment and calibration of monitoring control system

[0065] Mechanical installation of optical detection unit of monitoring system: the designed spectrophotometric flow cell is connected to the bacteria liquid conveying pipeline with DN15 sanitary clamp, and is strictly inclined at an angle range of 20±1°, and a Y-type filter is provided to process the inlet and outlet. The flow cell is connected to the vibration motor through a flange, and the vibration amplitude is calibrated in the range of 50±5 μm by a laser displacement sensor.

[0066] Optical calibration: turn off the light source, collect 1000 samples and take the average value as the dark reference;

[0067] White calibration: pass sterile normal saline, adjust the photoelectric amplifier gain to force the output voltage to be 4.000 V with an up and down fluctuation of 0.001 V;

[0068] Standard curve establishment: a three-degree polynomial calibration curve is established with gradient diluted mycelium suspension with OD value in the range of 0.1-2.5.

[0069] 2. Sensor network deployment

[0070] Bacteria liquid parameter sensor group: sanitary plug-in installation is adopted, the pH electrode is 1.5 m away from the outlet of the flow cell, and the dissolved oxygen sensor (0-20 mg / L, ±0.1 mg / L) is installed at the highest point of the pipeline.

[0071] Environmental monitoring node: according to the three-dimensional space grid deployment of the culture room, one monitoring point is arranged every 8 m³, the node height is layered (0.5 m, 1.5 m, 2.5 m), and LoRaWAN networking (SF7, 125 kHz bandwidth) is adopted.

[0072] The hardware basis for implementing the technology includes: customized online spectrophotometric detection unit (measurement range OD: 0-2.5, resolution 0.001), industrial-grade PLC controller (Siemens S7-1200 series), Internet of Things sensor nodes (temperature and humidity, CO2, illuminance, etc.), and actuators (metering pump, electric regulating valve, etc.). The software system is composed of lower computer data acquisition program and upper computer analysis platform, the development language adopts mixed programming of C# and Python, and the machine learning framework uses TensorFlow Lite to adapt to the needs of edge computing.

[0073] 3. System initialization setting

[0074] (1) Process parameter configuration

[0075] A dedicated parameter library is established for different strains, including target OD value range, temperature change curve, viscosity compensation coefficient and other key parameters. For example, the target OD value of Pleurotus ostreatus strain is set to 0.9-1.1, the initial temperature of inoculation is maintained at 24°C, and the temperature is adjusted to 22.5°C after 48 hours.

[0076] (2) Control parameter setting

[0077] In theory, the initial parameters of the PID controller are determined by dynamic response analysis method. The proportional coefficient of the process time constant 1.2 minutes and the pure lag time 0.3 minutes is 2.8, the integral time is 1.5 minutes, and the differential time is 0.1 minute. However, in actual operation, these parameters need to be continuously optimized and need to be solved by self-tuning algorithm.

[0078] 4. System operation process

[0079] (1) Real-time detection cycle

[0080] Each detection cycle needs to be completed within 215 milliseconds, and three steps of ultrasonic cleaning, data collection and output correction value need to be completed in one cycle. Each detection cycle first starts the ultrasonic cleaning for 200 milliseconds, then preheats the LED light source for 5 milliseconds and synchronously collects the light intensity data of different wavelengths of 600 nanometers, 750 nanometers and 850 nanometers, and finally corrects the collected data and obtains the output OD value.

[0081] (2) Dynamic control strategy

[0082] Please refer to Figure 3 , in order to optimize the control of the culture room environment parameters, a closed-loop control system is formed which can automatically adjust the related parameters according to the change of OD value. At the same time, the system also forms a three-level response mechanism: when the OD value deviates from the target value by more than 0.2, the emergency mode of quickly adjusting the inoculation amount is started; when the deviation value is between 0.1-0.2, the fast adjustment mode is used; when there is a small range of deviation, PID control is used.

[0083] 5. Maintenance and verification

[0084] (1) Daily maintenance procedures

[0085] The system automatically performs periodic maintenance every 8 hours by flushing the pipeline with 0.5% NaOH aqueous solution for 3 minutes and calibrating the sensor every week. Similarly, the optical assembly needs to be checked for transmittance every month, and a maintenance alarm is triggered when the transmittance attenuation exceeds 5%.

[0086] (2) Performance verification method

[0087] In the system accuracy verification, five groups of standard bacteria liquid samples between 0.5-1.5 OD range need to be prepared to ensure the reliability and accuracy of the data and 30 repeated measurements are performed on each group of standard bacteria liquid samples. If the correlation coefficient of the verification result and the laboratory reference method is as high as 0.982, the measurement repeatability is better than 1.5%, and the whole process delay from detection to execution control is less than 500 milliseconds, it is considered to meet the standard.

[0088] 6. Implemented device configuration

[0089] The system core components include a spectrophotometric detection module (spectrophotometric detection measurement range is 0-2.5 OD, pressure resistance is 0.8 MPa), a precision metering pump (flow regulation range is 5-500 milliliters / minute), and an environmental control cabinet (32 independent PID control channels).

[0090] 7. Exception handling, please refer to Figure 4

[0091] (1) Bubble interference processing

[0092] The system will automatically close the inoculation valve and start the reverse flushing program when the detected signal mutation exceeds 10%, and the time of the reverse flushing program is calculated according to twice the volume of the pipeline divided by the current flow rate.

[0093] (2) Strain viability abnormality processing

[0094] Different measures need to be taken according to different production stages when the strain viability index decreases by more than 30% per hour. The standby strain is automatically switched during the middle of inoculation, and the inoculation parameters are adjusted. At the end of inoculation, 3 liters of disinfectant flushing program is started to ensure pipeline cleaning.

[0095] The system for precise control of edible mushroom inoculation process can flexibly adjust parameters according to the requirements of production process, not only has the ability to adapt to different strain characteristics, but also provides reliable technical support for industrialized production of edible mushrooms.

[0096] During implementation, the detection unit needs to be connected in series to the bacteria liquid delivery pipeline to ensure that the inlet and outlet of the flow cell are connected and installed in airtight connection with DN15 sterile silicone tubes. Standard turbidity liquid is used for three-point calibration (0, 1.0, 2.0 OD), and the calibration process automatically completes the calculation and storage of the photoelectric conversion coefficient. Each environmental sensor is arranged with a monitoring node at 20 square meters according to the volume of the culture room, and is evenly distributed. After the hardware installation is completed, the full link delay from data acquisition to the response of the actuator should be less than the set threshold value for verification to perform system commissioning test.

[0097] The system requires the bacteria solution to run through the quartz flow cell at a flow rate of 0.5-1.0 m / s, and the spectrophotometric module to collect light intensity data at a frequency of 10 Hz. The light source emits 600 nm and 750 nm light pulses in each cycle, and the signal received by the photodiode is transmitted to the processor after being converted by a 24-bit ADC. The data processing unit calculates the absorbance at each wavelength according to the formula OD=log 10 (I0 / I) and performs a sliding average filter on the raw signal to eliminate random noise. Finally, the corrected bacteria solution OD value is obtained by subtracting OD750 from OD600. The system usually uses an outlier rejection algorithm to eliminate bubble interference, where the outlier rejection algorithm refers to triggering a self-cleaning program when the fluctuation amplitude of three consecutive sampling points exceeds 5%.

[0098] The strain viability prediction model uses a ReLU activation function and Dropout regularization, and includes a 1D-CNN architecture with 3 convolutional layers and 2 fully connected layers. The model input includes a 14-dimensional feature vector containing the current OD value and its past 10-minute change rate, environmental temperature, and bacteria solution pH value, and the model output is a standardized viability index between 0 and 1 as a reference. If the OD measurement data and environmental parameters collected by the sensor are input into the pre-trained strain viability prediction model, and the output standardized viability index is less than 0.7, the system will automatically stop inoculation and issue a warning and determine that the strain viability is insufficient.

[0099] PID algorithm is used to control the inoculation concentration, and the change in bacteria solution delivery volume controller adjusts the speed of the metering pump according to the deviation between the current OD value and the target set value. In the PID control algorithm, a feedforward compensation link is particularly introduced to automatically adjust the target parameters according to the inoculation concentration environmental control subsystem. When the OD value is detected to rise rapidly (ΔOD / Δt>0.05 / min), the valve opening is reduced in advance to offset the inertial effect of the increase in bacteria solution concentration. When the OD value is higher than 1.2, the culture room temperature set value needs to be increased by 0.5-1℃ to match the higher metabolic demand;

[0100] The system of the present application particularly designs a process expert library function that can store the optimal growth parameter curves of different strains, and the key process parameters need to be optimized according to the specific application scenario. For most wood-decaying fungi, it is recommended to control the inoculation OD value in the range of 0.8-1.2, and to apply 1% NaOH aqueous solution to the flow cell for continuous circulation cleaning once every 8 hours of production, and the neural network model suggests incremental training once every 500 batches of data accumulated to maintain prediction accuracy.

[0101] In the present application, a closed-loop control system of "optical detection-data analysis-automatic regulation" is constructed during the inoculation process of edible fungus sticks; the OD value of the fungus liquid is obtained in real time by a specially designed optical detection unit. Then, the temperature, humidity, CO2 concentration and other data collected by the environmental sensor are combined with the deep learning algorithm to establish a model for predicting the activity of the edible fungus strain in the laboratory. Finally, the PLC control system is used to dynamically adjust the culture room environment and the working parameters of the inoculation machine; it can predict the trend of strain activity in advance, and then adjust the culture parameters in time, so that the mycelium survival rate is greatly increased and stable, as shown in Figure 5 As shown, under electron microscope, the mycelium cell wall is complete, uniform in thickness, smooth surface without damage; the mycelium is cylindrical, covered with fine nanoscale fibrous structure on the surface, uniform branching, transparent growth cone at the tip, showing vigorous metabolic activity; the intelligent environmental regulation system matches the power of air conditioner / humidifying equipment according to real-time demand, reducing the comprehensive energy consumption of the culture room.

[0102] Example two

[0103] Please refer to Figure 4 The present embodiment discloses an edible fungus stick inoculation parameter regulation system based on spectrophotometer, which can realize the method of the above-mentioned embodiments, including optical detection unit construction deployment calibration module, sensor network construction deployment module, fungus liquid environmental parameter acquisition module, strain activity prediction module, inoculation adjustment module, fungus liquid OD value measurement module and fungus liquid environmental parameter adjustment module.

[0104] The optical detection unit construction deployment calibration module constructs an anti-interference optical detection sub-module system, which adopts a three-beam differential detection architecture; and performs mechanical installation, optical calibration, white calibration and standard curve establishment operation on the differential detection architecture, to obtain the calibrated optical detection unit.

[0105] The sensor network construction deployment module constructs and deploys a multi-modal sensor network; the multi-modal sensor network includes three sensor nodes for identifying fungus liquid parameters and monitoring data, monitoring culture environment and monitoring equipment status in real time;

[0106] The fungus liquid environmental parameter acquisition module acquires the environmental data of the culture room corresponding to the fungus liquid to be tested by the multi-modal sensor network, to obtain the culture room environmental parameter data.

[0107] The strain activity prediction module predicts the strain activity of the fungus liquid to be tested by constructing a strain quality evaluation model and inputting the culture room environmental parameter data into the evaluation model, to obtain edible fungus activity prediction data.

[0108] The inoculation adjustment module adjusts the inoculation process according to the quality judgment result of the edible mushroom vitality prediction data, and obtains the adjusted strain;

[0109] The bacteria liquid OD value measurement module adopts the calibrated optical detection unit to collect the OD value of the bacteria liquid corresponding to the adjusted strain in real time, and obtains the real-time OD value of the bacteria liquid;

[0110] The bacteria liquid environment parameter adjustment module constructs an adaptive control system; in cooperation with the adaptive control system, when the real-time OD value of the bacteria liquid deviates from the target value by more than 0.2, an emergency mode of quickly adjusting the inoculation amount is started; when the real-time OD value of the bacteria liquid deviates from the target value by between 0.1 and 0.2, a fast adjustment mode is adopted; otherwise, a PID control is enabled.

[0111] In the description of the present specification, the description of the terms "one embodiment", "example", "specific example" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are contained in at least one embodiment or example of the invention. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0112] The preferred embodiments of the above disclosed invention are only used to help explain the invention. The preferred embodiments do not describe all the details and limit the invention to the specific embodiments described. Obviously, many modifications and changes can be made according to the content of the present specification. The present specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can well understand and utilize the invention.

Claims

1. A method for adjusting inoculation parameters of edible fungi substrate based on a spectrophotometer, characterized in that, Includes the following steps: S1. Construct a sub-module system for anti-interference optical detection, and perform mechanical installation, optical calibration, white calibration, and standard curve establishment operations on the sub-module system to obtain a calibrated optical detection unit; S2. Construct and deploy a multimodal sensor network; use the multimodal sensor network to collect environmental data of the culture chamber corresponding to the bacterial solution to be tested, and obtain the environmental parameter data of the culture chamber; S3. By constructing an evaluation model for strain quality, the environmental parameter data of the culture room are input into the evaluation model to predict the strain viability of the test bacterial solution and make a quality judgment. Then, the inoculation process is adjusted according to the quality judgment result to obtain the adjusted strain. S4. The OD value of the bacterial solution corresponding to the adjusted bacterial strain is collected in real time using the calibrated optical detection unit to obtain the real-time OD value of the bacterial solution. When the real-time OD value of the bacterial culture deviates from the target value by more than 0.2, the emergency mode for rapid adjustment of the inoculum amount is activated; when the real-time OD value of the bacterial culture deviates from the target value by between 0.1 and 0.2, the rapid adjustment mode is adopted; otherwise, PID control is activated.

2. The method for adjusting inoculation parameters of edible fungi substrate based on a spectrophotometer according to claim 1, characterized in that: The submodule system described in S1 adopts a three-beam differential detection architecture. This architecture includes a 600nm main detection beam, a 750nm turbidity compensation beam, and an 850nm reference beam. The optical units used for designing the main detection beam, turbidity compensation beam, and reference beam include a sapphire window with a transmittance greater than 98%, an LED light source with a stability of ±0.1% and temperature control, and a dark current less than or equal to 100pA and a NEP less than or equal to... Low-noise photodiode array; The differential detection architecture also includes a Z-shaped flow cell. The inner surface of the Z-shaped flow cell is treated with nano-level polishing and combined with a 20° tilt installation and an ultrasonic sampling system to generate a micro-amplitude vibration effect at a frequency of 40kHz.

3. The method for adjusting inoculation parameters of edible fungi substrate based on a spectrophotometer according to claim 2, characterized in that, The mechanical installation, optical calibration, white calibration, and standard curve establishment operations for the submodule system include the following steps: S11. The Z-shaped flow channel flow tank is connected to the bacterial liquid delivery pipeline using DN15 sanitary clamps, with an installation angle range of 20±1° inclined, and equipped with a Y-type filter to treat the inlet and outlet; the Z-shaped flow channel flow tank is connected to the vibration motor through a flange, and the vibration amplitude is calibrated within the range of 50±5μm by a laser displacement sensor. S12. Turn off the light source and collect 1000 samples, taking the average value as the dark baseline; S13. Introduce sterile saline solution and adjust the gain of the photoelectric amplifier so that the output voltage fluctuates around 4.000V by 0.001V. S14. Establish a cubic polynomial calibration curve using serially diluted mycelial suspensions with OD values ​​ranging from 0.1 to 2.

5.

4. The method for adjusting inoculation parameters of edible fungi substrate based on a spectrophotometer according to claim 1, characterized in that, Deploying in S2 involves the following steps: S21. The multimodal sensor network includes three sensing nodes, which are used to identify bacterial liquid parameters and perform data monitoring, monitor the culture environment, and monitor the equipment status in real time, respectively. The three sensing nodes of the multimodal sensor network are referred to as the first sensing node, the second sensing node, and the third sensing node, respectively. The first sensing node adopts a sanitary insertion type installation. The pH electrode in the first sensing node is located 1.5m away from the flow tank outlet of the Z-shaped flow channel, and the dissolved oxygen sensor is installed at the highest point of the bacterial liquid delivery pipeline. The second sensing node is deployed according to the three-dimensional spatial grid of the culture chamber, with the node height set in layers and LoRaWAN networking.

5. The method for adjusting inoculation parameters of edible fungi substrate based on a spectrophotometer according to claim 4, characterized in that: The data monitored in S21 includes pH, conductivity, and dissolved oxygen; the equipment status includes inoculation machine pressure and flow meter accuracy; the first sensor node, the second sensor node, and the third sensor node are networked via the TSCH time synchronization protocol, wherein the network latency is <50ms.

6. The method for adjusting inoculation parameters of edible fungi substrate based on a spectrophotometer according to claim 5, characterized in that: The evaluation model described in S3 includes an LSTM neural network model and a method combining convolutional networks with multiple sensors. The LSTM neural network model is used to analyze the change trajectory of bacterial OD values ​​in the time dimension. The method combining convolutional networks with multiple sensors is used to process the topological relationships between data in the spatial dimension. The LSTM neural network model and the convolutional network need to be trained before application. Specifically, by using a transfer learning strategy, the model is first pre-trained on data collected in multiple controlled laboratory environments, and then fine-tuned and verified using data collected in the actual production environment.

7. The method for adjusting inoculation parameters of edible fungi substrate based on a spectrophotometer according to claim 6, characterized in that, In S4, the calibrated optical detection unit is used to collect the OD value of the bacterial solution corresponding to the adjusted bacterial strain in real time. The process of obtaining the real-time OD value of the bacterial solution includes the following steps: The calibrated optical detection unit completes each detection cycle within 215 milliseconds. Each cycle requires three steps: ultrasonic cleaning, data acquisition, and output correction values. Each detection cycle first starts with ultrasonic cleaning that lasts for 200 milliseconds, then preheats the LED light source for 5 milliseconds and simultaneously acquires light intensity data at different wavelengths of 600 nm, 750 nm, and 850 nm. Finally, the collected data are background corrected to obtain the real-time OD value of the bacterial solution.

8. The method for adjusting inoculation parameters of edible fungi substrate based on a spectrophotometer according to claim 7, characterized in that, The steps to determine whether the OD value deviates from the target value in S4 include: S41. Establish a dedicated parameter library for different bacterial strains. The data in the dedicated parameter library includes the target OD value range, temperature change curve, and viscosity compensation coefficient. S42. Determine whether the OD value deviates from the target value based on the dedicated parameter library.

9. The method for adjusting inoculation parameters of edible fungi substrate based on a spectrophotometer according to claim 8, characterized in that: The PID control strategy described in S4 includes a first-level control strategy and an optimization-level control strategy. The first-level control strategy uses a fuzzy PID algorithm to adjust the seeding rate. The optimization-level control strategy uses model predictive control. A self-tuning algorithm is used to initially adjust and optimize the parameters in the process of determining the initial parameters of the PID controller using dynamic response analysis. The self-tuning algorithm includes a particle swarm optimization algorithm. The initial parameters include proportional gain, integral time, and derivative time. In S4, a fuzzy PID algorithm is used to adjust the fuzzy rule base corresponding to the inoculation amount, which contains 81 empirical rules; the model predictive control solves a multi-objective optimization problem every 5 minutes, while simultaneously optimizing the inoculation parameters and environmental settings; the actuator uses a pneumatic diaphragm pump in conjunction with a high-speed solenoid valve.

10. A system for implementing the method for adjusting inoculation parameters of edible fungi substrate based on a spectrophotometer as described in any one of claims 1-9.

Images