Online measurement and control device for automatic production of Liupu tea

Abstract

The invention discloses an on-line measurement and control device for automatic production of Liupao tea, the device comprises a sensor array support, a plurality of multispectral sensors, an edge calculation module and a ventilation and humidification module, the multispectral sensors are fixedly distributed on the sensor array support in a grid shape, the multispectral sensors are electrically connected with the edge calculation module, and the ventilation and humidification module is electrically connected with the edge calculation module. The ventilation and humidification module is electrically connected with the edge calculation module. The multi-spectral sensors are arranged in a gridding mode, the environments of different point positions in the pile fermentation are accurately monitored through the multi-spectral sensors, the environment parameter adjustment amount is calculated through the edge calculation module, the internal detection environment of the pile fermentation is adjusted in cooperation with the ventilation and humidification module, and the process control precision and uniformity are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic production calibration of tea, and particularly to an on-line measurement and control device for the automatic production of Liubao tea. Background Art

[0002] As a representative of traditional Chinese dark tea, the processing of Liubao tea includes steps such as green killing, rolling, piling fermentation, and drying. Its unique piling fermentation process is extremely sensitive to environmental parameters such as temperature, humidity, and oxygen content. Parameters such as temperature, humidity, and time in each step need to be precisely controlled. Currently, the piling fermentation technology has the following deficiencies:

[0003] 1. Insufficient parameter control accuracy: During the piling fermentation process, the microbial activity and polyphenol oxidase reaction are synergistically affected by temperature (T), relative humidity (RH), and oxygen concentration (O2%). Existing technologies mostly use single-point sensors to monitor surface parameters and cannot penetrate the pile body to detect the internal gradient distribution, resulting in poor fermentation uniformity and significant fluctuations in the theaflavin content of the finished tea.

[0004] 2. Lack of closed-loop feedback mechanism: Most devices only have data acquisition functions (such as temperature and humidity recording), and do not form a "perception - decision - execution" closed loop. Key parameters still need to be detected offline, with poor real-time performance and frequent manual intervention. It relies on manual experience to judge temperature and humidity changes, and the control of microbial activity is unstable. Summary of the Invention

[0005] The purpose of the present invention is to: in view of the above problems, provide an on-line measurement and control device for the automatic production of Liubao tea, which arranges multiple multispectral sensors in a grid pattern, uses the multiple multispectral sensors to accurately monitor the environment at different points inside the pile, calculates the adjustment amount of environmental parameters through an edge computing module, and cooperates with a ventilation and humidification module to adjust the detected environment inside the pile, improving the process control accuracy and uniformity.

[0006] In order to achieve the above invention purpose, the technical solution adopted by the present invention is as follows:

[0007] According to one aspect of the present invention, there is provided an on-line measurement and control device for the automatic production of Liubao tea, including a sensor array bracket, multispectral sensors, an edge computing module, and a ventilation and humidification module;

[0008] A plurality of the multispectral sensors are fixedly distributed in a grid pattern on the sensor array bracket;

[0009] The multispectral sensors are electrically connected to the edge computing module;

[0010] The ventilation and humidification module is connected to the sensor array bracket, and the ventilation and humidification module is electrically connected to the edge computing module.

[0011] Preferably, the sensor array bracket includes a base, a riser pipe, a lead screw, a rotating motor, a moving block, and a grid fixing pipe rack;

[0012] The riser pipe is fixedly arranged on the base, the lead screw is arranged in the riser pipe, the upper end of the lead screw is rotatably connected to the riser pipe, the lower end of the lead screw is fixedly connected to the rotating motor, the moving block is threadedly connected to the lead screw, a sliding hole is formed in the riser pipe, the grid fixing pipe rack passes through the sliding hole and is fixedly connected to the moving block, the grid fixing pipe rack is formed by connecting a plurality of branch pipes to form a grid-like structure, a support pipe is arranged at the grid intersection point of the grid fixing pipe rack, the support pipe is communicated with the grid fixing pipe rack, and the multispectral sensor is fixedly arranged on the support pipe.

[0013] Preferably, the wavelength bands of the multispectral sensor are 1550 nm wavelength band, 940 nm wavelength band, and 3.5 - 4 μm wavelength band.

[0014] Preferably, the edge computing module includes a polyphenol oxidase activity calculation unit and an environmental disturbance prediction unit;

[0015] The polyphenol oxidase activity calculation unit includes the following method steps:

[0016] Receive the temperature, humidity, and oxygen concentration parameter values at the current position of the multispectral sensor;

[0017] Establish a coupling model, and use the coupling model to calculate the polyphenol oxidase activity value;

[0018] Compare the calculated polyphenol oxidase activity value with a preset threshold. If the polyphenol oxidase activity value deviates from the preset threshold, generate the standard values of temperature, humidity, and oxygen concentration;

[0019] The environmental disturbance prediction unit includes the following method steps:

[0020] Collect the temperature, humidity, and oxygen concentration data of a continuous time series;

[0021] Input the data into a pre-trained LSTM neural network, and the LSTM neural network includes a dual-channel input structure and an attention fusion layer;

[0022] Generate a feedforward compensation signal according to the prediction deviation output by the LSTM neural network.

[0023] Preferably, the coupling model is represented by the following formula:

[0024]

[0025] Where, A tis the enzyme activity index at time t; X1 = T (temperature); X2 = RH (humidity); X3 = O2 (oxygen concentration); K i is the half-saturation constant; n is the Hill coefficient; λ is the attenuation coefficient; is the power term of environmental parameters.

[0026] Preferably, it further includes a cross-term correction term for the temperature-humidity synergistic effect, and the cross-term correction term is introduced into the coupling model. The cross-term correction term is expressed by the following formula:

[0027] Cross-term correction term = 1 + α·(T - T ref )·(RH - RH ref )

[0028] where α is the synergistic effect coefficient; T ref = 35°C is the reference temperature; RH ref = 80% is the reference humidity.

[0029] Preferably, the dual-channel input structure of the LSTM neural network includes a main channel and an auxiliary channel. The main channel is used to process the temperature, humidity, and oxygen concentration data of the original time series; the auxiliary channel is used to process the external weather forecast data; the attention fusion layer uses the attention mechanism to dynamically allocate weights.

[0030] Preferably, the ventilation and humidification module includes a PID control system, a variable-frequency ventilation component, an atomizing humidification component, an air duct, a first solenoid valve, a main pipe, an atomizing pipe, and a second solenoid valve;

[0031] The PID control system is electrically connected to the edge computing module. The variable-frequency ventilation component, the atomizing humidification component, the first solenoid valve, and the second solenoid valve are electrically connected to the PID control system. The variable-frequency ventilation component is connected to the air duct. The air duct is connected to the main pipe through the first solenoid valve. The atomizing humidification component is connected to the atomizing pipe. The atomizing pipe is connected to the main pipe through the second solenoid valve. The main pipe is communicated with the grid fixed pipe rack, and a plurality of through holes are opened on the support pipe.

[0032] Preferably, the variable-frequency ventilation component includes a frequency converter and a fan motor. The frequency converter is electrically connected to the PID control system. The frequency converter is electrically connected to the fan motor. The fan motor is connected to the air duct.

[0033] Preferably, the atomizing humidification component includes a water tank, a water pump, and a high-frequency ultrasonic atomizer. The water tank is connected to the water pump. The water pump is connected to the high-frequency ultrasonic atomizer. The ultrasonic atomizer is connected to the atomizing pipe.

[0034] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are as follows:

[0035] The present invention arranges multiple multispectral sensors in a grid pattern, uses the multiple multispectral sensors to accurately monitor the environment at different positions inside the piling, calculates the adjustment amount of environmental parameters through the edge computing module, and cooperates with the ventilation and humidification module to adjust the detection environment inside the piling, improving the process control accuracy and uniformity. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 is a schematic structural diagram of the sensor array bracket of the present invention;

[0037] Figure 2 is a schematic structural diagram of the grid fixing pipe rack of the present invention.

[0038] In the drawings, 1, base; 2, multispectral sensor; 3, riser pipe; 4, screw rod; 5, rotating motor; 6, moving block; 7, grid fixing pipe rack; 8, sliding hole; 9, branch pipe; 10, support pipe. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the following preferred embodiments are cited with reference to the accompanying drawings for further detailed description of the present invention. However, it should be noted that many details listed in the specification are only for enabling the reader to have a thorough understanding of one or more aspects of the invention, and these aspects of the present invention can be implemented even without these specific details.

[0040] Please refer to Figures 1 to 2 , the present invention provides an on-line measurement and control device for the automated production of Liubao tea, and the technical solution is as follows:

[0041] An on-line measurement and control device for the automated production of Liubao tea, comprising a sensor array bracket, a multispectral sensor 2, an edge computing module and a ventilation and humidification module. The sensor array bracket includes a base 1, a riser 3, a lead screw 4, a rotating motor 5, a moving block 6 and a grid fixing pipe rack 7. The riser 3 is fixedly arranged on the base 1, the lead screw 4 is arranged in the riser 3, the upper end of the lead screw 4 is rotatably connected to the inner side wall of the upper end of the riser 3, and the lower end of the lead screw 4 is fixedly connected to the output end of the rotating motor 5. The rotating motor 5 is fixedly arranged on the base 1. The moving block 6 is arranged on the lead screw 4, and the moving block 6 is threadedly connected to the lead screw 4. A sliding hole 8 is formed in the riser 3, and the sliding hole 8 is arranged along the axial direction of the riser 3. In this embodiment, 4 sliding holes 8 are arranged, and the 4 sliding holes 8 are evenly spaced along the circumferential direction of the riser 3. The grid fixing pipe rack 7 includes a plurality of branch pipes 9, and the plurality of branch pipes 9 are staggered and communicated and extend radially outwards to form a grid-like structure. The branch pipes 9 inside the grid fixing pipe rack 7 pass through the sliding holes 8 and are fixedly connected to the moving block 6. A support pipe 10 is arranged at the intersection of the grid-like structure, and the lower end of the support pipe 10 is communicated with the branch pipe 9 at the intersection. The multispectral sensor 2 is arranged on the support pipe 10 to form a grid-type distributed multispectral sensor array. The grid-type distributed multispectral sensor array can be used to detect multiple horizontal points of the pile fermentation, improving the accuracy of the detection data. And by driving the lead screw 4 to rotate through the rotating motor 5, the rotation of the lead screw 4 causes the moving block 6 to move up and down, and the moving block 6 drives the grid fixing pipe rack 7 and the sensor array to move together, so as to detect the vertical points of the pile fermentation. By accurately detecting the environment at different points, the small-range environment at different points can be adjusted.

[0042] The multispectral sensor is used to detect data of temperature, humidity and oxygen concentration. In this embodiment, the bands of the multispectral sensor adopt the 1550nm band, the 760nm band and the 3.5-4μm band. The 1550nm band is used to detect the moisture content; the 760nm band is used to detect the oxygen content; the 3.5-4μm band is used to detect the temperature. The temperature, humidity and oxygen concentration are monitored by the multispectral sensor to obtain the monitoring data, and the monitoring data is transmitted to the edge computing module for the edge computing module to process the data.

[0043] The edge computing module includes a polyphenol oxidase activity calculation unit and an environmental disturbance prediction unit. The polyphenol oxidase activity calculation unit is used to calculate the polyphenol oxidase activity detection value at the detected position point according to the temperature, humidity and oxygen concentration values at the detected position point. By comparing the polyphenol oxidase activity detection value with a preset threshold, if the polyphenol oxidase activity detection value deviates from the preset threshold, the corresponding standard temperature, standard humidity and standard oxygen concentration values are generated and fed back to the ventilation and humidification module, and the temperature, humidity and oxygen concentration are adjusted by the ventilation and humidification module.

[0044] Specifically, the polyphenol oxidase activity calculation unit includes the following method steps:

[0045] 1. Receive the parameter values of temperature, humidity, and oxygen concentration at the current position of the multispectral sensor;

[0046] 2. Establish a coupling model and use the coupling model to calculate the polyphenol oxidase activity value:

[0047] The coupling model is represented by the following formula:

[0048]

[0049] where, A t is the enzyme activity index at time t; X1 = T (temperature); X2 = RH (humidity); X3 = O2 (oxygen concentration); K i is the half-saturation constant; n is the Hill coefficient; λ is the attenuation coefficient; is the environmental parameter power term.

[0050] 3. Compare the calculated polyphenol oxidase activity value with a preset threshold. If the polyphenol oxidase activity value deviates from the preset threshold, generate the standard values of temperature, humidity, and oxygen concentration.

[0051] The coupling model correlates temperature (T), humidity (RH), and oxygen concentration (O2) with polyphenol oxidase activity (A) through a dynamic equation, revealing the non-linear interaction effects among the parameters. By presetting the standard threshold of polyphenol oxidase activity, according to the comparison result between the detected value calculated in real time and the preset value, the standard environmental parameters (temperature, humidity, and oxygen concentration values) can be fed back, and the environmental adjustment component (ventilation and humidification module) can be called to correct the environmental parameters.

[0052] Furthermore, for parameter self-learning and environmental compensation mechanism, significantly enhancing the robustness of the model to complex working conditions, in this embodiment, it also includes a temperature and humidity synergistic effect cross-term correction term, which is introduced into the coupling model to correct the coupling model. The cross-term correction term is represented by the following formula:

[0053] Cross-term correction term = 1 + α · (T - T ref ) · (RH - RH ref )

[0054] where, α is the synergistic effect coefficient; T ref = 35 °C is the reference temperature; RH ref = 80% is the reference humidity.

[0055] Embed the correction term into the integral term of the original equation in the form of a multiplicative factor:

[0056]

[0057] The correction mechanism is as follows: when T > T ref and RH > RH ref the correction term value > 1, amplifying the effect of enzyme activity improvement; when any parameter is lower than the reference value, the correction term value < 1, inhibiting the growth of enzyme activity to prevent overcompensation.

[0058] The environmental disturbance prediction unit includes the following method steps:

[0059] 1. Collect temperature, humidity, and oxygen concentration data in a continuous time series.

[0060] 2. Input the data into a pre-trained LSTM neural network, which includes a dual-channel input structure and an attention fusion layer. The dual-channel input structure of the LSTM neural network includes a main channel and an auxiliary channel. The main channel is used to process the temperature, humidity, and oxygen concentration data of the original time series. The auxiliary channel is used to process external weather forecast data. The attention fusion layer uses an attention mechanism to dynamically allocate weights.

[0061] 3. Generate a feedforward compensation signal based on the prediction deviation output by the LSTM neural network.

[0062] The LSTM neural network predicts the future change trend of environmental parameters based on historical data (temperature, humidity, oxygen concentration). The input of the LSTM neural network is historical time series data (temperature, humidity, oxygen concentration), and the output is the predicted values of environmental parameters (temperature, humidity, oxygen concentration) in the next 5 - 10 minutes. For example, when it is predicted that the temperature is about to exceed the standard, the heating power is reduced in advance and ventilation is started to avoid excessive fermentation caused by lag. The LSTM captures environmental disturbances in advance and provides a feedforward input for the coupling model. The prediction results are input into the coupling model as a feedforward compensation signal to adjust the control strategy in advance (such as starting the ventilation and humidification module in advance).

[0063] The ventilation and humidification module includes a PID control system, a variable-frequency ventilation component, an atomization humidification component, an air duct, a first solenoid valve, a main pipe, an atomization pipe, and a second solenoid valve. The PID control system is electrically connected to the polyphenol oxidase activity calculation unit. The variable-frequency ventilation component includes an inverter and a fan motor. The inverter is electrically connected to the PID control system, the inverter is electrically connected to the fan motor, the fan motor is connected to the air duct, the air duct is connected to the main pipe through the first solenoid valve, the main pipe is communicated with the grid fixed pipe rack, and a plurality of through holes are formed in the support pipe. The atomization humidification component includes a water tank, a water pump, and a high-frequency ultrasonic atomizer. The water tank is connected to the water pump, the water pump is connected to the high-frequency ultrasonic atomizer, the ultrasonic atomizer is connected to the atomization pipe, and the atomization pipe is connected to the main pipe through the second solenoid valve.

[0064] After receiving the standard values ​​of temperature, humidity and oxygen concentration fed back by the polyphenol oxidase activity calculation unit, the PID control system controls the frequency converter, the first solenoid valve, the water pump, the high-frequency ultrasonic atomizer and the second solenoid valve and other components to control the air volume and atomizing steam volume derived from the through hole of the support tube at the current position, thereby adjusting the environment of the fermentation pile.

[0065] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An on-line measurement and control device for the automatic production of Liubao tea, characterized in that It includes a sensor array bracket, a multispectral sensor, an edge computing module, and a ventilation and humidification module; A plurality of the multispectral sensors are fixedly distributed in a grid pattern on the sensor array bracket; The multispectral sensor is electrically connected to the edge computing module; The ventilation and humidification module is connected to the sensor array bracket, and the ventilation and humidification module is electrically connected to the edge computing module.

2. The on-line measurement and control device for the automated production of Liubao tea according to claim 1, characterized in that: The sensor array bracket includes a base, a riser pipe, a lead screw, a rotating motor, a moving block, and a grid fixing pipe rack; The riser pipe is fixedly arranged on the base, the lead screw is arranged in the riser pipe, the upper end of the lead screw is rotatably connected to the riser pipe, the lower end of the lead screw is fixedly connected to the rotating motor, the moving block is threadedly connected to the lead screw, a sliding hole is formed in the riser pipe, the grid fixing pipe rack passes through the sliding hole and is fixedly connected to the moving block, the grid fixing pipe rack is formed by connecting a plurality of branch pipes to form a grid-like structure, a support pipe is arranged at the grid intersection of the grid fixing pipe rack, the support pipe is communicated with the grid fixing pipe rack, and the multispectral sensor is fixedly arranged on the support pipe.

3. The on-line measurement and control device for the automated production of Liubao tea according to claim 1, characterized in that: The wavelength bands of the multispectral sensor are 1550nm band, 940nm band, and 3.5 - 4μm band.

4. An on-line measurement and control device for the automatic production of Liubao tea according to claim 1, characterized in that: The edge computing module includes a polyphenol oxidase activity calculation unit and an environmental disturbance prediction unit; The polyphenol oxidase activity calculation unit includes the following method steps: Receive the temperature, humidity, and oxygen concentration parameter values at the current position of the multispectral sensor; Establish a coupling model and calculate the polyphenol oxidase activity value using the coupling model; Compare the calculated polyphenol oxidase activity value with a preset threshold. If the polyphenol oxidase activity value deviates from the preset threshold, generate the standard values of temperature, humidity, and oxygen concentration; The environmental disturbance prediction unit includes the following method steps: Collect the temperature, humidity, and oxygen concentration data of a continuous time series; Input the data into a pre-trained LSTM neural network, and the LSTM neural network includes a dual-channel input structure and an attention fusion layer; Generate a feedforward compensation signal according to the prediction deviation output by the LSTM neural network.

5. An on-line measurement and control device for the automated production of Liubao tea according to claim 4, characterized in that: The coupling model is represented by the following formula: Among them, A t is the enzyme activity index at time t; X1 = T (temperature); X2 = RH (humidity); X3 = O2 (oxygen concentration); K i is the half-saturation constant; n is the Hill coefficient; λ is the attenuation coefficient; is the environmental parameter power term.

6. The on-line measurement and control device for the automated production of Liubao tea according to claim 5, characterized in that: It also includes a temperature and humidity synergistic effect cross-term correction term, and the cross-term correction term is introduced into the coupling model. The cross-term correction term is represented by the following formula: Cross-term correction term = 1 + α·(T - T ref )·(RH - RH ref ) Among them, α is the synergy coefficient; T ref = 35°C is the reference temperature; RH ref = 80% is the reference humidity.

7. An on-line measurement and control device for the automated production of Liubao tea according to claim 4, characterized in that: The dual-channel input structure of the LSTM neural network includes a main channel and an auxiliary channel. The main channel is used to process the temperature, humidity, and oxygen concentration data of the original time series; the auxiliary channel is used to process the external weather forecast data; the attention fusion layer uses an attention mechanism to dynamically allocate weights.

8. The on-line measurement and control device for the automated production of Liubao tea according to claim 2, characterized in that: The ventilation and humidification module includes a PID control system, a variable-frequency ventilation component, an atomization humidification component, an air duct, a first solenoid valve, a main pipe, an atomization pipe, and a second solenoid valve; The PID control system is electrically connected to the edge computing module. The variable-frequency ventilation component, the atomizing humidification component, the first solenoid valve, and the second solenoid valve are electrically connected to the PID control system. The variable-frequency ventilation component is connected to the air duct. The air duct is connected to the main pipe through the first solenoid valve. The atomizing humidification component is connected to the atomizing pipe. The atomizing pipe is connected to the main pipe through the second solenoid valve. The main pipe is communicated with the grid fixed pipe rack. A plurality of through holes are formed in the support pipe.

9. An on-line measurement and control device for the automatic production of Liubao tea according to claim 8, characterized in that: The variable-frequency ventilation component includes a frequency converter and a fan motor. The frequency converter is electrically connected to the PID control system. The frequency converter is electrically connected to the fan motor. The fan motor is connected to the air duct.

10. The on-line measurement and control device for the automated production of Liubao tea according to claim 8, characterized in that: The atomizing humidification component includes a water tank, a water pump, and a high-frequency ultrasonic atomizer. The water tank is connected to the water pump. The water pump is connected to the high-frequency ultrasonic atomizer. The ultrasonic atomizer is connected to the atomizing pipe.

Images