Degradation rate prediction method and device based on physical model and deep learning

Abstract

The invention provides a degradation rate prediction method and device based on a physical model and deep learning, and the method comprises the steps: building a data set according to key parameters in a process of processing a solid-phase substance polluted by an organic matter through a plasma processing method; wherein the key parameters are parameters influencing the organic matter degradation rate; constructing a neural network model; a neural network model is trained according to the data set, and a degradation rate prediction model of the solid-phase matter polluted by the organic matter is obtained; constructing a physical model; obtaining a degradation rate hybrid prediction model according to the degradation rate prediction model and the physical model; key parameters in the process of treating the solid-phase matter polluted by the organic matter through a plasma treatment method are obtained in real time; and according to the key parameters, obtaining a hybrid prediction organic matter degradation rate through a degradation rate hybrid prediction model.

Description

Technical Field

[0001] The present invention belongs to the field of environmental protection, and particularly relates to a degradation rate prediction method and device based on a physical model and deep learning. Background Art

[0002] Existing oil-based drill cuttings treatment technologies mainly include: pyrolysis treatment, biodegradation, chemical oxidation, and dielectric barrier discharge (DBD) plasma treatment.

[0003] Dielectric barrier discharge (DBD) plasma treatment treats oil-based drill cuttings through a high-voltage power supply, discharge electrodes, a discharge gap (usually with dielectric isolation), and a reaction chamber. Its principle is that under the action of a high-voltage electric field, a plasma is formed in the discharge gap between the electrodes. High-energy electrons react with petroleum hydrocarbon molecules, inducing oxidation, ionization, and cracking reactions, and degrading petroleum hydrocarbons into carbon dioxide, water, and other small molecules.

[0004] However, existing dielectric barrier discharge (DBD) plasma treatment does not have a suitable on-line monitoring means. The treatment efficiency of dielectric barrier discharge (DBD) plasma treatment is affected by the parameters during the dielectric barrier discharge (DBD) plasma treatment process, which easily leads to large fluctuations in the treatment efficiency of dielectric barrier discharge (DBD) plasma treatment. Summary of the Invention

[0005] The purpose of the present invention is to provide a degradation rate prediction method based on a physical model and deep learning, so as to achieve high-precision prediction of the degradation rate.

[0006] On the one hand, a degradation rate prediction method based on a physical model and deep learning is provided. The degradation rate prediction method based on a physical model and deep learning includes: Construct a data set according to the key parameters in the process of treating solid-phase substances contaminated with organic matter by the plasma treatment method; wherein, the key parameters are the parameters that affect the organic matter degradation rate; Construct a neural network model; Train the neural network model according to the data set to obtain a degradation rate prediction model for solid-phase substances contaminated with organic matter; Construct a physical model; According to the degradation rate prediction model and the physical model, obtain a degradation rate hybrid prediction model; Real-time obtain the key parameters in the process of treating solid-phase substances contaminated with organic matter by the plasma treatment method; According to the key parameters, obtain a hybrid predicted organic matter degradation rate through the degradation rate hybrid prediction model.

[0007] Optionally, the key parameters include: Input power of the power supply, discharge power of the power supply, temperature of the discharge gap, temperature of the inlet of the plasma reactor, temperature of the outlet of the plasma reactor, temperature of the ground electrode, treatment time.

[0008] Optionally, the physical model includes: Thermal power and temperature rise model for calculating the energy input corresponding to the effective discharge power; Oxidation reaction model for calculating the oxidation rate; First organic matter degradation rate prediction model for calculating the first predicted organic matter degradation rate of the organic matter based on the energy input and the oxidation rate.

[0009] Optionally, the thermal power and temperature rise model includes the following formulas: Input power balance formula: , where represents the input energy, represents the input power of the power supply; Temperature rise formula: , where represents the temperature difference, represents the temperature of the outlet of the plasma reactor, represents the temperature of the inlet of the plasma reactor; Thermal loss formula: , represents the lost energy, represents the heat loss coefficient, represents the treatment time; Calculation formula for the effective energy input: .

[0010] Optionally, the oxidation reaction model includes the following formulas:

[0011] where represents the oxidation rate of the organic matter, represents the reaction rate constant of the organic matter, represents the activation energy of the organic matter, represents the gas constant, represents the discharge gap temperature, represents the petroleum hydrocarbon concentration.

[0012] Optionally, the first organic matter degradation rate prediction model includes the following formulas:

[0013] where represents the first predicted organic matter degradation rate, represents the coupling coefficient between the discharge process and the reaction process.

[0014] Optionally, according to the key parameters, the steps of obtaining the mixed predicted organic matter degradation rate through the degradation rate mixing prediction model include: Calculating the first predicted organic matter degradation rate through a physical model ; Calculating the second predicted organic matter degradation rate through a degradation rate prediction model : According to the formula , calculating the mixed predicted organic matter degradation rate , where represents the correction coefficient.

[0015] Optionally, the method further includes: Obtaining the true organic matter degradation rate ; According to the true organic matter degradation rate and the mixed organic matter degradation rate , determining the degradation rate prediction deviation ; According to the degradation rate prediction deviation , adjusting the key parameters to improve the prediction accuracy of the mixed organic matter degradation rate.

[0016] Optionally, according to the degradation rate prediction deviation , the steps of adjusting the key parameters to improve the prediction accuracy of the mixed organic matter degradation rate include: According to the formula calculating the adjusted value of the discharge power ; Taking as in the input power balance formula for calculating to correct the prediction deviation of the physical model; Using the error backpropagation algorithm to dynamically adjust the weights and biases of the neural network to correct the prediction deviation of the degradation rate prediction model; where the loss function is defined as .

[0017] On the other hand, a degradation rate prediction device based on a physical model and deep learning is provided, including: A dataset construction module, configured to construct a dataset according to the key parameters in the process of treating solid-phase substances contaminated with organic matter by the plasma treatment method; where the key parameters are parameters that affect the organic matter degradation rate; A first model construction module, configured to construct a neural network model; training the neural network model according to the dataset to obtain a degradation rate prediction model for solid-phase substances contaminated with organic matter; The second model construction module is used to construct a physical model; The hybrid model construction module is used to obtain a degradation rate hybrid prediction model according to the degradation rate prediction model and the physical model; The acquisition module is used to acquire in real time the key parameters during the process of treating the solid-phase substance contaminated by organic matter by the plasma treatment method; The prediction module is used to obtain the hybrid predicted organic matter degradation rate through the degradation rate hybrid prediction model according to the key parameters.

[0018] The unexpectedly technical effect of the technical solution provided by the present invention is: The present invention provides a degradation rate prediction method based on a physical model and deep learning. By constructing a data set according to the key parameters in the plasma treatment method, a neural network model is trained according to the data set. At the same time, a physical model is constructed, and the neural network model and the physical model are combined to obtain a degradation rate hybrid prediction model to improve the prediction accuracy of the degradation rate. Through the degradation rate hybrid prediction model, it is possible to realize the real-time prediction of the degradation rate when treating the solid-phase substance contaminated by organic matter by the plasma treatment method. Description of the Drawings

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to these drawings without creative efforts.

[0020] Figure 1 It is a flowchart of a degradation rate prediction method based on a physical model and deep learning provided by the present invention; Figure 2 It is a schematic structural diagram of a plasma treatment device provided by the present invention; Figure 3 It is a flowchart of another degradation rate prediction method based on a physical model and deep learning provided by the present invention; Figure 4 It is a structural block diagram of a degradation rate prediction device based on a physical model and deep learning provided by the present invention; Figure 5 It is a structural block diagram of an electronic device provided by the present invention.

[0021] The reference numerals are as follows: 1: Discharge assembly; 11: Power supply; 12: First electrode; 13: Second electrode; 14: Dielectric sheet support; 2: Moving component; 21: First electrode support plate; 211: Integrated handle; 22: Second electrode support plate; 23: First guide post; 24: Second guide post; 241: Silicone cap; 25: Driving part; 251: Electric push rod; 252: Push rod controller; 3: Monitoring component; 31: Oscilloscope; 32: High-voltage probe; 33: Distance sensor; 34: Inlet thermocouple temperature sensor; 35: Outlet thermocouple temperature sensor; 36: Ground electrode thermocouple temperature sensor; 37: Non-contact temperature sensor; 4: Prediction component.

[0022] 51: Dataset construction module; 52: First model construction module; 53: Second model construction module; 54: Hybrid model construction module; 55: Acquisition module; 56: Prediction module; 61: Processor; 62: Memory. Detailed implementation manners

[0023] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0024] Figure 1 It is a flowchart of a degradation rate prediction method based on a physical model and deep learning provided by the present invention. Refer to Figure 1 and the method steps include: S101. Construct a dataset according to the key parameters in the process of treating solid substances contaminated with organic matter by the plasma treatment method; wherein, the key parameters are the parameters that affect the degradation rate of organic matter.

[0025] In one example, the key parameters include: Input power of the power supply and discharge power of the power supply temperature of the discharge gap temperature of the inlet of the plasma reactor temperature of the outlet of the plasma reactor temperature of the ground electrode treatment time .

[0026] Refer to Figure 2 , Figure 2 It is a structural schematic diagram of a plasma treatment device provided by the present invention, and the plasma treatment device includes: The discharge component 1 includes a power supply 11, a first electrode 12, a second electrode 13, and a dielectric sheet tray 14 disposed between the first electrode 12 and the second electrode 13. A plurality of dielectric sheets are carried on the dielectric sheet tray 14, and a solid-phase substance contaminated with an organic substance is carried on the second electrode 13; The moving component 2 drives the first electrode 12 and the second electrode 13 to approach or separate from each other; The monitoring component 3 is used to obtain voltage and current waveform data between the first electrode 12 and the second electrode 13; The prediction component 4 is used to predict the degradation rate according to key parameters.

[0027] In this embodiment, the power supply 11 is a high-voltage power supply.

[0028] In this embodiment, the first electrode 12 is a high-voltage electrode, and the second electrode 13 is a ground electrode.

[0029] In this embodiment, the first electrode 12 is a metal block made of stainless steel and is connected to the stud through an M8 thread, which is convenient for replacement.

[0030] In this embodiment, the second electrode 13 is a metal block made of stainless steel and is connected to the stud through an M8 thread, which is convenient for replacement.

[0031] In this embodiment, the first electrode 12 has a first electrode terminal 121 connected to the first electrode 12, and the second electrode 13 has a second electrode connection line 131 connected to the second electrode 13.

[0032] In this embodiment, the dielectric disk tray 14 is used for loading quartz wafers, is provided with different diameter steps, is adapted to different quartz wafers, and is made of ABS material.

[0033] In this embodiment, the moving component 2 includes: A first electrode tray 21 for fixing the first electrode 12; A second electrode tray 22 for fixing the second electrode 13; A first guide post 23, and a plurality of the first guide posts 23 penetrate through the first electrode tray 21 and the second electrode tray 22; A second guide post 24, and a plurality of the second guide posts 24 penetrate through the first electrode tray 21 and the dielectric sheet tray 14; A driving member 25, and the driving member 25 is used to drive the first electrode tray 21 to approach or separate from the second electrode tray 22.

[0034] In this embodiment, an integrated handle 211 is provided on the first electrode tray 21, which is convenient for pulling out the first electrode and is made of nylon material.

[0035] In this embodiment, there is a gap between the first electrode support plate 21 and the second electrode support plate 22, and the materials are both black ABS.

[0036] In this embodiment, there are 4 first guide posts 23, which are used to maintain the stability of the motor support plate during the up and down movement, and are made of stainless steel.

[0037] In this embodiment, there are 4 second guide posts 24, which are used to maintain the stability of the dielectric disk support plate during the up and down movement, and are made of FR4 material.

[0038] In this embodiment, a silicone cap 241 is provided on the second guide post 24, which is used for warning when the electrode support plate is lifted. When the support plate rises to the lower end of the silicone cap 241, the lifting stops.

[0039] In this embodiment, the driving member 25 includes an electric push rod 251 and a push rod controller 252.

[0040] In this embodiment, the electric push rod 251 is the power device for the up and down movement of the first electrode support plate, with a speed of about 0.4 mm / s and made of aluminum alloy material.

[0041] In this embodiment, the push rod controller 252 is built-in with a radio frequency wireless control module to control the electric push rod to move up or down in a jog mode, and the jog distance at the middle position is 0.2 mm.

[0042] In this embodiment, the monitoring component 3 includes an oscilloscope 31, a high-voltage probe 32, a distance sensor 33, an intake port thermocouple temperature sensor 34, an outlet port thermocouple temperature sensor 35, a ground electrode thermocouple temperature sensor 36, and a non-contact temperature sensor 37 (for measuring the temperature of the discharge gap).

[0043] In this embodiment, the monitoring component 3 further includes that the distance sensor 33 is used to obtain the distance between the first electrode 12 and the second electrode 13.

[0044] In this embodiment, the distance sensor 33 is a wire-pulling distance sensor, which is used to display the lifting distance of the first electrode support plate, and the specific value is judged by the zero calibration operation.

[0045] S102. Build a neural network model.

[0046] In one example, the neural network model includes: Input layer: 5 nodes (corresponding to the input feature vector )

[0047] Among them, the feature vector , .

[0048] Hidden layer: including the first layer and the second layer; among them, the first layer includes 32 neurons, and the activation function is ReLU. The second layer includes 16 neurons, and the activation function is ReLU.

[0049] Output layer: 1 node, the activation function is Sigmoid, and the predicted value of the degradation rate of organic matter (petroleum hydrocarbons) is output:

[0050] S103. Train the neural network model according to the data set to obtain a prediction model for the degradation rate of solid substances contaminated by organic matter.

[0051] Among them, the loss function is defined as: ; The optimization algorithm uses the Adam optimizer, and the learning rate is set to 0.001. At the same time, the number of training rounds during the training process is set to 500 rounds, and the early stopping strategy is used to avoid overfitting.

[0052] S104. Build a physical model.

[0053] In one example, the physical model includes: Thermal power and temperature rise model, used to calculate the energy input corresponding to the effective discharge power; Oxidation reaction model, used to calculate the oxidation rate; The first organic matter degradation rate prediction model, used to calculate the first predicted organic matter degradation rate according to the energy input and the oxidation rate.

[0054] In this embodiment, the thermal power and temperature rise model includes the following formula: Input power balance formula: , where represents the input energy, represents the input power of the power supply; Temperature rise formula: , where represents the temperature difference, represents the temperature at the outlet of the plasma reactor, represents the temperature at the inlet of the plasma reactor; Thermal loss formula: , represents the lost energy, represents the heat loss coefficient, represents the processing time; Calculation formula for effective energy input: .

[0055] In this embodiment, the oxidation reaction model includes the following formula:

[0056] Among them, represents the oxidation rate of the organic matter, represents the reaction rate constant of the organic matter, represents the activation energy of the organic matter, represents the gas constant, represents the temperature of the discharge gap, represents the concentration of petroleum hydrocarbons.

[0057] In this embodiment, the first organic matter degradation rate prediction model includes the following formula:

[0058] Among them, represents the predicted degradation rate of the first predicted organic matter, represents the coupling coefficient between the discharge process and the reaction process.

[0059] S105. According to the degradation rate prediction model and the physical model, a degradation rate hybrid prediction model is obtained.

[0060] S106. Key parameters during the process of treating the solid-phase substance contaminated with organic matter by the plasma treatment method are obtained in real time.

[0061] S107. According to the key parameters, the hybrid predicted organic matter degradation rate is obtained through the degradation rate hybrid prediction model.

[0062] The first step is to calculate the predicted degradation rate of the first predicted organic matter through the physical model ; The second step is to calculate the predicted degradation rate of the second predicted organic matter through the degradation rate prediction model : The third step is to calculate the hybrid predicted organic matter degradation rate according to the formula , where , among them, represents the correction coefficient.

[0063] Among them, to prevent excessive power adjustment, the correction coefficient can also be dynamically adjusted according to the measurement data, and the formula is as follows:

[0064] Among them, represents the updated correction coefficient, represents the correction coefficient before update, represents the adjustment step size.

[0065] Figure 3 is the flowchart of another degradation rate prediction method based on a physical model and deep learning provided by the present invention. Refer to Figure 3 , the steps of this method include: S201. Construct a data set according to the key parameters in the process of treating solid-phase substances contaminated with organic matter by the plasma treatment method; among them, the key parameters are the parameters that affect the degradation rate of organic matter.

[0066] S202. Construct a neural network model.

[0067] S203. Train the neural network model according to the data set to obtain a degradation rate prediction model for solid-phase substances contaminated with organic matter.

[0068] S204. Construct a physical model.

[0069] S205. Obtain a hybrid degradation rate prediction model according to the degradation rate prediction model and the physical model.

[0070] S206. Real-time obtain the key parameters in the process of treating solid-phase substances contaminated with organic matter by the plasma treatment method.

[0071] S207. According to the key parameters, obtain the hybrid predicted organic matter degradation rate through the hybrid degradation rate prediction model.

[0072] S208. Obtain the real organic matter degradation rate.

[0073] S209. Determine the degradation rate prediction deviation according to the real organic matter degradation rate and the hybrid organic matter degradation rate.

[0074] When the degradation rate prediction deviation is higher than the threshold, it is determined that the key parameters need to be adjusted. Among them, as an example, the threshold can be 5%.

[0075] S210. Adjust the key parameters according to the degradation rate prediction deviation to improve the prediction accuracy of the hybrid organic matter degradation rate.

[0076] In one example, step S210 includes: The first step. According to the formula Calculate the adjusted value of the discharge power .

[0077] Of course, if continues to be higher than the threshold, appropriately extend the treatment time to obtain , and use as in the physical model, where:

[0078] is the adjustment time interval, calibrated according to experiments.

[0079] The second step. Use As in the input power balance formula , it is used to calculate to correct the prediction deviation of the physical model.

[0080] Step 3: Use the error backpropagation algorithm to dynamically adjust the weights and biases of the neural network to correct the prediction deviation of the degradation rate prediction model; among them, the loss function is defined as .

[0081] Among them, the update formulas for the weights and biases are as follows:

[0082]

[0083] Among them, , respectively represent the weights and biases of the th layer in the neural network.

[0084] The present invention also provides some embodiments: Embodiment 1: Experimental conditions: Input power ; Actually measured degradation rate: ; Mixed predicted organic matter degradation rate: ; Error analysis:

[0085] Power adjustment (in the case where the error is less than the threshold):

[0086] Subsequently, correct the deviations of the physical model and the neural network according to the above to ensure that the predicted degradation rate value is close to the actual degradation rate value.

[0087] Embodiment 2: Laboratory small-scale test Experimental conditions: The sample is oil-based drill cuttings with an oil content of 20% and a mass of 100 g. The equipment is a laboratory-scale DBD plasma reactor with a discharge gap of 10 mm. The input power is set to 5.5 kW, the discharge frequency is set to 10 kHz, and the treatment time is 20 minutes. Measurement and calculation: Temperature rise: Inlet air temperature = 25°C, outlet air temperature = 370°C.

[0088] Discharge gap temperature: = 340°C Ground electrode temperature: = 320°C Physical model prediction:

[0089] Deep learning correction:

[0090] Final prediction result: 。

[0091] Actual result: Experimental measurement shows that the degradation rate is 96%, which is consistent with the model prediction, verifying the model accuracy.

[0092] Example 3: Dynamic feedback and process optimization test.

[0093] Experimental conditions: Sample: Oil-based drill cuttings with an oil content of 25%, mass 500 g.

[0094] Operating parameters: Initial input power 7.0 kW, discharge power 6.5 kW, discharge frequency 12 kHz, reaction time 25 minutes.

[0095] Initial prediction: 。

[0096] First feedback correction: Deep learning correction: 。

[0097] Obtain the mixed predicted degradation rate: 。

[0098] The actually measured degradation rate is 97%, and the discharge power is adjusted to 6.8 kW.

[0099] Second feedback correction: Corrected prediction: 。

[0100] Actual result: The degradation rate is 98.4%, and the error is less than 1%.

[0101] Figure 4 The structure block diagram of a degradation rate prediction device based on a physical model and deep learning provided by the present invention is shown in Figure 4 , including: A data set construction module 51, configured to construct a data set according to key parameters in the process of treating solid-phase substances contaminated with organic matter by a plasma treatment method; wherein, the key parameters are parameters that affect the organic matter degradation rate; A first model construction module 52, configured to construct a neural network model; train the neural network model according to the data set to obtain a degradation rate prediction model for solid-phase substances contaminated with organic matter; The second model construction module 53 is used to construct a physical model; The hybrid model construction module 54 is used to obtain a degradation rate hybrid prediction model according to the degradation rate prediction model and the physical model; The acquisition module 55 is used to acquire key parameters in real time during the process of treating solid-phase substances contaminated with organic matter by the plasma treatment method; The prediction module 56 is used to obtain the hybrid predicted organic matter degradation rate through the degradation rate hybrid prediction model according to the key parameters.

[0102] Figure 5 This is a structural block diagram of an electronic device provided by the present invention. Refer to Figure 5 , the electronic device may include Figure 4 The degradation rate prediction device based on the physical model and deep learning described above. Generally, the electronic device includes: a processor 61 and a memory 62.

[0103] The processor 61 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 61 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 61 may also include a main processor and a coprocessor. The main processor is a processor used to process data in the wake state, also known as the CPU (Central Processing Unit); the coprocessor is a low-power processor used to process data in the standby state. The memory 62 may include one or more computer-readable storage media, and the computer-readable storage media may be non-transitory. The memory 62 may also include a high-speed random access memory and a non-volatile memory, such as one or more disk storage devices and flash storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 62 is used to store at least one instruction, and the at least one instruction is used to be executed by the processor 61 to implement the degradation rate prediction method based on the physical model and deep learning executed by the electronic device provided in the method embodiments of the present application.

[0104] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A degradation rate prediction method based on physical models and deep learning, characterized in that, The degradation rate prediction method based on physical models and deep learning includes: Construct a data set according to the key parameters in the process of treating solid-phase substances contaminated by organic matter using the plasma treatment method; among them, the key parameters are the parameters that affect the degradation rate of organic matter; Construct a neural network model; Train the neural network model according to the data set to obtain a degradation rate prediction model for solid-phase substances contaminated by organic matter; Construct a physical model; Obtain a degradation rate hybrid prediction model according to the degradation rate prediction model and the physical model; Obtain the key parameters in real time during the process of treating solid-phase substances contaminated by organic matter using the plasma treatment method; According to the key parameters, obtain the hybrid predicted degradation rate of organic matter through the degradation rate hybrid prediction model.

2. The degradation rate prediction method based on physical models and deep learning according to claim 1, characterized in that The key parameters include: The input power of the power supply, the discharge power of the power supply, the temperature of the discharge gap, the temperature of the inlet of the plasma reactor, the temperature of the outlet of the plasma reactor, the temperature of the ground electrode, and the treatment time.

3. The degradation rate prediction method based on physical models and deep learning according to claim 1, wherein The physical model includes: A thermal power and temperature rise model for calculating the energy input corresponding to the effective discharge power; An oxidation reaction model for calculating the oxidation rate; A first organic matter degradation rate prediction model for calculating the first predicted degradation rate of organic matter according to the energy input and the oxidation rate.

4. The degradation rate prediction method based on physical models and deep learning according to claim 3, characterized in that The thermal power and temperature rise model includes the following formula: Input power balance formula: , where represents the input energy, represents the input power of the power supply; Temperature rise formula: , where represents the temperature difference, represents the temperature at the outlet of the plasma reactor, represents the temperature at the inlet of the plasma reactor; Heat loss formula: , represents the lost energy, represents the heat loss coefficient, represents the processing time; Calculation formula for effective energy input: .

5. The degradation rate prediction method based on physical models and deep learning according to claim 3, characterized in that The oxidation reaction model includes the following formula: Among them, represents the oxidation rate of organic matter, represents the reaction rate constant of organic matter, represents the activation energy of organic matter, represents the gas constant, represents the temperature of the discharge gap, represents the petroleum hydrocarbon concentration.

6. The degradation rate prediction method based on physical models and deep learning according to claim 3, characterized in that The first organic matter degradation rate prediction model includes the following formula: Among them, represents the first predicted organic matter degradation rate, represents the coupling coefficient of the discharge process and the reaction process.

7. The degradation rate prediction method based on physical models and deep learning according to any one of claims 1 to 6, characterized in that, The steps of obtaining the hybrid predicted degradation rate of organic matter through the degradation rate hybrid prediction model according to the key parameters include: The first predicted organic matter degradation rate is obtained through physical model calculation ; The second predicted organic matter degradation rate is calculated by the degradation rate prediction model : According to the formula , the mixed predicted organic matter degradation rate is calculated, where represents the correction coefficient.

8. The degradation rate prediction method based on physical models and deep learning according to any one of claims 1 to 6, characterized in that The method further includes: Obtain the true organic matter degradation rate ; According to the true organic matter degradation rate and the mixed organic matter degradation rate , determine the prediction deviation of the degradation rate ; Based on the prediction deviation of the degradation rate , adjust the key parameters to improve the prediction accuracy of the degradation rate of the mixed organic matter.

9. The degradation rate prediction method based on physical model and deep learning according to claim 8, characterized in that According to the prediction deviation of the degradation rate , the steps of adjusting key parameters to improve the prediction accuracy of the degradation rate of mixed organic substances include: According to the formula the adjusted value of the discharge power is calculated ; Take as the in the input power balance formula, and use it to calculate to correct the prediction deviation of the physical model; The weights and biases of the neural network are dynamically adjusted using the error backpropagation algorithm to correct the prediction deviation of the degradation rate prediction model; among them, the loss function is defined as .

10. A degradation rate prediction device based on a physical model and deep learning, characterized in that, including: A data set construction module for constructing a data set according to the key parameters in the process of treating solid-phase substances contaminated by organic matter using the plasma treatment method; among them, the key parameters are the parameters that affect the degradation rate of organic matter; A first model construction module for constructing a neural network model; training the neural network model according to the data set to obtain a degradation rate prediction model for solid-phase substances contaminated by organic matter; A second model construction module for constructing a physical model; A hybrid model construction module for obtaining a degradation rate hybrid prediction model according to the degradation rate prediction model and the physical model; An acquisition module for obtaining the key parameters in real time during the process of treating solid-phase substances contaminated by organic matter using the plasma treatment method; A prediction module for obtaining the hybrid predicted degradation rate of organic matter through the degradation rate hybrid prediction model according to the key parameters.

Images