Decision support system for biomass pyrolysis process

Abstract

The invention relates to a decision support system for a biomass pyrolysis process, which is characterized in that on the basis of arranging a biomass pyrolysis device and an acquisition device, a control module and a large language model module are further arranged; the control module is used for obtaining a temperature prediction result and a pressure prediction result of the pyrolysis reactor at the next time point, a temperature adjustment amplitude required for enabling the temperature in the pyrolysis reactor to return to a safety range and a pressure adjustment amplitude required for enabling the pressure in the pyrolysis reactor to return to the safety range; and when the temperature adjustment amplitude and / or the pressure adjustment amplitude exceed the threshold value, a decision request instruction is sent out, the decision request instruction is processed by the large language model module and then early warning is sent to a supervisor, and meanwhile, operation suggestions are provided for the supervisor to make decisions, so that the uninterpretability of artificial intelligence is reduced, and the reliability is improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of biomass pyrolysis and artificial intelligence. Specifically, it relates to a decision support system for the biomass pyrolysis process. Background Art

[0002] To assist supervisors in achieving intelligent and refined control of biomass pyrolysis devices during the pyrolysis process, a real-time control technology integrating advanced neural network algorithms is disclosed in the prior art. This technology provides efficient and reliable decision support for high-dynamic working conditions.

[0003] However, the application of this real-time control technology based on neural network algorithms in industry usually faces a trade-off between accuracy and interpretability. For relatively large adjustments (the adjustment range is more than 5 units) of temperature and pressure in the biomass pyrolysis process involving high temperature and pressure, manual supervision is mainly relied on, and the adjustment range highly depends on the experience of supervisors. Summary of the Invention

[0004] The technical problem to be solved by the present invention is how to overcome the technical defect in the prior art that the interpretability of relatively large adjustments during the biomass pyrolysis process mainly relies on manual supervision. To overcome the above defects of the prior art, the present invention provides a decision support system for the biomass pyrolysis process.

[0005] A decision support system for the biomass pyrolysis process provided by the present invention includes:

[0006] A biomass pyrolysis device, including a feeding device, a pyrolysis reactor, a gas collection device, and a heat recovery device that form a loop structure;

[0007] A collection device for respectively obtaining temperature information, pressure information, and gas component information in the pyrolysis reactor and the gas collection device;

[0008] A control module configured to obtain, through a deep neural network algorithm, a temperature prediction result, a pressure prediction result, a temperature adjustment range required to return the temperature in the pyrolysis reactor to a safe range, and a pressure adjustment range required to return the pressure in the pyrolysis reactor to a safe range based on the temperature information, pressure information, and gas component information currently obtained by the collection device, and issue a request for decision-making instructions when the temperature adjustment range and / or the pressure adjustment range exceeds a threshold, and respectively execute adjusting the temperature in the pyrolysis reactor according to the temperature adjustment range and adjusting the pressure in the pyrolysis reactor according to the pressure adjustment range when both the temperature adjustment range and the pressure adjustment range do not exceed the threshold;

[0009] The large language model module is configured to issue a warning to the supervisor after receiving the request decision instruction, and at the same time provide operation suggestions for the supervisor to make a decision, and adjust the temperature and pressure in the pyrolysis reactor through the control module according to the decision of the supervisor;

[0010] Among them,

[0011] The control module is electrically connected to the acquisition device, and the large language model module is electrically connected to the control module.

[0012] The decision support system for the biomass pyrolysis process disclosed in the present invention aims at the technical problems of the present invention. By setting up a biomass pyrolysis device and an acquisition device, a control module and a large language model module are further set up. The control module obtains the temperature prediction result, pressure prediction result, temperature adjustment range required to return the temperature in the pyrolysis reactor to the safe range, and pressure adjustment range required to return the pressure in the pyrolysis reactor to the safe range at the next time point, and issues a request decision instruction when the temperature adjustment range and / or pressure adjustment range exceeds the threshold, and respectively executes adjusting the temperature in the pyrolysis reactor according to the temperature adjustment range and adjusting the pressure in the pyrolysis reactor according to the pressure adjustment range when both the temperature adjustment range and the pressure adjustment range do not exceed the threshold. For the request decision instruction, it is processed by the large language model module to issue a warning to the supervisor, and at the same time provide operation suggestions for the supervisor to make a decision, thereby making up for the respective deficiencies of the existing technology in using personnel alone and using neural network algorithms alone, and at the same time giving play to the universality of knowledge of neural network algorithms and the supervision of human personnel. Furthermore, it reduces the inexplicability of artificial intelligence and increases reliability, thus overcoming the technical defect that the interpretability of the existing technology that requires large adjustments and manipulations mainly relies on manual supervision.

[0013] In a possible implementation manner, the acquisition device includes:

[0014] A first temperature sensor, electrically connected to the control module, for obtaining temperature information in the pyrolysis reactor;

[0015] A second temperature sensor, electrically connected to the control module, for obtaining temperature information in the gas collection device;

[0016] A first pressure sensor, electrically connected to the control module, for obtaining pressure information in the pyrolysis reactor;

[0017] A second pressure sensor, electrically connected to the control module, for obtaining pressure information in the gas collection device;

[0018] A first gas composition analyzer, electrically connected to the control module, for obtaining gas composition information in the pyrolysis reactor;

[0019] A second gas composition analyzer, electrically connected to the control module, for obtaining gas composition information in the gas collection device;

[0020] The setting of the temperature sensor ensures that the pyrolysis reaction proceeds within a set temperature range, i.e., a safe range; the pressure sensor is responsible for monitoring the pressure changes inside and outside the pyrolysis reactor to prevent equipment damage caused by overpressure; the gas composition analyzer can analyze the components in the collected gas samples in real time, thus enabling real-time detection and control of temperature and pressure during the biomass pyrolysis process, achieving high-precision control of the reactor during the pyrolysis process.

[0021] In a possible implementation manner, the control module includes:

[0022] A multi-layer perception algorithm unit, configured to obtain the temperature prediction result, the pressure prediction result, the temperature adjustment amplitude, and the pressure adjustment amplitude by using the multi-layer perception algorithm with the temperature information, pressure information, and gas composition information currently obtained by the acquisition device;

[0023] A central processing unit, configured to determine whether both the temperature adjustment amplitude and the pressure adjustment amplitude exceed a threshold;

[0024] If so, respectively execute a first adjustment instruction according to the temperature adjustment amplitude and a second adjustment instruction according to the pressure adjustment amplitude;

[0025] If not, issue the request decision instruction;

[0026] A temperature adjustment device, for performing temperature adjustment in the pyrolysis reactor according to the first adjustment instruction;

[0027] A pressure adjustment device, for performing pressure adjustment in the pyrolysis reactor according to the second adjustment instruction;

[0028] Wherein,

[0029] The multi-layer perception algorithm unit is simultaneously electrically connected to the first temperature sensor, the second temperature sensor, the first pressure sensor, the second pressure sensor, the first gas composition analyzer, and the second gas composition analyzer, the central processing unit is simultaneously electrically connected to the multi-layer perception algorithm unit and the large language model module, the temperature adjustment device is electrically connected to the central processing unit, and the pressure adjustment device is electrically connected to the central processing unit;

[0030] This solution combines a neural network algorithm based on deep learning with a central processing unit, which can not only achieve dynamic adjustment of pyrolysis process parameters, but also facilitate real-time decision-making for assisting supervisors.

[0031] In a possible implementation, the multi-layer perceptron algorithm unit includes:

[0032] A preprocessing device, configured to perform normalization processing on the temperature information, pressure information, and gas composition information obtained by the acquisition device to obtain a normalization result;

[0033] A multi-layer perceptron, configured to obtain the temperature prediction result, the pressure prediction result, the temperature adjustment range, and the pressure adjustment range by using the multi-layer perceptron algorithm with the normalization result;

[0034] Wherein,

[0035] The preprocessing device is electrically connected to the first temperature sensor, the second temperature sensor, the first pressure sensor, the second pressure sensor, the first gas composition analyzer, and the second gas composition analyzer at the same time, and the multi-layer perceptron is electrically connected to the central processor and the preprocessing device at the same time;

[0036] The multi-layer perceptron algorithm adopted in this solution can effectively extract features from complex time series data and predict system behavior, and is suitable for processing numerical time series data (such as temperature, pressure, and gas composition, etc.). This algorithm maps the input data to a higher-dimensional space through a non-linear activation function (such as ReLU) to extract features and learn complex relationships.

[0037] In a possible implementation, the multi-layer perceptron algorithm unit further includes:

[0038] An optimizer, electrically connected to the multi-layer perceptron, configured to optimize the parameters of the multi-layer perceptron through a mean square error loss function;

[0039] An optimization accelerator, electrically connected to the optimizer, configured to accelerate the operation of the parameter optimization of the optimizer through an operation acceleration algorithm;

[0040] This solution can support the training and real-time inference of the multi-layer perceptron, ensure the rapid processing of big data. At the same time, using the optimization accelerator can accelerate the training process of the neural network, significantly reduce the parameter optimization time, and improve the inference speed.

[0041] In a possible implementation, the large language model module includes:

[0042] A language assistant unit, configured to issue a warning after receiving the request decision instruction, and generate operation suggestions for the supervisor to select by using the request decision instruction through the Char AI language assistant algorithm;

[0043] A display unit, which is configured to display the warnings and operation suggestions issued by the language assistant unit, and adjust the temperature in the pyrolysis reactor through the temperature adjustment device and / or adjust the pressure in the pyrolysis reactor through the pressure adjustment device after the supervisor makes a decision;

[0044] Wherein,

[0045] The language assistant unit is electrically connected to the central processing unit, and the display unit is simultaneously electrically connected to the language assistant unit, the temperature adjustment device, and the pressure adjustment device;

[0046] This solution makes the operation of the pyrolysis process more simple and efficient by setting up a language assistant unit based on the Char AI language assistant algorithm. Supervisors only need to obtain real-time data through a simple interface, reducing the need for complex operations. The simplification of the operation process reduces the requirements for operators, enabling new employees to quickly get started. At the same time, it can reduce the labor intensity of manual work. The intelligent detection system reduces manual intervention, lowering the labor intensity of operators and improving work efficiency and safety. Meanwhile, the real-time detection ability of Char AI can enhance the safety of the entire pyrolysis process, effectively increasing the safety of operations and avoiding potential risks brought by the automation process. Brief Description of the Drawings

[0047] Figure 1 It is a schematic structural diagram of a decision support system for a biomass pyrolysis process disclosed in an embodiment of the present application;

[0048] Figure 2 It is a schematic structural diagram of a biomass pyrolysis device disclosed in an embodiment of the present application;

[0049] Figure 3 It is a schematic structural diagram of a collection device, a control module, and a large language model module disclosed in an embodiment of the present application;

[0050] Figure 4 It is a schematic structural diagram of a multi-layer perceptron algorithm unit disclosed in an embodiment of the present application;

[0051] Figure 5 It is a schematic diagram of the operation process of a multi-layer perceptron disclosed in an embodiment of the present application. Detailed Embodiments

[0052] First of all, those skilled in the art should understand that these embodiments are only used to explain the technical principles of the embodiments of the present application, and are not intended to limit the protection scope of the embodiments of the present application. Those skilled in the art can make adjustments according to needs to adapt to specific application scenarios.

[0053] In the description of the embodiments of the present application, it should be noted that, unless otherwise clearly specified and defined, the terms "electrically connected" and "electric connection relationship" should be understood in a broad sense, that is, a connection method with an electrical relationship. For example, it can be a circuit connection through a wire, or an electrical connection through a radio signal channel (channel), or a combination of both. In addition, the "electrically connected" and "electric connection relationship" can be based on a mechanical connection (such as a wire arranged in a connection key); it can be directly connected or indirectly connected through an intermediate medium. For those of ordinary skill in the art, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific circumstances..

[0054] The present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] See Figures 1 to 5 , an embodiment of the present application discloses a decision support system for the biomass pyrolysis process, Figure 1 is a schematic diagram of the overall structure of the system. The system includes a biomass pyrolysis device, a collection device, a control module, and a large language model module. Among them, the control module is electrically connected to the collection device, and the large language model module is electrically connected to the control module.

[0056] See Figure 2 , in this system, the biomass pyrolysis device includes a feeding device, a pyrolysis reactor, a gas collection device, and a heat recovery device. Among them, the feeding device is provided with a feeding inlet, an air flow inlet, and an air flow outlet. The air flow outlet of the feeding device is connected to the inlet of the pyrolysis reactor through a pipeline. The outlet of the pyrolysis reactor is connected to the inlet of the gas collection device through a pipeline. The outlet of the gas collection device is connected to the inlet of the heat recovery device through a pipeline. The outlet of the heat recovery device is connected to the air flow inlet of the feeding device through a pipeline, thus forming a closed loop.

[0057] See Figure 3, in this system, the acquisition device is used to obtain the temperature information, pressure information, and gas composition information in the pyrolysis reactor and the gas collection device respectively. The acquisition device includes a first temperature sensor, a second temperature sensor, a first pressure sensor, a second pressure sensor, a first gas composition analyzer, and a second gas composition analyzer; the first temperature sensor is disposed on the pyrolysis reactor and electrically connected to the control module for obtaining the temperature information in the pyrolysis reactor; the second temperature sensor is disposed on the gas collection device and electrically connected to the control module for obtaining the temperature information in the gas collection device; the first pressure sensor is disposed on the pyrolysis reactor and electrically connected to the control module for obtaining the pressure information in the pyrolysis reactor; the second pressure sensor is disposed on the gas collection device and electrically connected to the control module for obtaining the pressure information in the gas collection device; the first gas composition analyzer is disposed on the pyrolysis reactor and electrically connected to the control module for obtaining the gas composition information in the pyrolysis reactor; the second gas composition analyzer is disposed on the gas collection device and electrically connected to the control module for obtaining the gas composition information in the gas collection device.

[0058] In this embodiment, both the first temperature sensor and the second temperature sensor adopt high-precision temperature sensors of model ADT7410, both the first pressure sensor and the second pressure sensor adopt high-precision pressure sensors of model VPG-2TH, and both the first gas composition analyzer and the second gas composition analyzer are analysis devices for analyzing components such as CO, CO2, H2, and CH4 in the gas to optimize the reaction conditions.

[0059] Please continue to refer to Figure 3 As shown, in this system, the control module is configured to use the temperature information, pressure information, and gas composition information currently obtained by the acquisition device through a deep neural network algorithm to obtain the temperature prediction result, pressure prediction result, temperature adjustment range required to return the temperature in the pyrolysis reactor to the safe range, and pressure adjustment range required to return the pressure in the pyrolysis reactor to the safe range at the next time point, and issue a request decision instruction when the temperature adjustment range and / or pressure adjustment range exceeds the threshold, and respectively execute adjusting the temperature in the pyrolysis reactor according to the temperature adjustment range and adjusting the pressure in the pyrolysis reactor according to the pressure adjustment range when both the temperature adjustment range and the pressure adjustment range do not exceed the threshold.

[0060] In this embodiment, the control module includes a multi-layer perception algorithm unit, a central processing unit, a temperature regulation device, and a pressure regulation device. Among them, the multi-layer perception algorithm unit is electrically connected to the first temperature sensor, the second temperature sensor, the first pressure sensor, the second pressure sensor, the first gas composition analyzer, and the second gas composition analyzer respectively. The central processing unit is electrically connected to the multi-layer perception algorithm unit. The temperature regulation device is electrically connected to the central processing unit. The pressure regulation device is electrically connected to the central processing unit.

[0061] In the control module, the multi-layer perception algorithm unit is configured to use the multi-layer perception algorithm to obtain temperature prediction results, pressure prediction results, temperature adjustment ranges, and pressure adjustment ranges by using the temperature information, pressure information, and gas composition information obtained by the acquisition device. The central processing unit is configured to determine whether both the temperature adjustment range and the pressure adjustment range do not exceed the threshold. If so, the first adjustment instruction is issued according to the temperature adjustment range and the second adjustment instruction is issued according to the pressure adjustment range respectively. If not, the request decision instruction is issued. The temperature regulation device is arranged on the pyrolysis reactor and is used to perform temperature regulation in the pyrolysis reactor according to the first adjustment instruction. The pressure regulation device is arranged on the pyrolysis reactor and is used to perform pressure regulation in the pyrolysis reactor according to the second adjustment instruction.

[0062] See Figure 4 , the multi-layer perception algorithm unit includes a preprocessing device, a multi-layer perceptron, an optimizer, and an optimization accelerator. Among them, the preprocessing device is electrically connected to the first temperature sensor, the second temperature sensor, the first pressure sensor, the second pressure sensor, the first gas composition analyzer, and the second gas composition analyzer respectively. The multi-layer perceptron is electrically connected to the preprocessing device.

[0063] In the multi-layer perception algorithm unit, the preprocessing device is configured to perform normalization processing on the temperature information, pressure information, and gas composition information obtained by the acquisition device to obtain a normalization result. The multi-layer perceptron is configured to use the multi-layer perception algorithm to obtain temperature prediction results, pressure prediction results, temperature adjustment ranges, and pressure adjustment ranges by using the normalization result. The optimizer is electrically connected to the multi-layer perceptron and is configured to perform parameter optimization on the multi-layer perceptron through the mean square error loss function. The optimization accelerator is electrically connected to the optimizer and is configured to perform arithmetic acceleration on the parameter optimization of the optimizer through an arithmetic acceleration algorithm. This structure can support the training and real-time inference of the multi-layer perceptron, ensure the rapid processing of big data. At the same time, using the optimization accelerator can accelerate the training process of the neural network, significantly reduce the parameter optimization time, and improve the inference speed.

[0064] To accelerate training and avoid local optima, some optimization algorithms are usually adopted to adjust the learning rate of the neural network. In this embodiment, the adam optimizer is used as the optimizer to adaptively adjust the learning rate of each parameter of the multi-layer perceptron during training, and quickly converge to the optimal solution. During the entire training cycle, an appropriate number of training epochs need to be set according to the error changes during training. Generally speaking, a small dataset may require thousands to tens of thousands of data, while a large dataset requires millions of data and more training time. Generally, it is more reasonable to set the number of training epochs to 20 - 100. Usually, during training, the error of the validation set is regularly evaluated, and the hyperparameters of the model (such as the learning rate) are adjusted according to the performance of the validation set. Similarly, the technique of cross-validation can be adopted to further ensure the generalization ability of the model and prevent overfitting on a single data. At the same time, the optimization accelerator in this embodiment uses a GPU cluster to accelerate the training process of the multi-layer perceptron. Through multi-GPU parallel computing, the model training time is significantly reduced and the inference speed is increased.

[0065] Please continue to refer to Figure 3 As shown, in this system, the large language model module is set to issue a warning to the supervisor after receiving a request decision instruction, and at the same time provide operation suggestions for the supervisor to make a decision, and adjust the temperature and pressure in the pyrolysis reactor through the control module according to the decision of the supervisor.

[0066] In this embodiment, the large language model module includes a language assistant unit and a display unit. Among them, the language assistant unit is electrically connected to the central processor, and the display unit is electrically connected to the language assistant unit, the temperature adjustment device, and the pressure adjustment device. The language assistant unit is set to issue a warning after receiving a request decision instruction, and generate operation suggestions for the supervisor to select by using the request decision instruction through the Char AI language assistant algorithm; the display unit is set to display the warning and operation suggestions issued by the language assistant unit, and adjust the temperature in the pyrolysis reactor through the temperature adjustment device and / or adjust the pressure in the pyrolysis reactor through the pressure adjustment device after the supervisor makes a decision.

[0067] The decision support system for the biomass pyrolysis process disclosed in this embodiment, based on the setting of the biomass pyrolysis device and the acquisition device, further sets a control module and a large language model module. The control module obtains the temperature prediction result, pressure prediction result, the temperature adjustment range required to return the temperature in the pyrolysis reactor to the safe range, and the pressure adjustment range required to return the pressure in the pyrolysis reactor to the safe range at the next time point, and issues a request for decision-making instruction when the temperature adjustment range and / or pressure adjustment range exceeds the threshold. When both the temperature adjustment range and the pressure adjustment range do not exceed the threshold, it respectively executes adjusting the temperature in the pyrolysis reactor according to the temperature adjustment range and adjusting the pressure in the pyrolysis reactor according to the pressure adjustment range. For the request for decision-making instruction, it is processed by the large language model module to issue a warning to the supervisor and provide operation suggestions for the supervisor to make a decision, thereby making up for the respective deficiencies of separately using personnel and separately using neural network algorithms in the prior art, and at the same time giving play to the universality of knowledge of the neural network algorithm and the supervision of human personnel. Furthermore, it reduces the unexplainability of artificial intelligence and increases reliability, thus overcoming the technical defect in the prior art that the explainability of large adjustments and manipulations mainly relies on manual supervision.

[0068] In the description of the embodiments of the present application, it should be noted that in the description of the present application, the terms such as "inside", "outside", etc., indicating the direction or positional relationship are based on the direction or positional relationship shown in the drawings. This is only for the convenience of description and does not indicate or imply that the device or component must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present application.

[0069] In the description of the present application, the description referring to terms such as "one embodiment", "some embodiments", "in this embodiment", "specific examples", or "some examples", etc., means that the specific features, mechanisms, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, mechanisms, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0070] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A decision support system for biomass pyrolysis process, characterized in that: include: A biomass pyrolysis device, comprising a feeding device, a pyrolysis reactor, a gas collecting device and a heat recovery device forming a loop structure; A collection device, used to obtain temperature information, pressure information, and gas composition information in the pyrolysis reactor and the gas collection device respectively; The control module is configured to obtain the temperature prediction result and pressure prediction result of the pyrolysis reactor at the next time point, the temperature adjustment amplitude required to return the temperature in the pyrolysis reactor to a safe range, and the pressure adjustment amplitude required to return the pressure in the pyrolysis reactor to a safe range by using the temperature information, pressure information, and gas composition information currently obtained by the acquisition device through a deep neural network algorithm, and issue a request decision instruction when the temperature adjustment amplitude and / or the pressure adjustment amplitude exceed a threshold value, and respectively adjust the temperature in the pyrolysis reactor according to the temperature adjustment amplitude and adjust the pressure in the pyrolysis reactor according to the pressure adjustment amplitude when both the temperature adjustment amplitude and the pressure adjustment amplitude do not exceed the threshold value; a large language model module, configured to issue an early warning to a supervisor after receiving the request decision instruction, and provide operation suggestions for the supervisor to make a decision, and adjust the temperature and pressure in the pyrolysis reactor through the control module according to the supervisor's decision; in, The control module is electrically connected to the acquisition device, and the large language model module is electrically connected to the control module.

2. The decision support system for biomass pyrolysis process according to claim 1, characterized in that: The collection device comprises: A first temperature sensor, electrically connected to the control module, for obtaining temperature information in the pyrolysis reactor; a second temperature sensor, electrically connected to the control module, for acquiring temperature information in the gas collection device; A first pressure sensor, electrically connected to the control module, for obtaining pressure information in the pyrolysis reactor; A second pressure sensor, electrically connected to the control module, for obtaining pressure information in the gas collection device; a first gas composition analyzer, electrically connected to the control module, for obtaining gas composition information in the pyrolysis reactor; The second gas composition analyzer is electrically connected to the control module and is used to obtain gas composition information in the gas collection device.

3. The decision support system for biomass pyrolysis process according to claim 2, characterized in that: The control module comprises: A multi-layer perception algorithm unit is configured to obtain the temperature prediction result, the pressure prediction result, the temperature adjustment range and the pressure adjustment range by using the temperature information, pressure information and gas composition information currently obtained by the acquisition device through a multi-layer perception algorithm; A central processing unit is configured to determine whether the temperature adjustment amplitude and the pressure adjustment amplitude both do not exceed a threshold value; If so, issuing a first adjustment instruction according to the temperature adjustment range and issuing a second adjustment instruction according to the pressure adjustment range respectively; If not, issuing the request decision instruction; a temperature regulating device, configured to perform temperature regulation in the pyrolysis reactor according to the first adjustment instruction; a pressure regulating device, configured to perform pressure regulation in the pyrolysis reactor according to the second adjustment instruction; in, The multi-layer perception algorithm unit is electrically connected to the first temperature sensor, the second temperature sensor, the first pressure sensor, the second pressure sensor, the first gas composition analyzer and the second gas composition analyzer at the same time, the central processing unit is electrically connected to the large language model module and the multi-layer perception algorithm unit at the same time, the temperature regulating device is electrically connected to the central processing unit, and the pressure regulating device is electrically connected to the central processing unit.

4. The decision support system for biomass pyrolysis process according to claim 3, characterized in that: The multi-layer perception algorithm unit comprises: A preprocessing device, configured to perform normalization processing on the temperature information, pressure information, and gas composition information obtained by the acquisition device to obtain a normalized result; A multi-layer perceptron is configured to obtain the temperature prediction result, the pressure prediction result, the temperature adjustment amplitude and the pressure adjustment amplitude by using the normalization result through a multi-layer perceptron algorithm; in, The preprocessing device is electrically connected to the first temperature sensor, the second temperature sensor, the first pressure sensor, the second pressure sensor, the first gas composition analyzer and the second gas composition analyzer at the same time, and the multilayer sensor is electrically connected to the preprocessing device and the central processing unit at the same time.

5. The decision support system for biomass pyrolysis process according to claim 4, characterized in that: The multi-layer perception algorithm unit also includes: an optimizer, electrically connected to the multilayer perceptron, and configured to optimize parameters of the multilayer perceptron through a mean square error loss function; The optimization accelerator is electrically connected to the optimizer and is configured to perform operation acceleration on the parameter optimization of the optimizer through an operation acceleration algorithm.

6. The decision support system for biomass pyrolysis process according to claim 4 or 5, characterized in that: The large language model module includes: A language assistant unit is configured to issue an early warning upon receiving the request decision instruction, and generate an operation suggestion for the supervisor to select using the request decision instruction through the Char AI language assistant algorithm; A display unit is configured to display the warning and operation suggestions issued by the language assistant unit, and to adjust the temperature in the pyrolysis reactor through the temperature regulating device and / or adjust the pressure in the pyrolysis reactor through the pressure regulating device after the supervisor makes a decision; in, The language assistant unit is electrically connected to the central processing unit, and the display unit is electrically connected to the language assistant unit, the temperature regulating device and the pressure regulating device at the same time.

Images