Intelligent controller for sulfuric acid production and intelligent control method for acid production

Abstract

The invention provides an intelligent controller for sulfuric acid production and an intelligent control method for acid making. The method comprises the following steps: 1, establishing an intelligent control model capable of self-identifying environment deviation; 2, evaluating and testing the intelligent control model; the method specifically comprises the steps that a control instruction and a monitoring instruction of production data serve as model input, model output and an actual result are subjected to curve visual comparison, and the accuracy of a model is verified; and 3, switching the successfully tested intelligent control model to a use state, issuing a control strategy formed by a new intelligent control model as an instruction to an actual production control process, and performing acid making control. The invention discloses an intelligent control model capable of self-identifying environment deviation, which not only improves the automation level of an acid production system, but also enhances the capability of coping with external changes and internal fluctuations. The intelligent level and the working efficiency of the control system can be greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of acid production, and in particular to an intelligent controller for sulfuric acid production and an intelligent control method for acid production. Background Art

[0002] Sulfuric acid is one of the most important and largest-used inorganic strong acids in the chemical industry. It is widely used in fertilizer manufacturing, petroleum refining, metal processing, pharmaceuticals and other fields. The process of sulfuric acid preparation includes incineration, conversion, dry absorption and tail gas treatment procedures.

[0003] The traditional control method is to use a basic model, that is, a control model established based on historical data and machine learning or deep learning. The control strategy formed is to issue instructions to a general intelligent controller to control production. Although it has certain control effects and convenience, the actual situation is usually more complicated. Due to hardware aging and some external factors, the working conditions of the system often change. Such changes will cause the model to be inaccurate, making the optimal parameters searched in the model no longer valid. The direct result is a decrease in control effect. The more serious consequence is shutdown, and even accidents. Therefore, this control method requires regular shutdowns, and control experts are required to debug and test the controller on site to ensure the safe operation of the control system. This process will introduce a lot of labor costs, downtime costs, and maintenance costs. In fact, such costs can be avoided. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention aims to provide a shorthand auxiliary device for learning, which can solve the existing problems.

[0005] In order to achieve the above object, the technical solution of the present invention is as follows:

[0006] The present invention is realized through the following technical scheme: an intelligent control system, the intelligent control system is bidirectionally connected to a universal intelligent controller; the universal intelligent controller is bidirectionally connected to a PLC; the universal intelligent controller and the PLC are bidirectionally connected to a control signal switching module, and the control signal switching module is connected to a frequency conversion control cabinet and a valve group; the frequency conversion control cabinet and the valve group are connected to a sensor group, and the sensor group is connected to the PLC.

[0007] An intelligent control method for acid production, which is executed based on the intelligent controller described above; comprising:

[0008] Step 1: Establish an intelligent control model that can self-identify environmental deviations; specifically include:

[0009] Identify the degree of deviation of the environment,

[0010] Based on the degree of identified deviation, environmental adaptive training is performed, and a new intelligent control model capable of self-identifying environmental deviation is obtained by training on the original basic model.

[0011] Step 2: Evaluate and test the intelligent control model; specifically include:

[0012] Using the control instructions and monitoring instructions of production data as model input, the model output is compared with the actual results by visual curve to verify the accuracy of the model;

[0013] Step 3: Switch the successfully tested intelligent control model to the use state, and use the control strategy formed by the new intelligent control model as an instruction to send it to the actual production control process to control acid production.

[0014] Furthermore, the deviation degree of the recognition environment includes:

[0015] For the current acid production environment, a model baseline of the current production environment is set to characterize the state of the current environment;

[0016] Obtain various data in the sulfuric acid production process through sensor groups and real-time data acquisition equipment, and perform data processing;

[0017] Taking the establishment time of the model baseline as the starting time, based on all processed data after that time, using machine learning and deep learning algorithms, the model predicts the environmental status under production operation conditions, and the degree of environmental deviation is identified by comparing the deviation between the predicted deduction data and the historical actual data.

[0018] Furthermore, the deviation degree of the identification environment includes: the various data include temperature, pressure and flow; the data processing includes data cleaning, preprocessing and feature extraction.

[0019] Furthermore, the actual production control process includes: controlling the incineration temperature of the incinerator in the incineration process by controlling the sulfur feed amount and the air fan speed; stably controlling the heat exchange and heating of the converter in the conversion process by controlling the pressure inside the converter and the process gas; controlling the sulfuric acid flow and circulating water flow in the dry absorption process by controlling the valve; and controlling the high-level water tank liquid level in the purification process by controlling the water valve and the water pump.

[0020] Furthermore, the basic model includes a prediction basic model of the acid production process established based on historical data and a mechanism model using a machine learning or deep learning method.

[0021] An acid-making intelligent control system, comprising:

[0022] A basic model module, the basic model module is used to receive the current state and the control amount as input, and calculate and generate the state prediction at the next moment;

[0023] An adaptive residual fine-tuning module, which is used to receive the current state and control amount as input and accumulate input data; dynamically adjust the output of the basic model based on the input information and the historical error information of the basic model to generate a more accurate state prediction; and continuously learn and optimize the residual module based on a continuous learning algorithm;

[0024] Self-evolving modules, including

[0025] Data acquisition unit: real-time acquisition of system operation data, including status, control quantity and actual output;

[0026] Model training unit: Based on the collected data and continuous learning algorithm, the adaptive residual fine-tuning module is continuously trained to ensure that the overall model always adapts to the latest working conditions;

[0027] Model update unit: automatically determines whether the currently running model needs to be replaced, and replaces the currently running model when necessary.

[0028] A computer-readable storage medium stores a computer program, wherein the computer program stores instructions, and when the instructions are executed by one or more processors, the one or more processors perform any one of the methods described above.

[0029] Compared with the prior art, the beneficial effects of the present invention include:

[0030] The present invention discloses an intelligent control model capable of self-identifying environmental deviations, which not only improves the automation level of the acid production system, but also enhances the ability to cope with external changes and internal fluctuations; it can achieve a significant improvement in the intelligence level and work efficiency of the control system, and can maintain or further improve control efficiency, safety and reliability while reducing production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The disclosure of the present invention is described with reference to the accompanying drawings. It should be understood that the drawings are only for illustrative purposes and are not intended to limit the scope of protection of the present invention. Among them:

[0032] Figure 1 This is a block diagram of an intelligent controller system for sulfuric acid production according to the present invention;

[0033] Figure 2 This is a schematic diagram of the process of the intelligent control method for acid production according to the implementation case of the present invention; DETAILED DESCRIPTION

[0034] It is easy to understand that according to the technical solution of the present invention, without changing the essential spirit of the present invention, a person skilled in the art can propose a variety of interchangeable structural modes and implementation modes. Therefore, the following specific implementation modes and drawings are only exemplary descriptions of the technical solution of the present invention, and should not be regarded as the whole of the present invention or as a limitation or restriction to the technical solution of the present invention.

[0035] The present invention provides an intelligent controller for sulfuric acid production. The system block diagram of the intelligent controller is as follows: Figure 1 As shown, including:

[0036] An intelligent control system, the intelligent control system is used to issue comprehensive instructions and receive comprehensive signals; the intelligent control system is bidirectionally connected to a universal intelligent controller; the universal intelligent controller is bidirectionally connected to a PLC, and the universal intelligent controller communicates with the PLC; the universal intelligent controller and the PLC are bidirectionally connected to a control signal switching module, specifically, the universal intelligent controller is connected to the control signal switching module through a monitoring unit, the PLC sends start and stop commands and frequency conversion commands to the control signal switching module, and the control signal switching module is connected to a frequency conversion control cabinet and a valve group; the frequency conversion control cabinet and the valve group are connected to a sensor group, and the sensor group is connected to the PLC.

[0037] The use of intelligent controllers for acid production control is because traditional PLC controllers (general intelligent controllers) cannot run neural network models, so dedicated servers need to be deployed to support them. The AI ​​algorithm server (intelligent controller) of the present invention is a dedicated server that can run neural network control models developed specifically for the needs of intelligent industrial production. The intelligent controller of the present invention has a built-in complete control model deployment, debugging, and operation service system, which provides acceleration capabilities for the operation of the algorithm model and can meet the needs of industrial high-frequency processing. The intelligent controller of the present invention is based on proprietary fault-tolerant synchronization technology, and has H&i Server to realize the dual-active synchronous operation of redundant host units, and a single operating environment with precise and consistent memory dynamic data, I / O requests, and disk network read and write. Simply put, it is a fault-tolerant computing and storage solution built on the hardware bottom layer. Completely independent of the operating system and application, users can maintain the fault-tolerant server like using a stand-alone machine. If any component fails, the business system switches in zero seconds. The H&i Server redundant fault-tolerant server is designed to prevent downtime, data loss, and business interruption. This integrated, easy-to-operate, high-reliability server system is the best solution to replace traditional clusters and stand-alone servers. Compared with traditional server cluster solutions, the H&i Server system eliminates the reliance on expensive SAN storage devices and foreign cluster software. The large-capacity storage system with built-in real-time data mirroring protection can achieve safe deployment of business systems. The integrated virtualization function realizes online expansion of configuration, parallel operation of multiple operating systems and efficient integration of server resources, saving IT facility investment costs.

[0038] The present invention also provides an intelligent control method for acid production, which is executed based on the above-mentioned intelligent controller and comprises the following steps:

[0039] Step 1: Establish an intelligent control model that can self-identify environmental deviations; specifically include:

[0040] 1.1. Identify the degree of environmental deviation; In intelligent control, the ability of the model to self-identify environmental deviation is a key and important capability. In industrial production, especially in complex and high-risk processes such as sulfuric acid production, maintaining the stability and consistency of the production environment is crucial to production efficiency and product quality. Therefore, the intelligent control model that can self-identify environmental deviation not only improves the automation level of the production system, but also enhances the ability to cope with external changes and internal fluctuations.

[0041] The model's ability to self-identify environmental deviations depends on two parts, including benchmarks and data acquisition, which are described as follows: First: Benchmarks

[0042] Establishing a benchmark is an indispensable step in the data analysis and machine learning process. It not only helps to evaluate and optimize model performance, but also supports business decision-making and continuous improvement processes. Before the model is put into production control, the current state of the environment is learned and the model baseline of the current production environment is set to characterize the state of the current environment by using machine learning and deep learning algorithms for the current production environment.

[0043] Second: Data Acquisition

[0044] The model's ability to self-identify environmental excursions is based on advanced data acquisition, processing and analysis technologies. The system acquires various key parameters in the sulfuric acid production process, including but not limited to temperature, pressure and flow, through a sensor group and real-time data acquisition equipment. These data are sent to the data processing unit, and after cleaning, preprocessing and feature extraction, a high-quality data stream is formed.

[0045] Therefore, the simulation and strategy control model established by the software package has the ability to identify environmental deviations and can perceive environmental changes in real time. Taking the model benchmark establishment time as the starting time, based on all analyzed and processed historical data after this time, using machine learning and deep learning algorithms, the model predicts the environmental state under production operating conditions and identifies the degree of environmental deviation by comparing the deviation between the predicted deduction data and the historical actual data.

[0046] 1.2. Based on the degree of identified offset, environmental adaptive training is performed. A new intelligent control model that can self-identify environmental offset is trained on the original basic model. Specifically, after identifying environmental offset, automatic / manual environmental adaptive training can be performed. It supports continued training on the original basic model, and a new model can also be retrained. The training data will be trained using the latest historical data. The new model will re-establish a new benchmark to facilitate subsequent environmental offset identification.

[0047] The basic model is the model designed in the incineration end process, conversion process, dry absorption process and purification process of the incinerator in the acid production process;

[0048] Intelligent control of the incinerator process. As the core equipment of the incineration process, the incinerator focuses on parameters such as sulfur feed rate and air fan speed. The optimization control of the incineration process aims to achieve one-key load increase or decrease without reducing combustion efficiency. Combining mechanism models, industry expertise and AI data models, the main factors affecting the control objectives are analyzed. These models can realize automatic control of the incinerator feed, thereby stabilizing the control and optimizing the key temperatures in the incineration process to achieve energy saving and consumption reduction.

[0049] Intelligent control of the conversion process and the optimization control of the conversion process are aimed at maximizing the conversion effect. The mechanism model, industry expertise and neural network model are combined to analyze the main factors affecting the control target. To this end, relevant temperature feedforward feedback models, operating condition prediction neural network models and energy consumption optimization models are established to achieve autonomous control of the converter heat exchange, heating and other processes, stabilize the key temperature control in the conversion process, and achieve the goal of energy saving and consumption reduction. The specific control structure is as follows: The pressure and process gas temperature inside the converter are adjusted through coordinated control to maintain the conversion efficiency. Pressure regulation is mainly achieved through the adjustment of the fan, and the motor frequency of the SO2 fan is adjusted to control the tail of the incinerator to maintain a slightly negative pressure state. The inlet temperature of each section of the converter is mainly achieved through the adjustment of the electric heater, direct condensation valve and heat exchanger short-circuit valve.

[0050] Intelligent control of dry absorption process and optimization control of dry absorption process aim to achieve the best dry absorption effect and product output. This process combines mechanism model, industry experience and AI data model to analyze the key factors affecting the control target and establish a feedforward feedback model of liquid level and temperature. By automatically controlling parameters such as sulfuric acid flow rate and circulating water flow rate, the stability and efficiency of process operation are ensured. During the optimization process, special attention should be paid to the cyclic exchange between dry acid, primary absorption acid and secondary absorption acid to achieve an acid equilibrium state. By precisely adjusting the flow rate and temperature of each link, the effective recycling of acidic substances in the system is guaranteed, thereby improving the overall efficiency of the process and resource utilization.

[0051] The purification process section is intelligently controlled. The main purpose of the purification process is washing with dilute sulfuric acid and electro-fog demisting of FRP. The process is relatively simple and the main control point is the liquid level of the high-level water tank.

[0052] Therefore, it can be seen that the basic model is based on historical data, mechanism model and the use of machine learning or deep learning methods (such as support vector machine, random forest, neural network, etc.) to establish a basic prediction model for each process section of the acid production process. Combined with the prediction basic model, the control algorithm (such as PID, model predictive control MPC, etc.) is designed to achieve stable and efficient production process control.

[0053] Step 2: Evaluate and test the intelligent control model; specifically include:

[0054] The trained model can be tested for a period of time to further verify the model's predictive ability. During the test, the control instructions and monitoring instructions of the production data are used as the model input, and the model output is compared with the actual results by visual curve to verify the accuracy of the model. During the test, the basic model and the new intelligent control model will run simultaneously, but the control instructions of the new intelligent control model will not be directly issued to the actual production process.

[0055] The control instructions include but are not limited to start / stop instructions and frequency conversion instructions; the monitoring instructions include but are not limited to obtaining data from a sensor group.

[0056] Step 3: Switch the successfully tested intelligent control model to the use state, and use the control strategy formed by the new intelligent control model as an instruction to send it to the actual production control process to control acid production.

[0057] The present invention also provides an acid production intelligent control system, comprising:

[0058] A basic model module, the basic model module is used to receive the current state and the control amount as input, and calculate and generate the state prediction at the next moment;

[0059] An adaptive residual fine-tuning module, which is used to receive the current state and control amount as input and accumulate input data; dynamically adjust the output of the basic model based on the input information and the historical error information of the basic model to generate a more accurate state prediction; and continuously learn and optimize the residual module based on a continuous learning algorithm;

[0060] Self-evolving modules, including

[0061] Data acquisition unit: real-time acquisition of system operation data, including status, control quantity and actual output;

[0062] Model training unit: Based on the collected data and continuous learning algorithm, the adaptive residual fine-tuning module is continuously trained to ensure that the overall model always adapts to the latest working conditions;

[0063] Model update unit: automatically determines whether the currently running model needs to be replaced, and replaces the currently running model when necessary.

[0064] This technical solution is a set of adaptive AI algorithms based on self-evolving models, which has a higher level of automation and intelligence than traditional models. This solution is based on an artificial intelligence method of continuous learning, also known as lifelong learning. During operation, it can combine observation data to adapt to the basic model, the residual error of the basic model, the state control quantity of the next moment, the state control quantity + self-evolving model, the state of the traditional model at the next moment, and autonomously adapt to changes in working conditions. Manual access is required to upgrade the model. The downtime is one to two weeks and the upgrade time is two to three days. Skilled operators are required to monitor the artificial intelligence model in real time instead of manually updating its own model according to actual conditions, so as to achieve continuous and rapid adaptation to changes in system working conditions. This solution aims to realize automatic upgrades and adaptive adjustments of controllers through self-evolving models, thereby reducing manual intervention and maintenance costs and improving the response speed and accuracy of the system.

[0065] The input and output interfaces of the self-evolution model are consistent with those of the traditional model. The state quantity and control quantity are input, and the state quantity of the next moment is output. Unlike the traditional model, the self-evolution model introduces an additional adaptive residual fine-tuning module on top of the basic model in the traditional model to correct the output of the basic model in real time. The traditional model is consistent with the basic model in the self-evolution model to be backward compatible, maintaining the advantage of the easy acquisition of the traditional model. The adaptive residual fine-tuning module introduced in the self-evolution model is continuously updated using the continuous learning method to continuously adapt to and match the changes in the system. The core mechanism of the self-evolution model is: continuously accumulate the latest observation and control data, and then use the collected data to update the residual fine-tuning module. The goal of the update is: output a residual quantity so that the sum of the residual quantity and the basic model output can be consistent with the real data. The model can quickly adapt to new data and automatically determine whether a rapid upgrade of the model is required. The benefits of correcting the basic model based on the residual module include: the amount of data required for model update is small, and only sparse updates are required; the complexity required for the residual model is low, which can meet the needs of rapid adaptation.

[0066] The following is a further introduction to the system architecture, module functions and self-evolution mechanism in the solution.

[0067] 1) System architecture

[0068] The system architecture of the self-evolving model includes the following two key parts:

[0069] Basic model: used to process basic states and control quantities and generate preliminary state predictions.

[0070] Adaptive residual fine-tuning module: Based on historical data and current operating conditions, the basic model output is fine-tuned to improve the accuracy of model predictions.

[0071] 2) Module function description

[0072] Basic model:

[0073] a) Receive the current state and control quantity as input.

[0074] b) Calculate and generate the state prediction for the next moment.

[0075] Adaptive residual fine-tuning module:

[0076] a) Receive the current state and control quantity as input and accumulate input data.

[0077] b) Based on the input information and the historical error information of the basic model, the output of the basic model is dynamically adjusted to generate more accurate state predictions.

[0078] c) Based on the continuous learning algorithm, the residual module is continuously learned and optimized to further improve the accuracy of the overall model.

[0079] 3) Self-evolution mechanism

[0080] Data acquisition: Real-time acquisition of system operation data, including status, control quantity and actual output.

[0081] Model training: Based on the collected data and continuous learning algorithm, the adaptive residual fine-tuning module is continuously trained to ensure that the overall model always adapts to the latest working conditions.

[0082] Model update: Automatically determine whether the currently running model needs to be replaced, and replace the currently running model when necessary.

[0083] Comparison between traditional model and self-evolution model Since the self-evolution model introduces a residual model based on the traditional model, the residual module will be initialized to all zeros at the beginning. Therefore, in the factory stage, the self-evolution model inherits all the advantages of the traditional model and can efficiently solve the control task. When the working conditions and hardware change, the self-evolution model can continue to follow the changes of the system to avoid performance degradation, which is difficult for the traditional model to achieve.

[0084] Such a technical advantage can bring improvements in many aspects, such as process level, economic level, and labor level. When the system working conditions change, the traditional model needs the intervention of control experts to upgrade the model, and the upgrade process takes a lot of time, ranging from a week to half a month. During this period, continuous production cannot be guaranteed, which will greatly affect the output. In contrast, the self-evolution model can continuously accumulate historical data and continuously update the model. Therefore, even if the system changes, it will not cause significant performance degradation, which can greatly reduce the frequency of shutdown adjustments. On the other hand, if the system needs to replace parts (such as motors), the traditional model needs to be trained from scratch to adapt to the new parts. This process also requires a long shutdown, while the self-evolution model only needs to learn the residual fine-tuning module, which only needs a small amount of adjustment to adapt to the new parts, greatly reducing downtime. Finally, in daily maintenance, the self-evolution model has a lower failure rate due to its strong adaptive ability, so it no longer requires a large number of on-site personnel for maintenance and monitoring, and labor costs can also be greatly reduced.

[0085] The system of the present invention introduces a self-evolution model solution, which can significantly improve the intelligence level and work efficiency of the control system by introducing an adaptive residual fine-tuning module and an automatic model upgrade mechanism, and can maintain or further improve control efficiency, safety and reliability while reducing production costs.

[0086] The present invention discloses an intelligent control model capable of self-identifying environmental deviations, which not only improves the automation level of the production system, but also enhances the ability to cope with external changes and internal fluctuations; it can achieve a significant improvement in the intelligence level and work efficiency of the control system, and can maintain or further improve control efficiency, safety and reliability while reducing production costs.

[0087] Those of ordinary skill in the art will appreciate that the unit-level algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this article.

[0088] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0089] In the several embodiments provided herein, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic, for example, the division of the units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or can be electrical, mechanical or other forms of connection.

[0090] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of this article.

[0091] In addition, each functional unit in each embodiment of this invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional unit.

[0092] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this article is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of this article. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0093] Specific embodiments are used in this article to illustrate the principles and implementation methods of this article. The description of the above embodiments is only used to help understand the methods and core ideas of this article. At the same time, for general technicians in this field, according to the ideas of this article, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on this article.

Claims

1. An intelligent controller for sulfuric acid production, characterized in that: The intelligent controller comprises: An intelligent control system, wherein the intelligent control system is bidirectionally connected to a universal intelligent controller; the universal intelligent controller is bidirectionally connected to a PLC; the universal intelligent controller and the PLC are bidirectionally connected to a control signal switching module, and the control signal switching module is connected to a frequency conversion control cabinet and a valve group; the frequency conversion control cabinet and the valve group are connected to a sensor group, and the sensor group is connected to the PLC.

2. An intelligent control method for acid production, which is executed based on the intelligent controller according to claim 1; characterized in that: include: Step 1: Establish an intelligent control model that can self-identify environmental deviations; specifically include: Identify the degree of deviation of the environment, Based on the degree of identified deviation, environmental adaptive training is performed, and a new intelligent control model capable of self-identifying environmental deviation is obtained by training on the original basic model. Step 2: Evaluate and test the intelligent control model; specifically include: Using the control instructions and monitoring instructions of production data as model input, the model output is compared with the actual results by visual curve to verify the accuracy of the model; Step 3: Switch the successfully tested intelligent control model to the use state, and use the control strategy formed by the new intelligent control model as an instruction to send it to the actual production control process to control acid production.

3. The intelligent control method for acid production according to claim 2, characterized in that: The deviation degree of the identification environment includes: For the current acid production environment, a model baseline of the current production environment is set to characterize the state of the current environment; Obtain various data in the sulfuric acid production process through sensor groups and real-time data acquisition equipment, and perform data processing; Taking the establishment time of the model baseline as the starting time, based on all processed data after that time, using machine learning and deep learning algorithms, the model predicts the environmental status under production operation conditions, and the degree of environmental deviation is identified by comparing the deviation between the predicted deduction data and the historical actual data.

4. The intelligent control method for acid production according to claim 3, characterized in that: The deviation degree of the identification environment includes: the various data include temperature, pressure and flow; the data processing includes data cleaning, preprocessing and feature extraction.

5. The intelligent control method for acid production according to claim 1, characterized in that: The actual production control process includes: controlling the incineration temperature of the incinerator in the incineration process by controlling the sulfur feed amount and the air fan speed; controlling the heat exchange and heating of the converter in the conversion process by controlling the pressure inside the converter and the process gas; controlling the sulfuric acid flow and circulating water flow in the dry absorption process by controlling the valve; and controlling the high-level water tank liquid level in the purification process by controlling the water valve and the water pump.

6. The intelligent control method for acid production according to claim 5, characterized in that: The basic model includes a prediction basic model of the acid production process established based on historical data and a mechanism model using a machine learning or deep learning method.

7. An intelligent control system for acid production, characterized in that: include: A basic model module, the basic model module is used to receive the current state and the control amount as input, and calculate and generate the state prediction at the next moment; An adaptive residual fine-tuning module, the adaptive residual fine-tuning module is used to receive the current state and the control amount as input and accumulate input data; Based on the input information and the historical error information of the basic model, the output of the basic model is dynamically adjusted to generate more accurate state predictions; Based on the continuous learning algorithm, the residual module is continuously learned and optimized; Self-evolving modules, including Data acquisition unit: real-time acquisition of system operation data, including status, control quantity and actual output; Model training unit: Based on the collected data and continuous learning algorithm, the adaptive residual fine-tuning module is continuously trained to ensure that the overall model always adapts to the latest working conditions; Model update unit: automatically determines whether the currently running model needs to be replaced, and replaces the currently running model when necessary.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program stores instructions which, when executed by one or more processors, cause the one or more processors to perform the method of any one of claims 2 to 6.

Images