Coal waste gas comprehensive utilization dynamic optimization method and system based on big data

Abstract

The invention belongs to the field of coal waste gas treatment, and particularly relates to a coal waste gas comprehensive utilization dynamic optimization method and system based on big data, and the dynamic optimization method comprises the following specific steps: S1, collecting and preprocessing multi-source data; s2, big data analysis and model construction: constructing a waste gas generation prediction model, a resource recovery potential evaluation model and a waste gas treatment effect evaluation model; and S3, dynamic optimization decision making and implementation: monitoring key indexes in real time and performing early warning, and through accurate waste gas generation prediction and real-time equipment performance monitoring, waste gas treatment process parameters can be adjusted in time, it is ensured that waste gas treatment equipment is always in an optimal operation state, the waste gas treatment efficiency is effectively improved, the pollutant emission concentration is reduced, and the energy consumption is reduced. The waste gas treatment quality is improved, so that the emission index more stably meets the environmental protection standard.

Description

Technical Field

[0001] The present invention belongs to the field of coal waste gas treatment, and specifically relates to a dynamic optimization method and system for comprehensive utilization of coal waste gas based on big data. Background Art

[0002] Coal is widely used in energy production and industrial manufacturing. However, when it burns, it will generate certain waste gas. Coal waste gas is a mixed gas emitted during the processes of coal combustion, processing, etc. Its composition is complex and it is highly harmful to the environment. Coal waste gas contains a large amount of pollutants, such as sulfur dioxide (SO2), nitrogen oxides (NO x ), particulate matter, etc., seriously endangering the environment and human health.

[0003] In the prior art, during long-term use and observation, it is found that the existing coal waste gas treatment methods often focus on end treatment, and the treatment process is relatively fixed, lacking the ability to respond to and optimize the dynamic factors in the production process in real time. At the same time, the recyclable resources contained in the waste gas, such as sulfur, nitrogen, etc., have not been fully and effectively utilized. Therefore, the present invention provides a dynamic optimization method and system for comprehensive utilization of coal waste gas based on big data. Summary of the Invention

[0004] In order to make up for the deficiencies of the prior art and solve at least one of the technical problems proposed in the background art.

[0005] The technical solution adopted by the present invention to solve its technical problems is as follows: A dynamic optimization method for comprehensive utilization of coal waste gas based on big data according to the present invention is characterized in that the specific steps of the dynamic optimization method are as follows:

[0006] S1: Multi-source data collection and preprocessing: Collect multi-source data such as coal quality parameters, combustion process parameters, waste gas composition and emission parameters, equipment operation status parameters, and environmental parameters, and perform data cleaning, normalization, standardization, and feature extraction processing on the data;

[0007] S2: Big data analysis and model construction: Construct a waste gas generation prediction model, a resource recovery potential evaluation model, and a waste gas treatment effect evaluation model;

[0008] S3: Dynamic optimization decision-making and implementation: Real-time monitor key indicators and give early warnings, generate optimization strategies and implement them, and adjust the optimization strategies according to the feedback.

[0009] Preferably, according to the waste gas generation prediction model described in S2, use machine learning algorithms to predict the waste gas generation amount and composition considering coal characteristics, combustion conditions, and production load factors.

[0010] Preferably, according to the resource recovery potential evaluation model described in S2, analyze the content and distribution of recyclable resources in the waste gas, and evaluate the feasibility and economic benefits of the recovery plan in combination with market demand, resource prices, and recovery technology costs.

[0011] Preferably, according to the waste gas treatment effect evaluation model described in S2, evaluate the equipment performance and efficiency based on the operating data of the waste gas treatment equipment and the emission indicators of the treated waste gas, and predict the equipment failure risk.

[0012] Preferably, a dynamic optimization system for comprehensive utilization of coal waste gas based on big data, characterized in that it includes a central processing unit module, a power supply module, a data acquisition module, a data transmission module, a big data processing module, and an execution module;

[0013] The central processing module and the power supply module are electrically connected, the power supply module can supply power to the central processing module, and the central processing module can control the current output in the power supply module;

[0014] The central processing unit module and the data acquisition module are signal-connected, the data acquisition module consists of sensors, a data acquisition terminal, and a communication component, which collect data and transmit it to the data transmission module;

[0015] The central processing module and the data transmission module are signal-connected, the data transmission module can perform encryption, compression, and error correction processing on the data, has data caching and flow control functions, and transmits the data to the big data processing module;

[0016] The central processing module and the big data processing module are signal-connected, the big data processing module includes a data storage sub-module, a data processing sub-module, a model management sub-module, and an optimization decision sub-module, which are used to store, process, and analyze data, build and manage models, and generate optimization strategies;

[0017] The central processing module and the execution module are signal-connected, the execution module consists of a coal combustion device, a waste gas treatment device, a resource recovery device, and a control system, receives the optimization strategy and implements it, and feeds back the equipment operation status and treatment effect.

[0018] Preferably, the sensors of the data acquisition module are distributed at key positions in coal production and waste gas treatment, and the data acquisition terminal preprocesses the sensor signals.

[0019] Preferably, the data transmission module uses the AES encryption algorithm, GZIP compression algorithm, and CRC error correction code technology to transmit data through the MQTT or HTTP protocol.

[0020] Preferably, the data storage submodule of the big data processing module adopts distributed file system and database technology, the data processing submodule uses MapReduce and Spark frameworks, the model management submodule uses a machine learning framework to build and manage models, and the optimization decision submodule uses optimization algorithms to generate optimization strategies.

[0021] Preferably, the control system of the execution module adjusts the equipment operating parameters according to the received control instructions to achieve optimized control of the comprehensive utilization process of coal waste gas.

[0022] The beneficial effects of the present invention are as follows:

[0023] 1. The present invention describes a dynamic optimization method and system for comprehensive utilization of coal waste gas based on big data. Through accurate waste gas generation prediction and real-time equipment performance monitoring, it can timely adjust waste gas treatment process parameters to ensure that the waste gas treatment equipment is always in the best operating state, effectively improve waste gas treatment efficiency, reduce pollutant emission concentration, improve waste gas treatment quality, and make emission indicators more stably meet environmental protection standards.

[0024] 2. The present invention describes a dynamic optimization method and system for comprehensive utilization of coal waste gas based on big data. By using a resource recovery potential assessment model, the value of recyclable resources in waste gas is deeply explored, resource recovery plans are optimized, resource recovery rates are improved, and potential resources in waste gas are converted into economic benefits, thereby achieving efficient resource utilization and turning waste into treasure.

[0025] The present invention integrates multi-source data such as coal production, waste gas treatment, environmental monitoring and market, and uses advanced big data analysis technology and machine learning algorithms for in-depth mining. By constructing a multi-factor waste gas generation prediction model, a resource recovery potential assessment model and a waste gas treatment effect assessment model, it achieves a comprehensive and accurate understanding and prediction of the comprehensive utilization process of coal waste gas. Moreover, based on real-time data and prediction models, a multi-objective optimization algorithm is used to generate a dynamic optimization strategy, which can adjust the waste gas treatment and resource recovery plans in real time according to production conditions, environmental changes and market demand. This dynamic optimization decision-making mechanism breaks the static mode of traditional treatment methods, realizes the adaptive optimization of the system, and improves the flexibility and adaptability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The present invention will be further described below with reference to the accompanying drawings.

[0027] Figure 1 is a flow chart of the method of the present invention;

[0028] Figure 2 It is a schematic diagram of the system flow in the present invention. DETAILED DESCRIPTION

[0029] In order to make the technical means, creative features, achieved objectives and effects realized by the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

[0030] As Figure 1 shown, a dynamic optimization method for comprehensive utilization of coal waste gas based on big data according to an embodiment of the present invention, the specific steps of the dynamic optimization method are as follows:

[0031] S1: Multi-source data collection and preprocessing: Collect multi-source data of coal quality parameters, combustion process parameters, waste gas composition and emission parameters, equipment operation status parameters, and environmental parameters, and perform data cleaning, normalization, standardization, and feature extraction processing on the data;

[0032] S2: Big data analysis and model construction: Construct a waste gas generation prediction model, a resource recovery potential evaluation model, and a waste gas treatment effect evaluation model;

[0033] S3: Dynamic optimization decision-making and implementation: Real-time monitor key indicators and give early warnings, generate optimization strategies and implement them, and adjust the optimization strategies according to the feedback;

[0034] During operation, at S1, in each link of the coal industrial chain, sensors convert various physical and chemical quantities into electrical signals or digital signals based on physical, chemical, and biological principles. For example, using the thermoresistive effect, a temperature sensor can accurately convert the temperature change during coal combustion into a change in resistance value, and then output the corresponding electrical signal. At the exhaust gas outlet, an electrochemical gas sensor generates a current signal proportional to the gas concentration through the oxidation-reduction reaction between the target gas and the electrode to obtain exhaust gas composition data. The data acquisition terminal amplifies these signals to strengthen weak signals for subsequent processing; uses filtering technology to remove noise interference in the signals to ensure data accuracy; then converts the analog signals into digital signals through analog-to-digital conversion for digital system processing. Subsequently, according to specific communication protocols, such as Modbus or MQTT protocols, the data is packaged into a specified format and stably transmitted to the subsequent processing link via a wired network (such as industrial Ethernet) or a wireless network (such as 4G, 5G). When cleaning data, by setting reasonable threshold ranges, data consistency verification and other rules, noise data (such as abnormal high values generated by instantaneous sensor failures) and duplicate data (duplicate records caused by abnormal transmission) are identified and removed. Normalization processing eliminates the dimensional differences between different data features by mapping the data to a specific interval (such as [0, 1]). For example, the ash content (percentage) of coal and the exhaust gas flow rate (cubic meters per hour) are unified to the same scale. Standardization transforms the data based on the mean and standard deviation of the data to make the data have unified statistical characteristics. Feature engineering techniques extract key features that have a significant impact on the comprehensive utilization of exhaust gas and optimal decision-making from a large amount of data through methods such as correlation analysis and principal component analysis. For example, key features such as combustion temperature and excess air coefficient that play a key role in the generation of exhaust gas are extracted from complex combustion process data;

[0035] When constructing the waste gas generation prediction model in S2, the machine learning algorithm learns the complex non-linear relationships between input variables such as coal properties (e.g., volatile matter, fixed carbon content, etc.), combustion conditions (temperature, pressure, air flow rate), production load, etc. and the waste gas generation amount and composition from a large amount of historical data. Taking a neural network as an example, data enters from the input layer, undergoes non-linear transformations through multiple hidden layers (e.g., through the activation function ReLU), and the prediction result is obtained at the output layer. During the training process, using the backpropagation algorithm, according to the error between the predicted value and the actual value, the weights and biases between neurons are continuously adjusted to gradually reduce the error, so that the model can accurately capture the internal laws between these factors and waste gas generation. The resource recovery potential evaluation model first determines the content of recoverable resources (such as sulfur, nitrogen, carbon dioxide) in the waste gas through technical means such as spectral analysis and chromatographic analysis, and then uses the material flow analysis method to track the distribution path of resources in the waste gas treatment process. Combining the market demand trends for these resources (such as the demand fluctuations of sulfur in the chemical industry) and the economic indicators of relevant recovery technologies (equipment investment, operating costs, prices of recovered products), using the cost-benefit analysis model, combines the resource recovery amount with the above cost and benefit factors to evaluate the feasibility and economic benefits of different recovery schemes. The waste gas treatment effect evaluation model is based on the operating data of the waste gas treatment equipment (inlet and outlet pressure difference, flow rate, chemical agent usage, etc.) and the emission indicators of the treated waste gas (sulfur dioxide, nitrogen oxides, particulate matter concentration), and uses statistical analysis methods (such as calculating the mean and standard deviation to analyze data stability) and machine learning algorithms (such as support vector machines) to construct an evaluation model. The support vector machine divides the equipment operating state into categories such as high efficiency, medium efficiency, and low efficiency by finding an optimal classification hyperplane, quantifies the performance of the equipment, and predicts the change trend of the treatment efficiency of the equipment under different operating conditions and the possibility of faults through learning historical data;

[0036] In the real-time monitoring and early warning part of S3, the big data platform continuously receives real-time data from sensors in each link and compares it with the pre-set early warning thresholds. The early warning thresholds are determined based on environmental protection standards (such as the exhaust gas emission standards stipulated by the state), the safe operating range of equipment (such as the upper limit of pressure that the equipment can withstand), etc. Once the key indicators exceed the thresholds, the system immediately sends text messages through the SMS gateway, sends emails through the email server, or issues audible and visual alarms on the monitoring interface to send early warning signals to relevant personnel, informing them of the abnormal situation and possible impacts. When generating the optimization strategy, with the reduction of waste gas pollutants, the maximization of resource recovery, and the minimization of costs as multiple objectives, the real-time data and the results of the prediction model are used as inputs. The optimization algorithm (such as the multi-objective genetic algorithm) searches for the optimal solution in the solution space that meets various constraints (such as equipment operation restrictions, environmental protection regulations requirements, production process constraints). The multi-objective genetic algorithm simulates the biological evolution process and continuously iterates and optimizes through selection, crossover, and mutation operations to obtain a dynamic optimization strategy, including adjusting the operating parameters of combustion equipment (such as increasing the combustion temperature to promote complete combustion and reduce pollutant generation), optimizing the treatment process (such as changing the spraying mode of the desulfurization tower to improve desulfurization efficiency), selecting the best recovery plan (selecting the process for recovering sulfur or nitrogen based on market prices and resource content), etc. In the strategy implementation and feedback stage, the optimization strategy is converted into specific control instructions and sent to the corresponding equipment for execution through the industrial control system (such as the PLC control system). At the same time, the sensors on the equipment continue to monitor the operating status and treatment effect in real time, and transmit the feedback data back to the big data platform for evaluating the implementation effect of the strategy. If the expected goals are not achieved, for example, the exhaust gas emissions are still exceeding the standards or the resource recovery efficiency has not been improved, the system will re-adjust the optimization strategy and execute it again to form a closed-loop control and continuously improve the system performance.

[0037] As Figure 1 shown, according to the waste gas generation prediction model described in S2, the machine learning algorithm is used to predict the waste gas generation volume and composition by considering factors such as coal characteristics, combustion conditions, and production load;

[0038] During operation, the waste gas generation prediction model uses machine learning algorithms to predict the amount and composition of waste gas. In terms of coal characteristics, when burning coal with a high volatile content, more volatile substances will be released, increasing the content of components such as volatile organic compounds and carbon monoxide in the waste gas; the fixed carbon content determines the degree of coal combustion, thereby affecting the generation ratio of carbon dioxide and carbon monoxide. Among the combustion conditions, an increase in temperature will promote the generation of nitrogen oxides because nitrogen in the air is more likely to react with oxygen at high temperatures; an appropriate air flow rate can ensure complete combustion of coal. Too little air leads to incomplete combustion and the generation of carbon monoxide, while too much air will carry away too much heat, affecting the combustion efficiency and possibly increasing nitrogen oxide emissions. The production load directly reflects the amount of coal used and the combustion intensity. When the load increases, the amount of coal burned increases, and the corresponding amount of waste gas generated also rises. Taking the decision tree algorithm as an example, when constructing a decision tree model, starting from the root node, the data set is divided according to a certain feature of the data (such as the volatile content of coal), so that the divided sub-data sets have better consistency in the target variable of the waste gas generation amount or composition. By calculating indicators such as information gain and Gini index, the optimal division feature is selected, and the feature selection and data set division are continuously recursively performed until certain stopping conditions are met (such as the number of samples in the sub-data set is less than a certain threshold or all samples belong to the same category), and a complete decision tree model is constructed. During prediction, new data starts from the root node and is gradually classified downward according to the rules of the decision tree, and finally the prediction result is obtained at the leaf node. The random forest algorithm constructs multiple decision trees and obtains the final prediction result by integrating the prediction results of multiple decision trees (such as voting method or averaging method), improving the stability and accuracy of the model.

[0039] As Figure 1 shown, according to the resource recovery potential assessment model described in S2, analyze the content and distribution of recyclable resources in the waste gas, and evaluate the feasibility and economic benefits of the recovery plan in combination with market demand, resource prices, and recovery technology costs;

[0040] During operation, when the resource recovery potential assessment model is in operation, it first uses advanced analytical instruments (such as gas chromatography-mass spectrometry) to determine the content of recoverable resources in the waste gas. For sulfur, by detecting the concentration of sulfur dioxide in the waste gas and combining it with the total amount of waste gas, the content of sulfur is calculated using stoichiometry methods. For nitrogen, it is mainly determined based on the concentration of nitrogen oxides. Using the material flow analysis method, a material flow model of the waste gas treatment process is established to trace the flow path and transformation relationship of recoverable resources between different treatment units (such as desulfurization towers, denitration devices, adsorbers), and to clarify which links have a higher content of recoverable resources in the waste gas, providing a basis for the design of subsequent recovery processes. In terms of combining market demand, through market research, industry report analysis, etc., pay attention to the demand trends of these resources in industries such as chemical engineering and energy. For example, if the development of the new energy vehicle industry leads to a significant increase in the demand for lithium carbonate, a raw material for lithium batteries, and the waste gas contains compounds that can be converted into lithium, then the potential for recovering this resource is significantly increased. In terms of technical and economic indicators, consider the procurement cost of recovery equipment, including the price of the equipment itself, transportation and installation costs; energy consumption costs during operation, such as electricity consumption and steam usage; maintenance costs, covering regular equipment inspections and component replacement costs; and the market sales price of the recovered products. By constructing a cost-benefit analysis model, combine the resource recovery volume with the above cost and revenue factors to calculate the net income under different recovery scenarios. For example, for the scenario of recovering nitrogen resources from waste gas to prepare ammonium nitrate, if the investment in recovery equipment is large, the operating cost is high, and the market price of ammonium nitrate is low, and the net income calculated by the model is negative, then the feasibility of this scenario is relatively low; on the contrary, if the net income is positive and relatively high, then the scenario has good economic benefits and feasibility.

[0041] As Figure 1 shown, according to the waste gas treatment effect evaluation model described in S2, the performance and efficiency of the equipment are evaluated based on the operating data of the waste gas treatment equipment and the emission indicators of the treated waste gas, and the equipment failure risk is predicted;

[0042] During operation, the waste gas treatment effect evaluation model evaluates the performance and efficiency of the waste gas treatment equipment based on the operation data of the waste gas treatment equipment and the emission indicators of the treated waste gas. The pressure difference between the inlet and outlet in the operation data reflects the resistance of the waste gas flowing in the equipment. An excessive pressure difference may indicate blockage inside the equipment (such as blockage of the packing in the desulfurization tower), which affects the waste gas treatment effect. The flow rate data reflects the amount of waste gas passing through the equipment per unit time and is an important indicator to measure the treatment capacity of the equipment. Unstable flow rate may lead to insufficient treatment. The dosage of chemicals (such as the dosage of limestone in the desulfurization process and the injection volume of urea solution in the denitrification process) is closely related to the treatment effect. Appropriate chemical dosage can ensure the pollutant removal efficiency. Insufficient dosage cannot react fully, and excessive dosage causes waste and may produce secondary pollution. The emission indicators of the treated waste gas, such as sulfur dioxide, nitrogen oxides, and particulate matter concentration, are key parameters directly measuring the treatment effect. The evaluation model uses statistical analysis methods (such as time series analysis to observe the change trend of emission indicators over time) and machine learning algorithms (such as neural networks). Taking these operation data and emission indicators as inputs, an evaluation model is constructed. The neural network constructs a network structure including an input layer, a hidden layer, and an output layer. The input layer receives these data, and the hidden layer performs complex transformations and feature extractions on the data through non-linear activation functions. Finally, the equipment performance evaluation results (such as high efficiency, medium efficiency, low efficiency) and fault prediction results (such as the probability of equipment failure within the next week) are output at the output layer. Through learning and training on a large amount of historical data, the network parameters are continuously adjusted to enable the model to accurately evaluate the equipment performance and predict the fault risk, providing a basis for the maintenance and optimized operation of the equipment. For example, arranging the equipment maintenance plan in advance to avoid production interruptions and environmental pollution accidents caused by equipment failures.

[0043] As Figure 2 shown, a dynamic optimization system for comprehensive utilization of coal waste gas based on big data, characterized in that: it includes a central processing unit module, a power supply module, a data acquisition module, a data transmission module, a big data processing module, and an execution module;

[0044] The central processing module and the power supply module are electrically connected. The power supply module can supply power to the central processing module, and the central processing module can control the current output in the power supply module;

[0045] The central processor module and the data acquisition module are signal-connected. The data acquisition module consists of sensors, a data acquisition terminal, and a communication component, which collects data and transmits it to the data transmission module;

[0046] The central processing module and the data transmission module are signal-connected. The data transmission module can perform encryption, compression, and error correction processing on the data, has data caching and flow control functions, and transmits the data to the big data processing module;

[0047] The central processing module and the big data processing module are connected by signals. The big data processing module includes a data storage sub-module, a data processing sub-module, a model management sub-module, and an optimization decision sub-module, which are used to store, process, analyze data, build and manage models, and generate optimization strategies.

[0048] The central processing module and the execution module are connected by signals. The execution module consists of coal combustion equipment, waste gas treatment equipment, resource recovery equipment, and a control system, which receives optimization strategies and implements them, and feeds back the equipment operation status and treatment effects.

[0049] During operation, in the dynamic optimization system for comprehensive utilization of coal waste gas based on big data, each module collaborates closely. The data acquisition module, relying on sensors distributed at key positions in coal production and waste gas treatment, converts the collected physical and chemical quantities into signals. After preprocessing by the data acquisition terminal, the signals are packed according to a specific communication protocol and transmitted to the data transmission module through a combination of wired and wireless methods. The data transmission module uses encryption, compression, and error correction technologies to ensure the secure, efficient, and accurate transmission of data to the big data processing module. In the big data processing module, the data storage sub-module stores data using a distributed file system and a database. The data processing sub-module cleans and analyzes data with the help of a distributed computing framework, providing support for the model management sub-module to build and train various models. The central processing module provides computing support for the model building, training, and optimization decision sub-module of the big data processing module to generate dynamic optimization strategies, and converts the strategies into instructions and transmits them to the execution module. The execution module adjusts the equipment operation parameters according to the instructions, and its operation status and treatment effects are fed back to the data acquisition module, forming a closed loop. The power supply module provides stable power for each module of the system and dynamically adjusts the power distribution according to the needs of each module to ensure the stable and efficient operation of the entire system.

[0050] In the data acquisition module, sensors use physical, chemical, and biological principles, etc., to convert various physical and chemical quantities in the coal production and waste gas treatment processes into electrical signals or digital signals. For example, gas sensors detect the concentration of various components in the waste gas through chemical reactions, and temperature sensors measure temperature using the characteristics of thermistors or thermocouples. The data acquisition terminal processes the signals output by the sensors, such as amplification, filtering, and analog-to-digital conversion, and packs the data according to a specific communication protocol. The communication component uses wired and wireless communication technologies to transmit the packed data to the data transmission module, realizing the acquisition and preliminary transmission of data.

[0051] In the data transmission module, an encryption algorithm (such as AES) is used to encrypt the data to prevent the data from being stolen or tampered with during transmission. A compression algorithm (such as GZIP) is used to compress the data to reduce the amount of data transmitted and improve the transmission efficiency. Error correction code technology (such as cyclic redundancy check CRC) is used to correct errors in the data to ensure the accuracy of the data. It has data caching and flow control functions. When the network is congested, the data is first cached locally and then transmitted after the network returns to normal. The data is securely transmitted to the big data processing module through a data transmission protocol (such as MQTT, HTTP);

[0052] In the big data processing module, the data storage sub-module's distributed file system (such as HDFS) splits the data into multiple data blocks and stores them on different nodes to improve the reliability of the data through redundant storage. Databases (such as MySQL and MongoDB) are used to store structured and unstructured data, and an appropriate storage method is selected according to the characteristics and usage requirements of the data;

[0053] The data processing sub-module MapReduce divides the data processing task into two stages: Map and Reduce. In the Map stage, the input data is split into multiple small pieces, each small piece is processed in parallel, and key-value pairs are output. In the Reduce stage, the values with the same key are aggregated and processed to achieve data cleaning, transformation, and preliminary analysis. Spark is based on in-memory computing, loads the data into memory for iterative computing, greatly improving the data processing speed. Key features and patterns in the data are extracted through data mining algorithms (such as clustering analysis, association rule mining),

[0054] The model management sub-module uses machine learning frameworks (such as TensorFlow, PyTorch) to build waste gas generation prediction models, resource recovery potential assessment models, and waste gas treatment effect assessment models. The models are trained with a large amount of historical data, and optimization algorithms (such as stochastic gradient descent) are used to adjust the model parameters to improve the accuracy and generalization ability of the models. Version management of the models is carried out, and the training process and performance metrics of the models are recorded to facilitate model updates and backtracking,

[0055] The optimization decision sub-module generates dynamic optimization strategies according to real-time data and prediction models using optimization algorithms (such as multi-objective genetic algorithm, particle swarm optimization algorithm), converts the optimization strategies into specific control instructions, and sends them to the execution module. At the same time, the implementation effect of the optimization strategies is monitored and evaluated, and the strategies are adjusted and optimized according to the feedback information;

[0056] In the central processing module, it can provide powerful computing support for the big data processing module. During the model construction and training process, taking the training of a neural network model as an example, the central processor executes core operations such as matrix operations and data iteration in the model training algorithm. When training the waste gas generation prediction model, a large amount of historical data and real-time monitoring data are input into the model. The central processor, according to optimization algorithms such as stochastic gradient descent, continuously adjusts the weights and biases of the neural network, gradually reducing the error between the model prediction value and the actual value, and improving the accuracy and generalization ability of the model;

[0057] In the power supply module, it can provide stable and reliable power for the data acquisition module, data transmission module, big data processing module, central processor module, and execution module.

[0058] As Figure 2 shown, the sensors of the data acquisition module are distributed at key positions in coal production and waste gas treatment, and the data acquisition terminal preprocesses the sensor signals;

[0059] During operation, in the coal production link, such as at the underground coal mining equipment, dust sensors are installed to monitor the dust concentration generated during coal mining. Since dust not only affects the working environment but may also affect the waste gas composition during subsequent combustion. At the coal transportation belt, coal quality sensors are set to continuously monitor quality parameters such as the ash content and moisture content of coal. These parameters have an important impact on the combustion process and waste gas generation. At key positions of the waste gas treatment equipment, such as the inlet and outlet of the desulfurization tower, sulfur dioxide sensors are installed to accurately monitor the concentration change of sulfur dioxide in the waste gas before and after treatment to evaluate the desulfurization effect. The data acquisition terminal is equipped with a high-precision signal amplification circuit, which can amplify the weak signals output by the sensors to an appropriate level range for subsequent processing. Digital filtering algorithms, such as Kalman filtering, are used to remove noise interference in the signals and improve the accuracy of the data. The analog-to-digital conversion uses a high-resolution analog-to-digital converter to accurately convert the analog signals into digital signals, reducing the conversion error. The preprocessed signals are packed according to a specific communication protocol to ensure the integrity and correctness of the data during transmission, providing a reliable data source for subsequent data processing and analysis.

[0060] As Figure 2 shown, the data transmission module uses the AES encryption algorithm, GZIP compression algorithm, and CRC error correction code technology to transmit data through the MQTT or HTTP protocol;

[0061] During operation, the data transmission module uses the AES encryption algorithm to encrypt data with a 128-bit or 256-bit key. At the sending end, the data is divided into fixed-size blocks, and through a series of byte substitution, shifting, and mixing operations, the plaintext is converted into ciphertext. At the receiving end, the same key is used for reverse operations to restore the original data, effectively preventing the data from being stolen or tampered with during transmission. The GZIP compression algorithm is based on the DEFLATE algorithm. By searching for repeated byte sequences in the data and replacing them with shorter encodings, data compression is achieved. For example, for multiple consecutive identical bytes, the algorithm will represent them as a combination of the repetition count and the byte value, reducing the storage space and transmission volume of the data and improving the transmission efficiency. The cyclic redundancy check (CRC) adds redundant check bits to the transmitted data. At the receiving end, the data is verified according to the verification rules. If an error occurs during data transmission, the verification result will be inconsistent with that at the sending end, and the receiving end can detect the error and request retransmission to ensure the accuracy of the data. The data transmission protocol (such as MQTT) adopts the publish-subscribe mode. The data acquisition module acts as the publisher and publishes data to specific topics. The data processing module acts as the subscriber, subscribes to relevant topics, and receives the data to achieve reliable data transmission.

[0062] As Figure 2 shown, the data storage sub-module of the big data processing module uses distributed file system and database technologies, the data processing sub-module uses MapReduce and Spark frameworks, the model management sub-module uses machine learning frameworks to build and manage models, and the optimization decision sub-module uses optimization algorithms to generate optimization strategies;

[0063] During operation, in the data storage sub-module of the big data processing module, the distributed file system (such as HDFS) plays a key role. Its name node is responsible for managing the namespace of the entire file system and maintaining the mapping relationship between files and data blocks. When the client initiates a data storage request, the name node determines the specific data node where the data block should be stored based on the system's load conditions and the data block distribution strategy. After the data is divided into multiple data blocks, it is scattered and stored on different data nodes. Through the multi-copy redundancy storage mechanism, usually each data block will have multiple copies distributed on different nodes, thereby ensuring that even if some data nodes fail, the data can still be reliably obtained, greatly improving the reliability of data storage. For example, when a data node cannot be accessed due to hardware failure, the system can automatically read the data from other nodes that have copies of the data block to ensure the continuity and availability of the data.

[0064] Relational databases (such as MySQL) are optimized for storing structured data. By defining the table structure, the organization form of data is clarified. Each row in the table represents a record, and each column represents a specific data field. Each field has a clear data type. With the powerful query, insert, update, and delete operation functions of the SQL language, it can efficiently handle complex data association relationships and transaction logics. For example, when storing the operation parameters of coal waste gas treatment equipment, a table containing fields such as equipment number, operation time, temperature, and pressure can be created. Using SQL query statements, it is convenient to retrieve the operation data of a specific equipment within a certain period of time, or update relevant records according to the equipment operation status, providing structured data support for equipment performance analysis and fault diagnosis.

[0065] Non-relational databases (such as MongoDB) store unstructured or semi-structured data in the form of documents. Each document is similar to a flexible JSON object, and its structure can be freely defined according to the data characteristics without following a fixed table structure pattern. This high flexibility makes it very suitable for storing irregular data such as coal waste gas composition analysis reports and equipment operation logs. For example, the waste gas composition analysis report may contain concentration data of various pollutants at different detection times, description of detection methods, information of detection personnel, etc. Using MongoDB, the entire report can be directly stored in the form of a document, facilitating subsequent query and analysis without complex data format conversion and table structure adaptation.

[0066] The data processing sub-module uses the MapReduce framework to process large-scale data tasks. In the Map stage, the input data is split into multiple data blocks and distributed to different computing nodes for parallel processing. Each computing node performs the same operation on the data block it processes. For example, when cleaning waste gas data, each data block can be checked in parallel to see if the data conforms to the preset format, mark the data that does not meet the requirements as abnormal, and output it in the form of key-value pairs. In the Reduce stage, the values with the same key are collected together for summary processing, such as summarizing and outputting all the data marked as abnormal to complete the data cleaning task. Through this staged parallel processing method, the efficiency of large-scale data processing is greatly improved.

[0067] Due to its in-memory computing characteristics, Spark demonstrates outstanding advantages in data processing. When dealing with iterative computing tasks, such as training a machine learning model through multiple iterations, Spark loads the data into memory, and the data quickly circulates and is processed in memory, avoiding the time overhead caused by frequent disk reads and writes, and significantly improving the computing speed. For example, when performing principal component analysis using Spark's machine learning library, it can efficiently extract key features from a large amount of waste gas data, reduce the data dimension, and at the same time retain the main information of the data, providing high-quality feature data for subsequent model training and data analysis.

[0068] The model management sub-module constructs waste gas-related models relying on machine learning frameworks (such as TensorFlow). When building a waste gas generation prediction model, by defining the model structure, determining the number of layers of the neural network, the number of neurons in each layer, and the connection method between neurons, selecting a loss function suitable for regression problems (such as mean squared error) to measure the difference between the model prediction value and the actual value, and at the same time selecting an optimizer (such as Adam optimizer) to adjust the model parameters. During the training process, a large amount of historical data is used to continuously optimize the model. Through algorithms such as stochastic gradient descent, according to the model prediction error backpropagation, the weights and biases of the neural network are adjusted to gradually reduce the error, thereby improving the accuracy and generalization ability of the model, enabling it to accurately predict the waste gas generation situation under different working conditions.

[0069] For the resource recovery potential assessment model and the waste gas treatment effect assessment model, they are constructed and trained according to a similar process. The version management mechanism details information such as parameter changes, training data used, and various evaluation indicators during the model training process. When the model performance deteriorates due to changes in production working conditions or data distribution, it is possible to trace back to a previous model version with good performance based on the version management record, or retrain the model using new data to ensure that the model can always accurately reflect the actual situation and provide a reliable basis for the decision-making of coal waste gas comprehensive utilization.

[0070] The optimization decision sub-module makes optimization decisions based on real-time data and prediction models using the particle swarm optimization algorithm. This algorithm transforms objectives such as waste gas pollutant reduction, maximizing resource recovery, and minimizing costs into fitness functions. In the solution space, each particle represents a possible optimization strategy. The particles adjust their flight directions and speeds by continuously tracking their own historical optimal positions and the global optimal position of the group. In each iteration, the particles evaluate the quality of their own positions according to the fitness function and move towards a better solution. After multiple iterative searches, the solution that makes the fitness function optimal is finally found, that is, the dynamic optimization strategy. For example, when adjusting the operating parameters of combustion equipment, the particles may represent different combinations of parameters such as combustion temperature and air flow rate. By continuously optimizing these combinations, the best parameter settings that can both reduce waste gas pollutant emissions, improve resource recovery efficiency, and reduce costs are found. After converting the optimization strategy into specific control instructions such as digital signals recognizable by the PLC control system, they are sent to the execution module for execution. At the same time, closely monitor the data feedback from the execution module. If it is found that the waste gas emissions still do not meet the standards after adjusting the combustion temperature, re-adjust the temperature parameters in the optimization strategy according to the feedback information, and continuously optimize the decision-making process to improve the effect of coal waste gas comprehensive utilization.

[0071] As Figure 2 shown, the control system of the execution module adjusts the equipment operating parameters according to the received control instructions to achieve optimized control of the coal waste gas comprehensive utilization process;

[0072] During operation, the control system in the execution module (such as the DCS control system) serves as the core control unit, receiving control instructions sent from the big data processing module. The DCS control system integrates complex control algorithms internally, among which the PID control algorithm is widely used. When receiving an instruction to adjust the combustion temperature of the coal combustion equipment, the PID controller calculates the opening degree of the fuel supply valve of the burner that needs to be adjusted based on the deviation between the current actual temperature and the target temperature, as well as the change rate and integral value of the deviation. If the actual temperature is lower than the target temperature, the PID controller will increase the valve opening degree to increase the fuel supply volume, thereby increasing the combustion temperature; conversely, if the actual temperature is higher than the target temperature, it will decrease the valve opening degree to reduce the fuel supply volume. Through continuous real-time adjustment, the combustion temperature is stabilized near the target value, ensuring a more efficient coal combustion process and reducing the generation of waste gas pollutants.

[0073] For the waste gas treatment equipment, taking the control of the spray pump speed as an example, after the control system receives an instruction to optimize the spray volume to improve the desulfurization efficiency, it first obtains relevant data such as the current spray volume and the concentration of sulfur dioxide in the waste gas through sensors. According to these real-time data and the preset desulfurization efficiency target, it uses a control algorithm to calculate the appropriate spray pump speed. If the current sulfur dioxide concentration is high and the desulfurization efficiency fails to meet the expectation, the control system will increase the spray pump speed to increase the spray volume, enabling more desulfurizing agent to contact and react with the sulfur dioxide in the waste gas, improving the desulfurization effect; if the concentration is low and the desulfurization efficiency is good, it will appropriately reduce the spray pump speed to avoid resource waste and excessive operation of the equipment.

[0074] In terms of the resource recovery equipment, when the control system receives an instruction to adjust the resource recovery speed, it adjusts the recovery speed by controlling the operating frequency of the equipment according to the working principle and current operating state of the resource recovery equipment. For example, for the equipment that uses the adsorption method to recover the available resources in the waste gas, increasing the equipment operating frequency can increase the contact time and times between the adsorbent and the waste gas, thereby increasing the resource recovery volume; conversely, reducing the operating frequency can reduce energy consumption and equipment wear during the period when the resource content is low or during equipment maintenance.

[0075] Meanwhile, various sensors distributed on the device, such as temperature sensors, monitor the temperature of key parts of the device in real time to prevent the device from being damaged due to overheating; pressure sensors monitor the internal pressure of the device to ensure that the device operates within a safe pressure range; concentration sensors continuously detect the concentration of pollutants or recyclable resources in the waste gas and provide real-time feedback data to the control system. These sensors feed the collected data back to the big data processing module through the data acquisition module. The big data processing module evaluates the operation effect of the execution module and the implementation of the optimization strategy based on the feedback data. If it is found that there is a deviation between the actual operation effect and the expected goal, such as the waste gas emission index still not meeting the standard or the resource recovery efficiency not reaching the expectation, the big data processing module will regenerate the optimization strategy and send it to the execution module. The execution module adjusts the device operation parameters again to form a closed-loop control, continuously optimizing the comprehensive utilization process of coal waste gas, keeping it in the best operation state all the time, and achieving efficient waste gas treatment and resource recovery.

[0076] The present invention realizes a comprehensive and accurate understanding and prediction of the comprehensive utilization process of coal waste gas by integrating multi-source data such as coal production, waste gas treatment, environmental monitoring, and market, and deeply mining by using advanced big data analysis technology and machine learning algorithms. By constructing a multi-factor waste gas generation prediction model, a resource recovery potential evaluation model, and a waste gas treatment effect evaluation model, and based on real-time data and prediction models, a multi-objective optimization algorithm is used to generate a dynamic optimization strategy, which can adjust the waste gas treatment and resource recovery plan in real time according to production conditions, environmental changes, and market demands. This dynamic optimization decision-making mechanism breaks the static mode of traditional treatment methods, realizes the adaptive optimization of the system, and improves the flexibility and adaptability of the system.

[0077] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and all these changes and improvements fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A dynamic optimization method for comprehensive utilization of coal waste gas based on big data, characterized in that, The specific steps of this dynamic optimization method are as follows: S1: Multi-source data collection and preprocessing: Collect multi-source data including coal quality parameters, combustion process parameters, waste gas composition and emission parameters, equipment operation status parameters, and environmental parameters, and perform data cleaning, normalization, standardization, and feature extraction on the data; S2: Big data analysis and model construction: Construct a waste gas generation prediction model, a resource recovery potential assessment model, and a waste gas treatment effect assessment model; S3: Dynamic optimization decision-making and implementation: Real-time monitor key indicators and give early warnings, generate optimization strategies and implement them, and adjust the optimization strategies according to the feedback.

2. A dynamic optimization method for comprehensive utilization of coal waste gas based on big data according to claim 1, characterized in that: According to the waste gas generation prediction model described in S2, use machine learning algorithms to predict the waste gas generation volume and composition considering coal characteristics, combustion conditions, and production load factors.

3. A dynamic optimization method for comprehensive utilization of coal waste gas based on big data according to claim 1, characterized in that: According to the resource recovery potential assessment model described in S2, analyze the content and distribution of recoverable resources in the waste gas, and evaluate the feasibility and economic benefits of the recovery plan in combination with market demand, resource prices, and recovery technology costs.

4. A dynamic optimization method for comprehensive utilization of coal waste gas based on big data according to claim 1, characterized in that: According to the waste gas treatment effect assessment model described in S2, evaluate the equipment performance and efficiency based on the operation data of the waste gas treatment equipment and the emission indicators of the treated waste gas, and predict the equipment failure risk.

5. A dynamic optimization system for comprehensive utilization of coal waste gas based on big data, characterized in that: It includes a central processing unit module, a power supply module, a data collection module, a data transmission module, a big data processing module, and an execution module; The central processing module and the power supply module are electrically connected, the power supply module can supply power to the central processing module, and the central processing module can control the current output in the power supply module; The central processing unit module and the data collection module are signal-connected, the data collection module consists of sensors, a data collection terminal, and a communication component, collects data and transmits it to the data transmission module; The central processing module and the data transmission module are signal-connected, the data transmission module can perform data encryption, compression, and error correction processing, has data caching and flow control functions, and transmits the data to the big data processing module; The central processing module and the big data processing module are signal-connected, the big data processing module includes a data storage sub-module, a data processing sub-module, a model management sub-module, and an optimization decision-making sub-module, which are used to store, process, analyze data, construct and manage models, and generate optimization strategies; The central processing module and the execution module are signal-connected, the execution module consists of a coal combustion device, a waste gas treatment device, a resource recovery device, and a control system, receives the optimization strategy and implements it, and feedbacks the equipment operation status and treatment effect.

6. The dynamic optimization system for comprehensive utilization of coal waste gas based on big data according to claim 5, characterized in that: The sensors of the data collection module are distributed at key positions in coal production and waste gas treatment, and the data collection terminal preprocesses the sensor signals.

7. A dynamic optimization system for comprehensive utilization of coal waste gas based on big data according to claim 5, characterized in that: The data transmission module uses the AES encryption algorithm, the GZIP compression algorithm, and the CRC error correction code technology to transmit data through the MQTT or HTTP protocol.

8. The dynamic optimization system for comprehensive utilization of coal waste gas based on big data according to claim 5, characterized in that: The data storage sub-module of the big data processing module adopts distributed file system and database technologies. The data processing sub-module uses MapReduce and Spark frameworks. The model management sub-module uses machine learning frameworks to build and manage models. The optimization decision sub-module uses optimization algorithms to generate optimization strategies.

9. The dynamic optimization system for comprehensive utilization of coal waste gas based on big data according to claim 5, wherein: The control system of the execution module adjusts the device operation parameters according to the received control instructions to achieve the optimized control of the comprehensive utilization process of coal waste gas.

Images