Desulfurization device operation cost on-line calculation and energy efficiency optimization integration method and system

Abstract

The invention provides a desulfurization device operation cost online calculation and energy efficiency optimization integration method and system, and particularly relates to the technical field of industrial desulfurization, and the system comprises a data acquisition and preprocessing module, a dynamic cost calculation module, a self-adaptive optimization module, a closed-loop control module and a user interaction and visualization module. According to the invention, through integration and cooperation of the data acquisition and preprocessing module, the dynamic cost calculation module, the adaptive optimization module, the closed-loop control module and the user interaction and visualization module, real-time online calculation of the operation cost of the desulfurization device and global optimization of the energy efficiency are realized; an LSTM deep learning algorithm, a closed-loop control architecture and a multi-source data fusion technology are introduced, on the premise that emission reaches the standard, the operation cost is reduced, the intelligent level of the system is remarkably improved, and meanwhile, through integration of a predictive maintenance module, a block chain security module and a user behavior analysis module, the safety of the system is improved. The maintenance cost is further reduced; the data security is ensured; and the user experience is improved.

Description

Technical Field

[0001] This invention relates to the field of industrial desulfurization, and in particular to an integrated method and system for online calculation of desulfurization equipment operating costs and energy efficiency optimization. Background Technology

[0002] Coal-fired boiler desulfurization systems, as a key technology for achieving ultra-low emissions in the power energy industry, are widely used in the flue gas purification process of thermal power plants. With increasingly stringent environmental policies and continuous improvement in industrial automation, traditional desulfurization systems are gradually evolving towards intelligent and data-driven approaches. Among related technologies, a complete technical system has been constructed through the collaborative operation of edge sensing devices, data transmission networks, and big data analysis platforms, encompassing data acquisition, transmission, storage, and intelligent decision-making. Specifically, this system covers the entire process from flue gas composition monitoring and equipment status sensing to multi-objective optimization control, including key aspects such as high-frequency laser spectral analysis, current ripple detection, multispectral particulate matter inversion, protocol adaptive transmission, knowledge graph modeling, and multi-objective genetic algorithm optimization, aiming to improve the operating efficiency and stability of the desulfurization system.

[0003] However, existing desulfurization systems, which directly combine DCS control with manual experience-based adjustments, fail to achieve efficient fusion and dynamic optimization decision-making from multi-source heterogeneous data. This can lead to large fluctuations in desulfurization efficiency, high energy consumption and material costs, or accelerated equipment wear and frequent unplanned shutdowns, thus impacting the economic operation and environmental compliance of power plants. Furthermore, traditional systems lack early warning capabilities for equipment status and multi-objective collaborative optimization mechanisms when facing complex operating conditions such as changes in fuel characteristics and load fluctuations. This makes it difficult to balance desulfurization efficiency, operating costs, and equipment reliability, limiting the further development of desulfurization systems towards intelligence and green technology. Summary of the Invention

[0004] The present invention aims to at least partially solve one of the technical problems in the related art.

[0005] Therefore, the first objective of this invention is to propose an integrated method for online calculation of desulfurization unit operating costs and energy efficiency optimization.

[0006] The second objective of this invention is to propose an integrated device for online calculation of desulfurization unit operating costs and energy efficiency optimization.

[0007] To achieve the above objectives, a first aspect of the present invention proposes an integrated method for online calculation of desulfurization unit operating costs and energy efficiency optimization, comprising: S1 collects real-time operating parameters, environmental parameters, and market data of the desulfurization unit through IoT sensors, and performs outlier detection, noise filtering, and missing value filling on the collected data to form cleaned data. S2 aligns and integrates the cleaning operation parameters, environmental parameters and market data in time and space, and constructs structured data input to a multivariate dynamic cost model to calculate the operating cost of the desulfurization unit in real time. S3, based on the LSTM deep learning algorithm, trains historical operating data and optimization results to construct a nonlinear relationship model between desulfurization efficiency and cost, and adaptively optimizes the operating parameters of the desulfurization unit according to the real-time cost calculation results. S4 feeds back the optimized operating parameters to the desulfurization unit's control system via industrial bus or wireless communication protocol, forming a closed-loop control architecture to continuously optimize the operating status.

[0008] In one embodiment of the present invention, S1 includes: S11, anomaly detection is performed on continuously collected operating parameters using a sliding window algorithm, and data points that exceed the set threshold range are identified and marked; S12 uses a time-series-based interpolation algorithm to fill in missing values ​​and wavelet transform to denoise the noisy data.

[0009] In one embodiment of the present invention, S2 includes: S21, Time synchronization processing is performed on the operating parameters and market data of different timestamps, and data alignment is achieved by interpolation or delay compensation methods; S22 uses an association rule mining algorithm to identify the coupling relationship between operating parameters and environmental parameters, and constructs a multi-dimensional feature vector for input to the cost model.

[0010] In one embodiment of the present invention, S3 includes: S31 adopts a multi-layer LSTM network structure. The input layer includes the amount of desulfurizing agent, equipment power, flue gas flow rate, SO2 concentration and corresponding historical optimization results. S32 uses the gradient descent algorithm to fine-tune the LSTM model online to adapt to the changing trends of desulfurization efficiency and cost under current operating conditions.

[0011] In one embodiment of the present invention, it further includes: S5. Construct a user profile based on the user's historical operation data, and generate personalized optimization suggestions based on the user profile to improve user operation efficiency and system user experience.

[0012] To achieve the above objectives, a second aspect of the present invention provides an integrated device for online calculation of desulfurization unit operating costs and energy efficiency optimization, comprising: The data acquisition and cleaning module is used to collect the operating parameters, environmental parameters and market data of the desulfurization unit in real time through IoT sensors, and to perform outlier detection, noise filtering and missing value filling on the collected data to form cleaned data. The data alignment and fusion module is used to perform spatiotemporal alignment and correlation fusion of the cleaned operating parameters, environmental parameters and market data, and to build structured data input into a multivariate dynamic cost model to calculate the operating cost of the desulfurization unit in real time. The model training and optimization module is used to train historical operating data and optimization results based on the LSTM deep learning algorithm, build a nonlinear relationship model between desulfurization efficiency and cost, and adaptively optimize the operating parameters of the desulfurization unit based on real-time cost calculation results. The parameter feedback and control module is used to feed back the optimized operating parameters to the control system of the desulfurization unit through industrial bus or wireless communication protocol, forming a closed-loop control architecture to continuously optimize the operating status.

[0013] To achieve the above objectives, a third aspect of the present invention provides an integrated system for online calculation of desulfurization unit operating costs and energy efficiency optimization, comprising: The module comprises a data acquisition and preprocessing module, a dynamic cost calculation module, an adaptive optimization module, a closed-loop control module, and a user interaction and visualization module, among which: The data acquisition and preprocessing module is used to collect the operating parameters, environmental parameters and market data of the desulfurization unit in real time, and to clean, fuse and structure the data. The dynamic cost calculation module is connected to the output of the data acquisition and preprocessing module. It is used to build a multivariate dynamic cost model based on structured data, including desulfurizing agent costs, equipment power consumption, maintenance costs and labor costs, and to calculate the operating cost of the desulfurization unit in real time. The adaptive optimization module is connected to the output of the dynamic cost calculation module. It is used to train the historical operating data and optimization results based on the LSTM deep learning algorithm, build a nonlinear relationship model between desulfurization efficiency and cost, and adaptively optimize the operating parameters of the desulfurization unit according to the real-time cost calculation results. The closed-loop control module is connected to the output of the adaptive optimization module and is used to feed back the optimized operating parameters to the desulfurization unit to form a closed-loop control architecture. The user interaction and visualization module is bidirectionally connected to the data acquisition and preprocessing module, dynamic cost calculation module, adaptive optimization module, and closed-loop control module. It is used to display the optimization effect, operating status, and cost composition to the user and to receive the user's operation instructions.

[0014] Furthermore, the data acquisition and preprocessing module includes a multi-source data acquisition unit, a market data acquisition unit, a data cleaning unit, and a data fusion unit, wherein: The multi-source data acquisition unit is used to collect operating parameters such as desulfurizer dosage, equipment power, flue gas flow rate, SO2 concentration, and environmental parameters such as temperature, humidity, and atmospheric pressure through IoT sensors. The market data acquisition unit is used to collect market data on desulfurizing agent prices and electricity prices via API interface; The data cleaning unit is used to perform outlier detection, noise filtering, and missing value imputation on the collected data. The data fusion unit is used to perform spatiotemporal alignment and correlation fusion of cleaned operating parameters, environmental parameters, and market data to form structured data.

[0015] Furthermore, the dynamic cost calculation module includes a desulfurizer cost model unit, a maintenance cost model unit, a labor cost model unit, and a total cost calculation unit, wherein: The desulfurizing agent cost model unit is used to calculate the cost of desulfurizing agent based on the amount of desulfurizing agent used and the market price; the equipment power consumption model unit is used to calculate the equipment power consumption cost based on the equipment power and the electricity price. The maintenance cost model unit is used to calculate maintenance costs based on equipment uptime, failure frequency, and maintenance unit price. The labor cost model unit is used to calculate labor costs based on labor hours and wage levels. The total cost calculation unit is used to sum the calculation results of the above model units in a weighted manner to obtain the total operating cost of the desulfurization unit.

[0016] Furthermore, the adaptive optimization module includes an LSTM algorithm training unit and a real-time optimization unit, wherein: The LSTM algorithm training unit is used to train the LSTM network using historical operating data and corresponding optimization results, and to build a nonlinear relationship model between desulfurization efficiency and cost. The real-time optimization unit is used to adaptively optimize the operating parameters of the desulfurization unit based on the real-time cost calculation results and the nonlinear relationship model, using a gradient descent algorithm.

[0017] Furthermore, the closed-loop control module includes a feedback unit and a control unit, wherein: The feedback unit is used to feed back the optimized operating parameters to the control system of the desulfurization unit via industrial bus or wireless communication protocol; The control unit is used to control the desulfurization unit in real time based on the optimized operating parameters, forming a closed-loop control architecture of calculation-optimization-adjustment-recalculation. The adjusted operating parameters will be used as new input data to participate in the next round of cost calculation and optimization.

[0018] Furthermore, it also includes a predictive maintenance module, which is connected to the output of the data acquisition and preprocessing module. This module is used to predict potential failures of the desulfurization unit through vibration analysis and temperature monitoring, and to provide maintenance suggestions in conjunction with the equipment life prediction model.

[0019] Furthermore, it also includes a blockchain security module, which is connected to the output of the data acquisition and preprocessing module. This module is used to encrypt and store key information of the operation data and optimization results using blockchain distributed ledger technology, and to record them in an immutable manner. It also supports remote auditing and data verification by environmental regulatory authorities.

[0020] Furthermore, it also includes a user behavior analysis module, which is bidirectionally connected to the user interaction and visualization module. This module is used to analyze users' operating habits and maintenance preferences through historical user operation data and to provide personalized optimization suggestions in conjunction with user profiling technology.

[0021] The methods, apparatus, and systems of this invention can achieve real-time online calculation and adaptive optimization of desulfurization device operating costs while ensuring SO2 emissions meet standards, significantly reducing operating and maintenance costs and improving system intelligence and optimization response speed.

[0022] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0023] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart illustrating an integrated method for online calculation of desulfurization unit operating costs and energy efficiency optimization, provided in this application embodiment; Figure 2 A structural diagram of an integrated device for online calculation of desulfurization unit operating costs and energy efficiency optimization provided in this application embodiment; Figure 3 This is a structural diagram of an integrated system for online calculation of desulfurization device operating costs and energy efficiency optimization, provided in an embodiment of this application.

[0024] Labels: 1. Data Acquisition and Preprocessing Module; 2. Dynamic Cost Calculation Module; 3. Adaptive Optimization Module; 4. Closed-Loop Control Module; 5. User Interaction and Visualization Module; 6. Predictive Maintenance Module; 7. Blockchain Security Module; 8. User Behavior Analysis Module. Detailed Implementation

[0025] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0026] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0027] The following description, with reference to the accompanying drawings, describes an integrated method and apparatus for online calculation of desulfurization unit operating costs and energy efficiency optimization according to an embodiment of the present invention.

[0028] Example 1 Figure 1 This is a flowchart of an integrated method for online calculation of desulfurization unit operating costs and energy efficiency optimization according to an embodiment of the present invention, such as... Figure 1 As shown, it includes: S1 collects real-time operating parameters, environmental parameters, and market data of the desulfurization unit through IoT sensors, and performs outlier detection, noise filtering, and missing value filling on the collected data to form cleaned data.

[0029] Specifically, the step in this invention of "collecting real-time operating parameters, environmental parameters, and market data of the desulfurization unit through IoT sensors, and performing outlier detection, noise filtering, and missing value imputation on the collected data to form cleaned data" is a fundamental step for the entire system to achieve accurate cost calculation and energy efficiency optimization. At the technical implementation level, this step adopts a multi-source heterogeneous data acquisition architecture. Through IoT sensors (such as electrochemical sensors, infrared gas analyzers, and power monitoring modules) deployed at key nodes of the desulfurization unit, real-time operating parameters such as desulfurizing agent dosage, equipment power, flue gas flow rate, and SO2 concentration, as well as environmental parameters such as temperature, humidity, and atmospheric pressure, are acquired. Market data is obtained by connecting to external price databases (such as commodity trading platforms and real-time electricity market quotation systems) via API interfaces to obtain real-time price information for desulfurizing agents and electricity. The acquisition frequency can be set from 1 second to 5 minutes according to system response requirements to meet the real-time requirements of dynamic optimization.

[0030] During data cleaning, the system employs an outlier detection algorithm based on the Z-score method, setting a threshold of ±3σ to identify and remove data points exceeding the normal fluctuation range. Noise filtering utilizes sliding window mean filtering or wavelet transform methods to eliminate high-frequency interference signals during sensor sampling, ensuring data smoothness and continuity. For missing values, the system uses time series interpolation methods (such as linear interpolation and cubic spline interpolation) or LSTM-based prediction imputation methods to make reasonable estimates based on historical data trends, filling data gaps in missing periods and thus ensuring data integrity.

[0031] The cleaned data generated in this step is characterized by high accuracy, low noise, and no missing data, meeting the requirements of the ISO 8000 standard for industrial data preprocessing. This provides reliable input for subsequent dynamic cost modeling and adaptive optimization. In practical applications, this module is widely deployed in desulfurization systems of coal-fired power plants and chemical plants, supporting 24 / 7 uninterrupted operation. Its technical value lies in improving data quality, reducing model misjudgment rates, laying the data foundation for the system to achieve a closed-loop control architecture of "calculation-optimization-adjustment-recalculation," and significantly improving the intelligence and economic efficiency of desulfurization unit operation.

[0032] Furthermore, S1 includes: S11 uses a sliding window algorithm to detect outliers in continuously collected operating parameters, identifies data points that exceed a set threshold range, and marks them.

[0033] Specifically, in the data acquisition and preprocessing module, the sliding window algorithm is used to detect outliers in continuously acquired operating parameters. This is a key technical step to ensure the quality of system input data and improve the accuracy of subsequent cost calculations and optimizations. Based on the principle of time series data processing, this step uses a sliding window mechanism to dynamically analyze the real-time operating parameters of the desulfurization unit (such as desulfurizing agent dosage, equipment power, flue gas flow rate, SO2 concentration, etc.), and identify and mark data points that exceed the set threshold range.

[0034] In some implementations, the sliding window algorithm slides across the time series data point by point within a fixed-length time window (e.g., 30 seconds to 5 minutes), calculating statistical characteristics within the window, such as mean, standard deviation, median, and interquartile range (IQR). Outlier detection typically employs the 3σ principle or the IQR method. The 3σ principle is suitable for approximately normally distributed data, with a threshold set at the window mean ± 3 times the standard deviation. The IQR method is suitable for non-normally distributed data, with thresholds set from Q1 - 1.5IQR to Q3 + 1.5IQR. Furthermore, the system can set differentiated thresholds based on the characteristics of different parameters; for example, the threshold for SO2 concentration can be dynamically adjusted according to national emission standards (e.g., GB 13223-2011).

[0035] In practical applications, this step operates on edge computing nodes or central data processing servers, supporting real-time processing of high-frequency data streams (such as 100 data points per second) to ensure the timeliness of data cleaning. Through outlier labeling, the system can effectively remove abnormal data caused by sensor drift, communication interference, or equipment malfunctions, thereby improving the accuracy and stability of the dynamic cost model and providing high-quality data input for subsequent LSTM algorithm training and adaptive optimization, demonstrating significant engineering practical value.

[0036] S12 uses a time-series-based interpolation algorithm to fill in missing values ​​and wavelet transform to denoise the noisy data.

[0037] Specifically, in the data preprocessing stage, using time-series-based interpolation algorithms to impute missing values ​​and employing wavelet transform to reduce noise in the data are crucial steps to ensure the quality of input data for subsequent dynamic cost calculation and adaptive optimization modules. In some implementations, missing value imputation uses either linear interpolation or cubic spline interpolation, the choice depending on the continuity and trend of the data. For example, for parameters with strong time continuity, such as equipment power and flue gas flow rate, cubic spline interpolation can be used, typically with an interpolation order of 3 and an interpolation step size consistent with the sampling frequency to ensure data smoothness and physical plausibility. Interpolation algorithms in time-series processing must meet the requirements of ISO 14224 regarding data integrity and continuity, ensuring that the input data has no breaks in the time dimension.

[0038] Furthermore, noise filtering employs Discrete Wavelet Transform (DWT) for multi-scale decomposition and reconstruction. In practice, the system can use commonly used wavelet basis functions such as db4 or sym8 to perform a 5-level wavelet decomposition on the acquired signal, extracting approximation coefficients and detail coefficients for each frequency band. The high-frequency detail coefficients correspond to noise components, and a soft-thresholding method is used for noise reduction. The threshold calculation can be based on the Universal Threshold criterion or the SureShrink algorithm to suppress random noise while preserving the effective signal. The signal-to-noise ratio (SNR) after noise reduction can typically be improved to over 20 dB, meeting the data accuracy requirements in industrial control.

[0039] This step plays a crucial role in the integrated system for online calculation of desulfurization unit operating costs and energy efficiency optimization. Through high-quality data preprocessing, the system can accurately construct a dynamic cost model, improve the training accuracy and optimization effect of the LSTM algorithm, and thus achieve the minimization of operating costs and global optimization of energy efficiency while ensuring that SO2 emissions meet standards.

[0040] S2 aligns and integrates the cleaning operation parameters, environmental parameters, and market data in time and space to construct a structured data input to a multivariate dynamic cost model, and calculates the operating cost of the desulfurization unit in real time.

[0041] Specifically, aligning and fusing the cleaned operating parameters, environmental parameters, and market data in time and space is a key step in constructing a structured data input to a multivariate dynamic cost model. This process, based on the synchronous processing and semantic association of multi-source heterogeneous data, aims to eliminate inconsistencies in data timestamps and spatial dimensions, thereby improving the accuracy and timeliness of cost calculation.

[0042] This step first unifies data from different data sources (such as IoT sensors and API interfaces) to the same time base using timestamp alignment algorithms (such as sliding alignment based on time windows or synchronization methods based on interpolation). For example, device power data may be collected at a rate of seconds, while market data (such as desulfurizer prices) may be updated at a rate of minutes or hours. The system uses linear interpolation or nearest neighbor interpolation methods to map low-frequency data to high-frequency time series, ensuring data consistency in the time dimension. Simultaneously, spatial alignment is achieved through device topology mapping and sensor location encoding, ensuring accurate correspondence between parameters in physical space.

[0043] The system-aligned operating parameters include desulfurizing agent dosage (kg / h), equipment power (kW), and flue gas flow rate (Nm³). 3 / h), SO2 concentration (unit: mg / Nm³) 3 Environmental parameters include temperature (°C), humidity (%RH), and atmospheric pressure (kPa); market data includes desulfurizer prices (RMB / kg) and electricity prices (RMB / kWh). Alignment errors must be controlled within ±0.5 seconds, and the data missing rate should be less than 1% to meet the data integrity requirements of the ISO 50001 energy management system.

[0044] This step is widely used in the operating environments of desulfurization units in coal-fired power plants and chemical plants, and is especially suitable for complex scenarios with multiple time periods and operating conditions. Through the construction of structured data, the system can input data into a multivariate dynamic cost model in real time to achieve comprehensive calculation of desulfurizing agent costs, electricity consumption costs, maintenance costs, and labor costs, providing high-precision and timely data support for subsequent adaptive optimization.

[0045] Furthermore, S2 includes: S21 performs time synchronization processing on the operating parameters and market data at different timestamps, and uses interpolation or delay compensation methods to achieve data alignment.

[0046] Specifically, this step involves synchronizing the operating parameters of the desulfurization unit with market data over time to align multi-source heterogeneous data and provide a consistent time benchmark for subsequent dynamic cost calculation and adaptive optimization. In industrial desulfurization systems, operating parameters (such as desulfurizing agent dosage, equipment power, flue gas flow rate, and SO2 concentration) are typically collected in real time by IoT sensors with sampling frequencies ranging from 100ms to 1s; while market data (such as desulfurizing agent prices and electricity prices) are obtained through API interfaces, with update cycles generally on the order of minutes or hours. Due to the significant difference between sampling frequencies and update cycles, directly fusing data will result in inconsistent timestamps, affecting the accuracy of model inputs and the reliability of calculation results.

[0047] In some implementations, time synchronization processing employs linear interpolation or spline interpolation methods to complete the time dimension of low-frequency market data, aligning it with high-frequency operating parameters at the same point in time. For example, if market data is updated every 5 minutes, while operating parameters are collected every second, cubic spline interpolation can be used to interpolate the market data, ensuring a reasonable estimate at each second. Optionally, for market data with significant delays, a delay compensation algorithm can be introduced. This algorithm uses a sliding window and timestamp offset calculation to perform time correction on the data, eliminating errors caused by data transmission or update delays.

[0048] Furthermore, time synchronization processing must meet the accuracy requirements for time series data alignment in the ISO / IEC 20922 standard, typically requiring timestamp alignment errors to be less than 50ms and interpolation errors to be controlled within ±2%. This step plays a crucial bridging role in the entire system, ensuring that the dynamic cost model has a consistent time reference under multivariate inputs, thereby improving the real-time performance of cost prediction and the response speed of optimization strategies, laying a data foundation for achieving global energy efficiency optimization of the desulfurization unit.

[0049] S22 uses an association rule mining algorithm to identify the coupling relationship between operating parameters and environmental parameters, and constructs a multi-dimensional feature vector for input to the cost model.

[0050] Specifically, this step identifies the coupling relationship between the desulfurization unit's operating parameters and environmental parameters through association rule mining algorithms, and constructs a multi-dimensional feature vector for input into the cost model. This is a key step in achieving intelligent system optimization. At the technical implementation level, this step employs classic association rule mining methods such as the Apriori algorithm or the FP-Growth algorithm to analyze multi-source structured data from the data acquisition and preprocessing modules. Specifically, the system first jointly models operating parameters such as desulfurizer dosage, equipment power, flue gas flow rate, and SO2 concentration with environmental parameters such as temperature, humidity, and atmospheric pressure. By setting minimum support and minimum confidence thresholds (e.g., support ≥ 0.1, confidence ≥ 0.7), statistically significant parameter combinations are mined, revealing their synergistic variation patterns under different operating conditions.

[0051] The multidimensional feature vector constructed by the system typically includes, but is not limited to: desulfurizer dosage (unit: kg / h), flue gas flow rate (unit: Nm³). 3 / h), SO2 inlet concentration (unit: mg / Nm³) 3 The parameters include: circulating pump frequency (Hz), ambient temperature (°C), relative humidity (%RH), electricity price (yuan / kWh), and desulfurizer market price (yuan / kg). These parameters are standardized (e.g., Z-score normalization) and integrated into a unified feature vector, which serves as the input variable for the dynamic cost calculation module for real-time cost modeling and prediction.

[0052] This step is widely used in the online operation and management of desulfurization units in coal-fired power plants, chemical plants, and other similar facilities. By identifying the coupling relationship between operating parameters and environmental parameters, the system can dynamically adjust the desulfurizer dosing strategy and equipment operating mode to achieve optimal cost while meeting emission standards. For example, in high humidity and low oxygen environments, the system can automatically identify the pattern of declining desulfurizer utilization and adjust the dosage in the optimization module to avoid overuse.

[0053] This step effectively improves the system's adaptability to complex operating conditions and enhances the prediction accuracy and generalization ability of the cost model. By constructing high-dimensional, dynamic, and interpretable feature vectors, the system can more accurately reflect the nonlinear relationship between operating costs and environmental variables, providing high-quality input for subsequent LSTM model training and adaptive optimization, and significantly improving the overall system's intelligence level and operating efficiency.

[0054] S3 uses the LSTM deep learning algorithm to train on historical operating data and optimization results, constructs a nonlinear relationship model between desulfurization efficiency and cost, and adaptively optimizes the operating parameters of the desulfurization unit based on real-time cost calculation results.

[0055] Specifically, in this invention, the adaptive optimization module trains on historical operating data and optimization results using the LSTM (Long Short-Term Memory) deep learning algorithm to construct a nonlinear relationship model between desulfurization efficiency and operating cost, and adaptively optimizes the operating parameters of the desulfurization unit based on real-time cost calculation results. This step is the core link in realizing intelligent control and energy efficiency optimization of the system.

[0056] The LSTM algorithm training unit first acquires structured historical operational data from the data acquisition and preprocessing module, including desulfurizer dosage, circulating pump frequency, flue gas flow rate, SO2 inlet and outlet concentrations, equipment power, environmental parameters (such as temperature and humidity), and market data (such as desulfurizer price and electricity price). This data is organized in time series to form a multi-dimensional input feature vector. The LSTM network captures long-term dependencies in the time series through its gating mechanism (input gate, forget gate, output gate), thereby effectively modeling the dynamic nonlinear mapping between desulfurization efficiency and cost. The network structure typically includes 3-5 layers of LSTM units, each layer containing 64-256 neurons, with an input dimension of 12-20 and an output dimension of 1 (such as cost prediction or efficiency prediction). The Adam optimizer is used for model training, with a learning rate set to 0.001-0.0001 and training epochs controlled between 100-300 to avoid overfitting.

[0057] During model training, key performance indicators (KPIs) need to be set, such as mean squared error (MSE), mean absolute error (MAE), and R-squared. 2 The coefficient of determination is used to evaluate the model's predictive accuracy. During real-time operation, the optimization unit iteratively optimizes the operating parameters using a gradient descent algorithm (such as SGD or Adam) based on the current cost value output by the dynamic cost calculation module and the efficiency-cost curve predicted by the LSTM model. The objective function is to minimize the total operating cost while simultaneously meeting SO2 emission limits (e.g., ≤35 mg / Nm³). 3 () constraints.

[0058] This module can be deployed in desulfurization systems of coal-fired power plants, chemical plants, etc. By collecting and processing data in real time, it can dynamically adjust key parameters such as desulfurizing agent dosage, circulation pump frequency, and absorber level, thereby maintaining efficient and low-consumption operation under complex working conditions.

[0059] By combining deep learning modeling with real-time optimization algorithms, the system's adaptability to nonlinear and time-varying operating conditions has been significantly improved, achieving a dynamic balance between desulfurization efficiency and operating costs, and providing a solid foundation for the intelligent operation of desulfurization devices.

[0060] Furthermore, S3 includes: S31 employs a multi-layer LSTM network structure. The input layer includes the amount of desulfurizing agent, equipment power, flue gas flow rate, SO2 concentration, and corresponding historical optimization results.

[0061] Specifically, this step employs a multi-layer LSTM (Long Short-Term Memory) network structure. The input layer includes desulfurizer dosage, equipment power, flue gas flow rate, SO2 concentration, and corresponding historical optimization results. Its core lies in modeling the complex nonlinear relationship between desulfurization efficiency and operating costs through deep learning, thereby achieving adaptive optimization control. In some implementations, this LSTM network typically consists of 3 to 5 hidden layers, each containing 64 to 256 memory units to enhance the model's ability to model long-term dependencies on time-series data. After standardization, the input data is fed into the network in time-series format, where each time step includes the aforementioned five-dimensional features. The sampling frequency is typically set to 1 to 5 minutes to balance real-time performance and data continuity.

[0062] The LSTM network is trained using the Adam optimizer with a learning rate set between 0.001 and 0.0001 to balance convergence speed and stability. The loss function is either mean squared error (MSE) or weighted MSE to emphasize the importance of cost optimization. The model training dataset typically contains at least six months of historical operating data, covering desulfurizer dosing strategies, equipment operating status, and market electricity price fluctuations under different operating conditions. During model evaluation, R² (coefficient of determination) and MAE (mean absolute error) are used as the main performance indicators, requiring R² ≥ 0.95 and MAE ≤ 5% of cost fluctuation range to ensure the model has high-precision predictive capabilities.

[0063] This step is deployed in the adaptive optimization module of the desulfurization unit, receiving real-time operating cost data from the dynamic cost calculation module and iteratively updating the model based on historical optimization results. By constructing a nonlinear mapping relationship between desulfurization efficiency and cost, the system can automatically adjust key parameters such as desulfurizing agent dosage and circulation pump frequency under complex operating conditions such as sudden changes in flue gas flow, increased SO2 concentration, or fluctuating electricity prices, achieving the dual goals of minimizing costs and meeting emission standards.

[0064] S32 uses the gradient descent algorithm to fine-tune the LSTM model online to adapt to the changing trends of desulfurization efficiency and cost under current operating conditions.

[0065] Specifically, in the adaptive optimization module, online fine-tuning of the LSTM model using the gradient descent algorithm is a core technical step for dynamically optimizing the operating parameters of the desulfurization unit. This step continuously updates the parameters of the trained LSTM model based on real-time collected desulfurization efficiency and operating cost data, thereby improving the model's adaptability to current operating conditions.

[0066] LSTM (Long Short-Term Memory) networks, as deep learning models, are capable of processing time-series data and capturing the dynamic changes of desulfurization equipment at different operating stages. During online fine-tuning, the system takes current-time structured data (such as flue gas flow rate, SO2 concentration, desulfurizer dosage, and equipment power) as input and real-time calculated desulfurization efficiency and operating costs as output to construct a loss function (such as mean squared error (MSE) or a weighted loss function). The gradient descent algorithm, through backpropagation, iteratively updates the weight matrix (W, U, b) of the LSTM model based on the gradient of the loss function, thereby optimizing the model's prediction accuracy and control response capability for the current operating conditions.

[0067] During gradient descent, the learning rate is typically set between 0.001 and 0.01, employing an adaptive learning rate adjustment strategy (such as the Adam optimizer) to improve convergence efficiency. The batch size is set between 32 and 128, and the time step is set between 5 and 15 minutes based on the flue gas processing cycle. The convergence threshold of the loss function is generally set to 1e-4, and the number of epochs is dynamically adjusted based on model performance, typically not exceeding 10 epochs to ensure real-time online computation.

[0068] This step is applicable to real-time optimization scenarios for desulfurization units under complex operating conditions (such as sudden load changes, SO2 concentration fluctuations, and changes in electricity prices or desulfurizing agent prices). For example, in a flue gas desulfurization system of a coal-fired power plant, when the SO2 concentration in the flue gas suddenly increases, the system uses gradient descent to fine-tune the LSTM model online, quickly adjusting the desulfurizing agent dosage and circulation pump frequency to maintain emission compliance while minimizing operating costs.

[0069] By fine-tuning the LSTM model online using the gradient descent algorithm, the model's response speed and prediction accuracy to current operating conditions can be significantly improved, thereby achieving adaptive optimization of the desulfurization unit's operating parameters. This step effectively solves the lag problem of traditional fixed-parameter optimization methods in dynamic environments, improves the system's intelligence level and operational economy, and is a key technical support for realizing the "computation-optimization-adjustment-recomputation" closed-loop control architecture of this invention.

[0070] S4 feeds back the optimized operating parameters to the desulfurization unit's control system via industrial bus or wireless communication protocol, forming a closed-loop control architecture of "calculation-optimization-adjustment-recalculation" to continuously optimize the operating status.

[0071] Specifically, this step involves feeding back the optimized operating parameters to the desulfurization unit's control system via an industrial bus or wireless communication protocol, thereby constructing a closed-loop control architecture of "calculation-optimization-adjustment-recalculation". This step is the core link in achieving dynamic optimization and real-time control of the system, and its technical implementation is based on the deep integration of industrial automation communication standards and control logic.

[0072] The feedback unit interacts with the desulfurization unit's PLC (Programmable Logic Controller) or DCS (Distributed Control System) via industrial buses (such as Modbus TCP, PROFIBUS, PROFINET) or wireless communication protocols (such as MQTT, OPC UA over Wi-Fi, or 5G). Optimized operating parameters (such as desulfurizing agent dosage, circulating pump frequency, and absorber level setpoint) are output by the adaptive optimization module and sent to the control system after protocol encapsulation. Upon receiving the parameters, the control system adjusts the actuators (such as metering pumps, frequency converters, and regulating valves) in real time using PID control strategies or preset control logic to ensure that the desulfurization unit's operating status matches the optimization target.

[0073] The feedback process must meet real-time requirements of communication latency less than 50ms and data refresh rate not less than 1Hz to ensure the response speed of the control system. Simultaneously, parameter transmission must comply with industrial communication and information security standards such as IEC 61158 or IEC 62443 to ensure data integrity and anti-interference capabilities. The adjustment range of optimization parameters must be set according to the equipment performance boundaries. For example, the desulfurizing agent dosage should be controlled between 0.5 and 2.0 kg / m³ of flue gas, and the circulating pump frequency should be controlled within the range of 30 to 60 Hz to avoid equipment overload or a decrease in desulfurization efficiency.

[0074] This step is widely applicable to desulfurization systems that require continuous operation, such as those in coal-fired power plants, steel mills, and chemical plants. Through a closed-loop feedback mechanism, the system can automatically adjust its operating strategy under conditions such as flue gas load fluctuations, changes in desulfurizing agent prices, or equipment aging, thereby achieving dynamic cost minimization and emission compliance.

[0075] By constructing a closed-loop control architecture, the system can continuously optimize and adjust operating parameters in real time, significantly improving the energy efficiency and economy of the desulfurization unit. At the same time, the feedback mechanism enhances the system's adaptability and stability, providing a solid foundation for the intelligent operation of industrial desulfurization systems.

[0076] Also includes: S5. Construct a user profile based on the user's historical operation data, and generate personalized optimization suggestions based on the user profile to improve user operation efficiency and system user experience.

[0077] Specifically, the user behavior analysis module in this invention aims to improve user operation efficiency and system user experience by constructing user profiles and generating personalized optimization suggestions. This step is technically implemented based on in-depth mining and machine learning modeling of historical user operation data, specifically including four core stages: data collection, feature extraction, profile construction, and optimization strategy generation.

[0078] At the technical implementation level, the user behavior analysis module first obtains users' historical operation records from the user interaction and visualization module, including but not limited to parameter adjustment frequency, adjustment magnitude, operation time distribution, and maintenance task execution preferences. These data are then used for feature extraction through time series analysis and clustering algorithms (such as K-means or DBSCAN) to identify typical user operation patterns under different working conditions. Furthermore, the system employs rule-based feature engineering methods to extract key behavioral features such as "high-frequency adjustment periods," "parameter types preferred for manual intervention," and "delay time in responding to optimization suggestions," and combines these with user identity information (such as job role and operating permissions) to construct a multi-dimensional user profile.

[0079] User profile models typically include the following dimensions: operation frequency (times / hour), operation response time (seconds), parameter adjustment magnitude (percentage), maintenance task execution cycle (days), and optimization suggestion adoption rate (percentage). The system matches historical optimization strategies based on the similarity of user profiles, employing collaborative filtering or content-based recommendation algorithms to generate personalized parameter adjustment thresholds, maintenance reminder methods, and operation suggestion priorities.

[0080] This module is widely applicable to desulfurization unit operators, maintenance engineers, and system administrators. For example, for users who frequently manually adjust the desulfurizing agent dosage, the system can automatically recommend more refined automatic adjustment strategies; for users who prefer nighttime maintenance, the system can optimize maintenance task scheduling and reduce daytime production disruptions.

[0081] The technical benefits of this step lie in significantly improving the system's user-friendliness and operational efficiency. Through personalized optimization suggestions, users can respond more quickly to system changes, reduce operational errors, and improve the stability and economy of the desulfurization unit. Simultaneously, this module enhances the system's adaptability, making optimization strategies more aligned with actual user needs, thereby improving the overall level of intelligence and user satisfaction.

[0082] The online calculation method for desulfurization device operating costs and adaptive energy efficiency optimization of this invention enables real-time online calculation of desulfurization device operating costs and adaptive energy efficiency optimization, effectively reducing operating and maintenance costs while ensuring emission compliance, and improving the system's intelligence level and response efficiency.

[0083] Example 2 This invention proposes an integrated device 10 for online calculation of desulfurization unit operating costs and energy efficiency optimization, comprising: The data acquisition and cleaning module 100 is used to collect the operating parameters, environmental parameters and market data of the desulfurization unit in real time through IoT sensors, and to perform outlier detection, noise filtering and missing value filling on the collected data to form cleaned data. The data alignment and fusion module 200 is used to perform spatiotemporal alignment and correlation fusion of the cleaned operating parameters, environmental parameters and market data, construct structured data input to the multivariate dynamic cost model, and calculate the operating cost of the desulfurization unit in real time. The model training and optimization module 300 is used to train historical operating data and optimization results based on the LSTM deep learning algorithm, build a nonlinear relationship model between desulfurization efficiency and cost, and adaptively optimize the operating parameters of the desulfurization unit based on real-time cost calculation results. The parameter feedback and control module 400 is used to feed back the optimized operating parameters to the control system of the desulfurization unit through an industrial bus or wireless communication protocol, forming a closed-loop control architecture to continuously optimize the operating status.

[0084] The online calculation and energy efficiency adaptive optimization device for desulfurization equipment operation costs in this invention realizes real-time online calculation and energy efficiency adaptive optimization of desulfurization equipment operation costs, effectively reducing operation and maintenance costs while ensuring emission compliance, and improving the system's intelligence level and response efficiency.

[0085] Example 3 Traditional desulfurization (Desulfurization) system management systems suffer from the following technical deficiencies, hindering their intelligent and efficient development: In existing technologies, the operating costs of Desulfurization systems are typically calculated manually or through post-event accounting, failing to reflect real-time dynamic changes in desulfurizer usage, equipment power consumption, and maintenance costs. For example, key factors such as fluctuations in desulfurizer market prices and changes in flue gas flow cannot be promptly incorporated into the cost model, leading to lagging cost control strategies and increased unnecessary operating expenses. Traditional optimization methods rely on fixed parameter thresholds or manual experience adjustments, lacking adaptability to complex operating conditions. For instance, when the SO2 concentration in the flue gas suddenly increases, the system can only respond through a preset excessive desulfurizer dosage strategy, potentially leading to desulfurizer waste or emission exceeding risks, failing to minimize costs while ensuring compliance.

[0086] To overcome the aforementioned shortcomings of existing technologies, this invention provides an integrated system for online calculation of desulfurization unit operating costs and energy efficiency optimization. It includes a data acquisition and preprocessing module, a dynamic cost calculation module, an adaptive optimization module, a closed-loop control module, and a user interaction and visualization module, wherein: The data acquisition and preprocessing module is used to collect the operating parameters, environmental parameters and market data of the desulfurization unit in real time, and to clean, fuse and structure the data. The dynamic cost calculation module is connected to the output of the data acquisition and preprocessing module. It is used to build a multivariate dynamic cost model based on structured data, including desulfurizing agent costs, equipment power consumption, maintenance costs and labor costs, and to calculate the operating cost of the desulfurization unit in real time. The adaptive optimization module is connected to the output of the dynamic cost calculation module. It is used to train the historical operating data and optimization results based on the LSTM deep learning algorithm, build a nonlinear relationship model between desulfurization efficiency and cost, and adaptively optimize the operating parameters of the desulfurization unit according to the real-time cost calculation results. The closed-loop control module is connected to the output of the adaptive optimization module and is used to feed back the optimized operating parameters to the desulfurization unit, forming a closed-loop control architecture of "calculation-optimization-adjustment-recalculation". The user interaction and visualization module is bidirectionally connected to the data acquisition and preprocessing module, dynamic cost calculation module, adaptive optimization module, and closed-loop control module. It is used to display the optimization effect, operating status, and cost composition to the user and to receive the user's operation instructions.

[0087] As a further improvement to the technical solution of the present invention, the data acquisition and preprocessing module includes a multi-source data acquisition unit, a market data acquisition unit, a data cleaning unit, and a data fusion unit, wherein: The multi-source data acquisition unit is used to collect operating parameters such as desulfurizing agent dosage, equipment power, flue gas flow rate, and SO2 concentration, as well as environmental parameters such as temperature, humidity, and atmospheric pressure through IoT sensors. The market data acquisition unit is used to collect market data such as desulfurizer market prices and electricity prices through the API interface; The data cleaning unit is used to perform outlier detection, noise filtering, and missing value imputation on the collected data. The data fusion unit is used to perform spatiotemporal alignment and correlation fusion of cleaned operating parameters, environmental parameters, and market data to form structured data.

[0088] As a further improvement to the technical solution of the present invention, the dynamic cost calculation module includes a desulfurizing agent cost model unit, a maintenance cost model unit, a labor cost model unit, and a total cost calculation unit, wherein: The desulfurizing agent cost model unit is used to calculate the cost of desulfurizing agent based on the amount of desulfurizing agent used and the market price; the equipment power consumption model unit is used to calculate the equipment power consumption cost based on the equipment power and the electricity price. The maintenance cost model unit is used to calculate maintenance costs based on equipment uptime, failure frequency, and maintenance unit price. The labor cost model unit is used to calculate labor costs based on labor hours and wage levels. The total cost calculation unit is used to sum the calculation results of the above model units in a weighted manner to obtain the total operating cost of the desulfurization unit.

[0089] As a further improvement to the technical solution of the present invention, the adaptive optimization module includes an LSTM algorithm training unit and a real-time optimization unit, wherein: The LSTM algorithm training unit is used to train the LSTM network using historical operating data (such as desulfurizer dosage, equipment power, flue gas flow rate, SO2 concentration) and corresponding optimization results (such as cost reduction and efficiency improvement) to build a nonlinear relationship model between desulfurization efficiency and cost. The real-time optimization unit is used to adaptively optimize the operating parameters of the desulfurization unit (such as the amount of desulfurizing agent added and the frequency of the circulating pump) based on the real-time cost calculation results and the nonlinear relationship model, so as to minimize the cost while ensuring that the emissions meet the standards.

[0090] As a further improvement to the technical solution of the present invention, the closed-loop control module includes a feedback unit and a control unit, wherein: The feedback unit is used to feed back the optimized operating parameters (such as the amount of desulfurizing agent added and the frequency of the circulating pump) to the control system of the desulfurization unit through an industrial bus or wireless communication protocol. The control unit is used to control the desulfurization unit in real time based on the optimized operating parameters, forming a closed-loop control architecture of "calculation-optimization-adjustment-recalculation". The adjusted operating parameters will be used as new input data to participate in the next round of cost calculation and optimization.

[0091] As a further improvement to the technical solution of the present invention, a predictive maintenance module is also included. The predictive maintenance module is connected to the output end of the data acquisition and preprocessing module and is used to predict potential failures of the desulfurization device (such as pump wear and pipeline blockage) through vibration analysis, temperature monitoring and other means, and to provide maintenance suggestions (such as maintenance time and maintenance content) in combination with the equipment life prediction model, so as to reduce the risk of unplanned downtime and maintenance costs.

[0092] As a further improvement to the technical solution of the present invention, a blockchain security module is also included. The blockchain security module is connected to the output end of the data acquisition and preprocessing module and is used to encrypt and store key information such as operating data and optimization results through blockchain distributed ledger technology, and record them in an immutable manner to ensure data security and traceability, and support remote auditing and data verification by environmental regulatory authorities.

[0093] As a further improvement to the technical solution of the present invention, a user behavior analysis module is also included. The user behavior analysis module is bidirectionally connected to the user interaction and visualization module. It is used to analyze the user's operating habits and maintenance preferences through the user's historical operation data (such as parameter adjustment frequency and maintenance time selection), and to provide personalized optimization suggestions (such as parameter adjustment threshold and maintenance reminder method) in combination with user profiling technology, so as to reduce the operation threshold and improve the user experience.

[0094] This invention achieves real-time online calculation of desulfurization unit operating costs and global energy efficiency optimization through multi-source data fusion, LSTM deep learning-driven adaptive optimization, and closed-loop control architecture. While ensuring stable SO2 emission compliance, it reduces operating costs and improves desulfurization efficiency, offering faster optimization response compared to traditional methods. Simultaneously, it reduces unplanned downtime through a predictive maintenance module, ensures data immutability and compliance through a blockchain security module, and lowers the operational threshold through a user interaction module. This constructs an integrated system for online calculation of industrial desulfurization unit operating costs and energy efficiency optimization, combining precise calculation, intelligent optimization, closed-loop control, and secure interaction.

[0095] Example 4 like Figure 3 The desulfurization unit operating cost online calculation and energy efficiency optimization integrated system shown includes a data acquisition and preprocessing module 1, a dynamic cost calculation module 2, an adaptive optimization module 3, a closed-loop control module 4, and a user interaction and visualization module 5, wherein: Data acquisition and preprocessing module 1 is used to collect the operating parameters, environmental parameters and market data of the desulfurization unit in real time, and to clean, fuse and structure the data. The dynamic cost calculation module 2 is connected to the output of the data acquisition and preprocessing module 1. It is used to build a multivariate dynamic cost model based on structured data, including desulfurizing agent costs, equipment power consumption, maintenance costs and labor costs, and to calculate the operating cost of the desulfurization unit in real time. The adaptive optimization module 3 is connected to the output of the dynamic cost calculation module 2. It is used to train the historical operating data and optimization results based on the LSTM deep learning algorithm, build a nonlinear relationship model between desulfurization efficiency and cost, and adaptively optimize the operating parameters of the desulfurization unit according to the real-time cost calculation results. The closed-loop control module 4 is connected to the output of the adaptive optimization module 3 and is used to feed back the optimized operating parameters to the desulfurization unit, forming a closed-loop control architecture of "calculation-optimization-adjustment-recalculation". The user interaction and visualization module 5 is bidirectionally connected to the data acquisition and preprocessing module 1, the dynamic cost calculation module 2, the adaptive optimization module 3, and the closed-loop control module 4. It is used to display the optimization effect, operating status, and cost composition to the user and to receive the user's operation instructions.

[0096] Furthermore, the data acquisition and preprocessing module 1 includes a multi-source data acquisition unit, a market data acquisition unit, a data cleaning unit, and a data fusion unit, wherein: The multi-source data acquisition unit is used to collect operating parameters such as desulfurizing agent dosage, equipment power, flue gas flow rate, and SO2 concentration, as well as environmental parameters such as temperature, humidity, and atmospheric pressure through IoT sensors. The market data acquisition unit is used to collect market data such as desulfurizer market prices and electricity prices through the API interface; The data cleaning unit is used to perform outlier detection, noise filtering, and missing value imputation on the collected data. The data fusion unit is used to perform spatiotemporal alignment and correlation fusion of cleaned operating parameters, environmental parameters, and market data to form structured data.

[0097] Furthermore, the dynamic cost calculation module 2 includes a desulfurizer cost model unit, a maintenance cost model unit, a labor cost model unit, and a total cost calculation unit, wherein: The desulfurizing agent cost model unit is used to calculate the cost of desulfurizing agent based on the amount of desulfurizing agent used and the market price; the equipment power consumption model unit is used to calculate the equipment power consumption cost based on the equipment power and the electricity price. The maintenance cost model unit is used to calculate maintenance costs based on equipment uptime, failure frequency, and maintenance unit price. The labor cost model unit is used to calculate labor costs based on labor hours and wage levels. The total cost calculation unit is used to sum the calculation results of the above model units in a weighted manner to obtain the total operating cost of the desulfurization unit.

[0098] Furthermore, the adaptive optimization module 3 includes an LSTM algorithm training unit and a real-time optimization unit, wherein: The LSTM algorithm training unit is used to train the LSTM network using historical operating data (such as desulfurizer dosage, equipment power, flue gas flow rate, SO2 concentration) and corresponding optimization results (such as cost reduction and efficiency improvement) to build a nonlinear relationship model between desulfurization efficiency and cost. The real-time optimization unit is used to adaptively optimize the operating parameters of the desulfurization unit (such as the amount of desulfurizing agent added and the frequency of the circulating pump) based on the real-time cost calculation results and nonlinear relationship model through the gradient descent algorithm, so as to minimize the cost while ensuring that the emissions meet the standards.

[0099] Furthermore, the closed-loop control module 4 includes a feedback unit and a control unit, wherein: The feedback unit is used to feed back the optimized operating parameters (such as the amount of desulfurizing agent added and the frequency of the circulating pump) to the control system of the desulfurization unit through an industrial bus or wireless communication protocol. The control unit is used to control the desulfurization unit in real time based on the optimized operating parameters, forming a closed-loop control architecture of "calculation-optimization-adjustment-recalculation". The adjusted operating parameters will be used as new input data to participate in the next round of cost calculation and optimization.

[0100] Preferably, the system also includes a predictive maintenance module 6, which is connected to the output of the data acquisition and preprocessing module 1. It is used to predict potential failures of the desulfurization unit (such as pump wear and pipeline blockage) through vibration analysis, temperature monitoring and other means, and to provide maintenance suggestions (such as maintenance time and maintenance content) in combination with the equipment life prediction model, so as to reduce the risk of unplanned downtime and maintenance costs.

[0101] Preferably, the system also includes a blockchain security module 7, which is connected to the output of the data acquisition and preprocessing module 1. The blockchain security module 7 is used to encrypt and store key information such as operating data and optimization results through blockchain distributed ledger technology, and record them in an immutable manner to ensure data security and traceability, and support remote auditing and data verification by environmental regulatory authorities.

[0102] Preferably, the system also includes a user behavior analysis module 8, which is bidirectionally connected to the user interaction and visualization module 5. It is used to analyze the user's operating habits and maintenance preferences through the user's historical operation data (such as parameter adjustment frequency and maintenance time selection), and to provide personalized optimization suggestions (such as parameter adjustment threshold and maintenance reminder method) in combination with user profiling technology, so as to reduce the operation threshold and improve the user experience.

[0103] This invention is based on the integration and collaborative operation of the following modules: The data acquisition and preprocessing module 1 collects operating parameters such as desulfurizer dosage, equipment power, flue gas flow rate, and SO2 concentration in real time through IoT sensors, as well as environmental and market data such as temperature, humidity, and market prices; it cleans the data (outlier detection and noise filtering), fuses it (spatiotemporal alignment and correlation), and performs structured processing to provide high-quality input for subsequent calculations; Dynamic cost calculation module 2 constructs a multivariate dynamic cost model, integrating four sub-models: desulfurizer cost, equipment power consumption, maintenance cost, and labor cost; it calculates the total operating cost in real time and supports a visual display of cost composition. The adaptive optimization module 3 trains historical operating data based on the LSTM deep learning algorithm to establish a nonlinear relationship model between desulfurization efficiency and cost; based on real-time cost results, it automatically adjusts parameters such as desulfurizing agent dosage and circulation pump frequency to minimize costs while ensuring emission compliance. The closed-loop control module 4 feeds back the optimized parameters to the desulfurization unit control system, forming a dynamic cycle of "calculation-optimization-adjustment-recalculation"; through real-time adjustment, it ensures that the system is always in the optimal operating state. User interaction and visualization module 5 provides real-time data monitoring, optimization effect display and operation command input interface, supports personalized settings and remote collaboration, and reduces the user's operation threshold. Predictive maintenance module 6 is used to predict potential failures of the desulfurization unit (such as pump wear and pipeline blockage) through vibration analysis, temperature monitoring and other means, and to provide maintenance suggestions (such as maintenance time and maintenance content) in combination with the equipment life prediction model, so as to reduce the risk of unplanned downtime and maintenance costs.

[0104] Blockchain security module 7 is used to encrypt and store key information such as operational data and optimization results in an immutable manner through blockchain distributed ledger technology, ensuring data security and traceability, and supporting remote auditing and data verification by environmental regulatory authorities.

[0105] User behavior analysis module 8 is used to analyze users' operating habits and maintenance preferences through historical user operation data (such as parameter adjustment frequency and maintenance time selection), and to provide personalized optimization suggestions (such as parameter adjustment threshold and maintenance reminder method) in combination with user profiling technology, so as to reduce the operation threshold and improve the user experience.

[0106] This invention enables real-time online calculation of desulfurization unit operating costs and global optimization of energy efficiency. It introduces LSTM deep learning algorithm, closed-loop control architecture and multi-source data fusion technology to reduce operating costs and significantly improve the system's intelligence level while ensuring emission compliance. At the same time, by integrating predictive maintenance module, blockchain security module and user behavior analysis module, it further reduces maintenance costs, ensures data security and improves user experience.

[0107] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0108] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A method for online calculation of desulfurization unit operating costs and integrated energy efficiency optimization, characterized in that, include: S1 collects real-time operating parameters, environmental parameters, and market data of the desulfurization unit through IoT sensors, and performs outlier detection, noise filtering, and missing value filling on the collected data to form cleaned data. S2 aligns and integrates the cleaning operation parameters, environmental parameters and market data in time and space, and constructs structured data input to a multivariate dynamic cost model to calculate the operating cost of the desulfurization unit in real time. S3, based on the LSTM deep learning algorithm, trains historical operating data and optimization results to construct a nonlinear relationship model between desulfurization efficiency and cost, and adaptively optimizes the operating parameters of the desulfurization unit according to the real-time cost calculation results. S4 feeds back the optimized operating parameters to the desulfurization unit's control system via industrial bus or wireless communication protocol, forming a closed-loop control architecture to continuously optimize the operating status.

2. The method as described in claim 1, characterized in that, S1 includes: S11, anomaly detection is performed on continuously collected operating parameters using a sliding window algorithm, and data points that exceed the set threshold range are identified and marked; S12 uses a time-series-based interpolation algorithm to fill in missing values ​​and wavelet transform to denoise the noisy data.

3. The method as described in claim 1, characterized in that, The S2 includes: S21, Time synchronization processing is performed on the operating parameters and market data of different timestamps, and data alignment is achieved by interpolation or delay compensation methods; S22 uses an association rule mining algorithm to identify the coupling relationship between operating parameters and environmental parameters, and constructs a multi-dimensional feature vector for input to the cost model.

4. The method as described in claim 1, characterized in that, The S3 includes: S31 adopts a multi-layer LSTM network structure. The input layer includes the amount of desulfurizing agent, equipment power, flue gas flow rate, SO2 concentration and corresponding historical optimization results. S32 uses the gradient descent algorithm to fine-tune the LSTM model online to adapt to the changing trends of desulfurization efficiency and cost under current operating conditions.

5. The method as described in claim 1, characterized in that, Also includes: S5. Construct a user profile based on the user's historical operation data, and generate personalized optimization suggestions based on the user profile to improve user operation efficiency and system user experience.

6. An integrated device for online calculation of desulfurization unit operating costs and energy efficiency optimization, characterized in that, include: The data acquisition and cleaning module is used to collect the operating parameters, environmental parameters and market data of the desulfurization unit in real time through IoT sensors, and to perform outlier detection, noise filtering and missing value filling on the collected data to form cleaned data. The data alignment and fusion module is used to perform spatiotemporal alignment and correlation fusion of the cleaned operating parameters, environmental parameters and market data, and to build structured data input into a multivariate dynamic cost model to calculate the operating cost of the desulfurization unit in real time. The model training and optimization module is used to train historical operating data and optimization results based on the LSTM deep learning algorithm, build a nonlinear relationship model between desulfurization efficiency and cost, and adaptively optimize the operating parameters of the desulfurization unit based on real-time cost calculation results. The parameter feedback and control module is used to feed back the optimized operating parameters to the control system of the desulfurization unit through industrial bus or wireless communication protocol, forming a closed-loop control architecture to continuously optimize the operating status.

7. An integrated system for online calculation of desulfurization unit operating costs and energy efficiency optimization, characterized in that, It includes a data acquisition and preprocessing module, a dynamic cost calculation module, an adaptive optimization module, a closed-loop control module, and a user interaction and visualization module, among which: The data acquisition and preprocessing module is used to collect the operating parameters, environmental parameters and market data of the desulfurization unit in real time, and to clean, fuse and structure the data. The dynamic cost calculation module is connected to the output of the data acquisition and preprocessing module. It is used to build a multivariate dynamic cost model based on structured data, including desulfurizing agent costs, equipment power consumption, maintenance costs and labor costs, and to calculate the operating cost of the desulfurization unit in real time. The adaptive optimization module is connected to the output of the dynamic cost calculation module. It is used to train the historical operating data and optimization results based on the LSTM deep learning algorithm, build a nonlinear relationship model between desulfurization efficiency and cost, and adaptively optimize the operating parameters of the desulfurization unit according to the real-time cost calculation results. The closed-loop control module is connected to the output of the adaptive optimization module and is used to feed back the optimized operating parameters to the desulfurization unit to form a closed-loop control architecture. The user interaction and visualization module is bidirectionally connected to the data acquisition and preprocessing module, dynamic cost calculation module, adaptive optimization module, and closed-loop control module. It is used to display the optimization effect, operating status, and cost composition to the user and to receive the user's operation instructions.

8. The system as described in claim 7, characterized in that, The data acquisition and preprocessing module includes a multi-source data acquisition unit, a market data acquisition unit, a data cleaning unit, and a data fusion unit, wherein: The multi-source data acquisition unit is used to collect operating parameters such as desulfurizer dosage, equipment power, flue gas flow rate, SO2 concentration, and environmental parameters such as temperature, humidity, and atmospheric pressure through IoT sensors. The market data acquisition unit is used to collect market data on desulfurizing agent prices and electricity prices via API interface; The data cleaning unit is used to perform outlier detection, noise filtering, and missing value imputation on the collected data. The data fusion unit is used to perform spatiotemporal alignment and correlation fusion of cleaned operating parameters, environmental parameters, and market data to form structured data.

9. The system as described in claim 7, characterized in that, The dynamic cost calculation module includes a desulfurizer cost model unit, a maintenance cost model unit, a labor cost model unit, and a total cost calculation unit, wherein: The desulfurizing agent cost model unit is used to calculate the cost of desulfurizing agent based on the amount of desulfurizing agent used and the market price; the equipment power consumption model unit is used to calculate the equipment power consumption cost based on the equipment power and the electricity price. The maintenance cost model unit is used to calculate maintenance costs based on equipment uptime, failure frequency, and maintenance unit price. The labor cost model unit is used to calculate labor costs based on labor hours and wage levels. The total cost calculation unit is used to sum the calculation results of the above model units in a weighted manner to obtain the total operating cost of the desulfurization unit.

10. The system as described in claim 7, characterized in that, The adaptive optimization module includes an LSTM algorithm training unit and a real-time optimization unit, wherein: The LSTM algorithm training unit is used to train the LSTM network using historical operating data and corresponding optimization results, and to build a nonlinear relationship model between desulfurization efficiency and cost. The real-time optimization unit is used to adaptively optimize the operating parameters of the desulfurization unit based on the real-time cost calculation results and the nonlinear relationship model, using a gradient descent algorithm.

11. The system as described in claim 7, characterized in that, The closed-loop control module includes a feedback unit and a control unit, wherein: The feedback unit is used to feed back the optimized operating parameters to the control system of the desulfurization unit via industrial bus or wireless communication protocol; The control unit is used to control the desulfurization unit in real time based on the optimized operating parameters, forming a closed-loop control architecture of calculation-optimization-adjustment-recalculation. The adjusted operating parameters will be used as new input data to participate in the next round of cost calculation and optimization.

12. The system as described in claim 7, characterized in that, It also includes a predictive maintenance module, which is connected to the output of the data acquisition and preprocessing module. This module is used to predict potential failures of the desulfurization unit through vibration analysis and temperature monitoring, and to provide maintenance suggestions in conjunction with the equipment life prediction model.

13. The system as described in claim 7, characterized in that, It also includes a blockchain security module, which is connected to the output of the data acquisition and preprocessing module. The blockchain security module is used to encrypt and store key information of operation data and optimization results in an immutable manner through blockchain distributed ledger technology, and supports remote auditing and data verification by environmental regulatory authorities.

14. The system as described in claim 7, characterized in that, It also includes a user behavior analysis module, which is bidirectionally connected to the user interaction and visualization module. This module is used to analyze users' operating habits and maintenance preferences through historical user operation data and to provide personalized optimization suggestions in conjunction with user profiling technology.

Images