Energy-saving early warning control method and system in casting process

Abstract

The present invention discloses an energy-saving early warning control method and system for a casting process. The method comprises: obtaining casting environment data, power consumption data, fuel consumption data, and emission data, and performing data cleaning and normalization operations to obtain preprocessed multivariate time series data; inputting the multivariate time series data into a pretrained multivariate prediction model to obtain predicted values ​​for each variable, and comparing the predicted values ​​with actual measured values ​​to obtain prediction residuals; performing sequential Bayesian analysis based on the prediction residuals to obtain abnormal event probabilities for each variable, and performing probability fusion calculations and comprehensive evaluations to obtain an anomaly score; and generating an early warning signal when the anomaly score is determined to be greater than a preset anomaly threshold. The early warning signal is sent to the casting equipment, causing the casting equipment to automatically adjust operating parameters based on the early warning signal. This method can achieve accurate environmental early warning and real-time control during the casting process.

Description

Technical Field

[0001] The present invention relates to the field of casting, and in particular to an energy-saving early warning control method and system in a casting process. Background Art

[0002] As global environmental issues become increasingly severe, the foundry industry, a typical high-energy-consuming and high-emissions industry, faces immense pressure to conserve energy and reduce emissions. This has placed ever-stricter demands on foundries for energy efficiency and environmental pollution control. To meet these demands, the foundry industry urgently needs a control method that can monitor and provide early warnings for energy consumption and environmental data in real time, enabling timely implementation of energy-saving measures during production to reduce energy waste and environmental pollution.

[0003] In the existing technology, various environmental data and energy consumption data in the casting process are usually collected through sensors and monitoring instruments installed on the production line, and these data are transmitted to the data processing system for analysis, generating certain energy consumption reports and providing basic early warning functions.

[0004] In the process of data processing and analysis, existing technologies only use prediction models from a single data source, lacking in-depth analysis of the correlation between multivariate data under complex working conditions, which can easily lead to false alarms or missed alarms. The early warning mechanism lags behind the actual casting process and lacks real-time performance. Summary of the Invention

[0005] The present invention provides an energy-saving early warning control method and system in a casting process, so as to realize accurate environmental protection early warning and real-time control in the casting process.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides an energy-saving early warning control method in a casting process, comprising:

[0007] Obtain casting environment data, power consumption data, fuel consumption data, and emission data;

[0008] performing data cleaning and normalization operations on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain pre-processed multivariate time series data;

[0009] Input the multivariate time series data into a pre-trained multivariate prediction model to obtain a predicted value for each variable, and compare the predicted value with the actual measured value to obtain a prediction residual;

[0010] Based on the prediction residuals, sequential Bayesian analysis is performed to obtain the probability of abnormal events for each variable;

[0011] Based on the probability of the abnormal event, a probability fusion calculation is performed and a comprehensive evaluation is performed to obtain an abnormality score;

[0012] When it is determined that the abnormality score is greater than a preset abnormality threshold, an early warning signal is generated and sent to the casting equipment, so that the casting equipment automatically adjusts the operating parameters according to the early warning signal.

[0013] As an optional implementation manner, performing data cleaning and normalization operations based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain pre-processed multivariate time series data includes:

[0014] performing outlier detection and processing based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain noise reduction data;

[0015] interpolating the missing data in the noise-reduced data to obtain clean data;

[0016] Performing normalization processing on the clean data to obtain normalized data;

[0017] A time alignment operation is performed based on the normalized data to obtain preprocessed multivariate time series data.

[0018] As an optional implementation, the configuration process of the multivariate prediction model includes:

[0019] Taking historical multivariate time series data as input data, an initial multivariate prediction model is constructed based on the WOA-HKELM model;

[0020] Based on the initial multivariate prediction model, the parameter combination of the HKELM model is iterated through the whale optimization algorithm to perform model training;

[0021] When the number of training times is greater than or equal to the preset maximum number of training times, the training is determined to be completed and a multivariate prediction model is obtained.

[0022] As an optional implementation, the initial multivariate prediction model is based on the parameter combination of the HKELM model iterated by the whale optimization algorithm to perform model training, including:

[0023] Based on the initial multivariate prediction model, obtaining parameters of the whale optimization algorithm;

[0024] Based on the parameters of the whale optimization algorithm, randomly generate parameter combinations of the HKELM model;

[0025] Based on the parameter combination, fitness calculation is performed to obtain the fitness of each parameter combination;

[0026] Based on the fitness, whale optimization iteration is performed, and when the number of iterations is greater than or equal to the preset maximum number of iterations, the optimal parameter combination of the HKELM model is obtained;

[0027] Among them, the parameters of the whale optimization algorithm include population size and maximum number of iterations; the parameter combination of the HKELM model includes regularization parameter, kernel function parameter and kernel function weight.

[0028] As an optional implementation manner, performing fitness calculation based on the parameter combination to obtain the fitness of each parameter combination includes:

[0029] ;

[0030] in, Indicates parameter combination Adaptability; represents the regularization parameter; Represents the RBF kernel function parameters; and Represents the parameters of the Poly kernel function; represents the kernel function weight; Indicates the amount of sample data; Indicates the multivariate time series data, including the casting environment data, the power consumption data, the fuel consumption data, and the emission data; Indicates the The predicted value of multivariate time series data; Indicates the The actual value of the multivariate time series data.

[0031] As an optional implementation, performing sequential Bayesian analysis based on the prediction residuals to obtain the abnormal event probability of each variable includes:

[0032] ,

[0033] ,

[0034] in, Indicates at time The residuals of multivariate time series data are classified as outliers, and the probability of abnormal events is updated; Indicates at time The residuals of multivariate time series data are classified as normal values, and the probability of abnormal events is updated; It represents the probability that the system correctly detects an abnormal event, that is, the ratio of the system to successfully detect an abnormality when an abnormality actually occurs; It represents the probability of the system falsely detecting an abnormal event, that is, the ratio of the system falsely reporting an abnormality when no abnormality actually occurs; It represents the prior probability of an abnormal event occurring at time point t-1.

[0035] As an optional implementation, performing probability fusion calculation and comprehensive evaluation based on the abnormal event probability to obtain an abnormality score includes:

[0036] According to the abnormal event probability of each variable, the Bayesian network method is used to perform probability fusion calculation to obtain the fused comprehensive probability of the abnormal event;

[0037] Comparing the fused comprehensive probability of abnormal events with a preset probability threshold, performing a comprehensive evaluation, and obtaining an abnormality score;

[0038] The calculation formula for the comprehensive probability of abnormal events is as follows:

[0039] ;

[0040] in, represents the comprehensive probability of abnormal events after fusion; It represents the joint probability of abnormal events of each variable under given multivariate time series data; represents the abnormal event of the nth variable at time t; Indicates the total number of variables; Represents the multivariate time series data for the current observation.

[0041] As an optional embodiment, when it is determined that the abnormality score is greater than a preset abnormality threshold, a warning signal is generated, and the warning signal is sent to the casting equipment, so that the casting equipment automatically adjusts the operating parameters according to the warning signal, including:

[0042] Obtaining an anomaly score obtained from the multivariate prediction model;

[0043] Comparing the abnormality score with a preset abnormality threshold to determine whether the abnormality score is higher than the preset abnormality threshold;

[0044] When it is determined that the abnormality score is higher than the preset abnormality threshold, an early warning signal is generated and sent to the casting equipment through a pre-configured communication channel, so that the casting equipment automatically adjusts the operating parameters according to the early warning signal;

[0045] Iteratively updating the multivariate prediction model based on the warning signal and the current multivariate time series data as input data to obtain an updated multivariate prediction model;

[0046] The operating parameters include melting temperature, gas flow, pouring speed and cooling water flow.

[0047] In a second aspect, the present invention provides an energy-saving early warning control system for a casting process, comprising:

[0048] A data acquisition module is used to acquire casting environment data collected by sensor equipment, power consumption data and fuel consumption data collected by energy consumption metering equipment, and emission data collected by emission detection equipment;

[0049] a data cleaning module, configured to perform data cleaning and normalization operations based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain pre-processed multivariate time series data;

[0050] A model training module is used to construct an initial multivariate prediction model based on historical multivariate time series data as input data, and train the initial multivariate prediction model to obtain a multivariate prediction model;

[0051] The residual calculation module is used to input multivariate time series data into the pre-trained multivariate prediction model, obtain the predicted value of each variable and compare it with the actual measured value to obtain the prediction residual;

[0052] A probability calculation module is used to perform sequential Bayesian analysis based on the prediction residuals to obtain the probability of abnormal events for each variable;

[0053] A score calculation module is used to perform probability fusion calculation and comprehensive evaluation based on the probability of the abnormal event to obtain an abnormality score;

[0054] The early warning processing module is used to generate an early warning signal when it is determined that the abnormality score is greater than a preset abnormality threshold, and send the early warning signal to the casting equipment so that the casting equipment automatically adjusts the operating parameters according to the early warning signal.

[0055] In a third aspect, the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned energy-saving warning control methods in the casting process.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The present invention provides an energy-saving early warning control method in a casting process, which is executed by a server and includes: obtaining casting environment data collected by a sensor device, power consumption data and fuel consumption data collected by an energy consumption metering device, and emission data collected by an emission detection device; performing data cleaning and normalization operations based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain preprocessed multivariate time series data; inputting the multivariate time series data into a pre-trained multivariate prediction model to obtain a predicted value of each variable and performing a comparative calculation with the actual measured value to obtain a prediction residual; performing a sequential Bayesian analysis based on the prediction residual to obtain an abnormal event probability for each variable; performing a probability fusion calculation and a comprehensive evaluation based on the abnormal event probability to obtain an abnormal score; when it is determined that the abnormal score is greater than a preset abnormal threshold, generating a warning signal, and sending the warning signal to the casting equipment, so that the casting equipment automatically adjusts operating parameters according to the warning signal; wherein the multivariate prediction model is based on historical multivariate time series data as input data, constructing an initial multivariate prediction model, and training the initial multivariate prediction model to obtain a multivariate prediction model. The method collects environmental, energy consumption, and emissions data from the casting process, constructs and trains a multivariate prediction model, performs predictions and residual calculations on real-time monitoring data, calculates the probability of abnormal events through sequential Bayesian analysis, and then performs probability fusion and comprehensive evaluation to generate an anomaly score. When the anomaly score exceeds a preset threshold, the system generates a warning signal and sends it to the casting equipment, automatically adjusting operating parameters, thereby achieving real-time and accurate monitoring and early warning of abnormal emissions and energy consumption events. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is a flow chart of an energy-saving early warning control method in a casting process provided by an embodiment of the present invention;

[0059] Figure 2 1 is a flow chart of a multivariate prediction model configuration process provided by an embodiment of the present invention;

[0060] Figure 3 It is a structural schematic diagram of an energy-saving early warning control system in a casting process provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0062] As global environmental issues become increasingly severe, the foundry industry, a typical high-energy-consuming and high-emissions industry, faces tremendous pressure to save energy and reduce emissions. This places ever-stricter demands on foundries for energy efficiency and environmental pollution control. To meet these demands, the foundry industry urgently needs a control method that can monitor and provide early warnings for energy consumption and environmental data in real time, enabling timely implementation of energy-saving measures during production to reduce energy waste and environmental pollution.

[0063] In the existing technology, various environmental data and energy consumption data in the casting process are usually collected through sensors and monitoring instruments installed on the production line, and these data are transmitted to the data processing system for analysis, generating certain energy consumption reports and providing basic early warning functions.

[0064] In the process of data processing and analysis, existing technologies only use prediction models from a single data source, lacking in-depth analysis of the correlation between multivariate data under complex working conditions, which can easily lead to false alarms or missed alarms. The early warning mechanism lags behind the actual casting process and lacks real-time performance.

[0065] In order to solve the above problems, an energy-saving early warning control method in a casting process provided by an embodiment of the present application will be introduced and explained in detail through the following specific embodiments.

[0066] Reference Figure 1 The first embodiment of the present invention provides an energy-saving early warning control method in a casting process, comprising the following steps:

[0067] S11, obtaining casting environment data, power consumption data, fuel consumption data, and emission data;

[0068] S12, performing data cleaning and normalization operations based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain pre-processed multivariate time series data;

[0069] S13, inputting the multivariate time series data into a pre-trained multivariate prediction model to obtain a predicted value of each variable, and comparing the predicted value with the actual measured value to obtain a prediction residual;

[0070] S14, performing sequential Bayesian analysis based on the prediction residuals to obtain the abnormal event probability of each variable;

[0071] S15, performing probability fusion calculation and comprehensive evaluation based on the abnormal event probability to obtain an abnormality score;

[0072] S16, when it is determined that the abnormality score is greater than the preset abnormality threshold, a warning signal is generated, and the warning signal is sent to the casting equipment, so that the casting equipment automatically adjusts the operating parameters according to the warning signal.

[0073] In step S11 , casting environment data, power consumption data, fuel consumption data, and emission data are acquired.

[0074] It should be noted that casting environment data includes temperature and humidity data within the foundry, monitored by temperature and humidity sensors installed within the foundry. This data helps assess the impact of the working environment on workers and the stability of the casting process. Electricity consumption data is collected in real time through electricity meters and used to analyze energy efficiency, identify energy-saving potential, and identify cost control points. Fuel consumption data is monitored in real time through fuel metering equipment and used to assess energy costs and environmental impacts, guiding fuel selection and optimization. Emissions data is collected through emission monitoring equipment and used to assess the overall environmental impact of the casting process, thereby developing and implementing emission reduction measures and improving environmental management. This collected data allows for better management and optimization of the casting process, achieving environmentally friendly and sustainable development goals.

[0075] In step S12, data cleaning and normalization operations are performed based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain pre-processed multivariate time series data, including:

[0076] performing outlier detection and processing based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain noise reduction data;

[0077] interpolating the missing data in the noise-reduced data to obtain clean data;

[0078] Performing normalization processing on the clean data to obtain normalized data;

[0079] A time alignment operation is performed based on the normalized data to obtain preprocessed multivariate time series data.

[0080] Among them, outlier detection and processing is a data cleaning operation, which detects outliers by using statistical methods or machine learning methods, deletes them after finding them, or replaces them with interpolation methods to reduce the impact of noise on subsequent analysis. Missing data interpolation processing is to mark missing values ​​from the data and fill in the missing values ​​using methods such as linear interpolation, polynomial interpolation, and mean interpolation, thereby ensuring the integrity of the data. In the present invention, normalization is the process of converting features of different dimensions or ranges to the same dimension or range. Normalization processing can improve the stability and accuracy of the algorithm, especially when multimodal data is involved, ensuring that the data of each modality has equal importance in the analysis. Exemplarily, the normalization algorithm used in the present invention is the Z-score normalization algorithm. Of course, the normalization algorithm can also use the minimum-maximum normalization algorithm, the nonlinear normalization algorithm, etc., and the present invention is not limited to this.

[0081] It should be noted that the time alignment operation is a key step in ensuring the consistency of timestamps from different data sources, which is crucial for the analysis of multivariate time series data. Through the time alignment operation, the timestamps from different data sources can be ensured to be consistent, providing a reliable data basis for subsequent multivariate time series analysis. Exemplarily, the alignment method used in the present invention is the resampling method, which unifies the timestamps from different data sources to a fixed time interval. The time interval set in the present invention is hourly. Of course, the time interval can also be set to every minute, every second, etc., which is not limited by the present invention. In addition, the alignment method can also use linear interpolation, polynomial interpolation, etc., which is not limited by the present invention.

[0082] Reference Figure 2 In step S13, the multivariate time series data is input into a pre-trained multivariate prediction model to obtain the predicted value of each variable, and the predicted value is compared with the actual measured value to obtain the prediction residual. The configuration process of the multivariate prediction model includes:

[0083] S131, using historical multivariate time series data as input data, and building an initial multivariate prediction model based on the WOA-HKELM model;

[0084] S132, based on the initial multivariate prediction model, iterate the parameter combination of the HKELM model through the whale optimization algorithm to perform model training;

[0085] S133: When the number of training times is greater than or equal to the preset maximum number of training times, the training is determined to be completed, and a multivariate prediction model is obtained.

[0086] It should be noted that WOA stands for Whale Optimization Algorithm, a heuristic optimization algorithm that simulates the hunting behavior of whales to optimize parameters. HKELM stands for Hybrid Kernel Extreme Learning Machine, an improved algorithm based on the Extreme Learning Machine (ELM). By combining multiple kernel functions, it enhances the model's ability to fit complex data and is suitable for multivariate time series forecasting. The present invention uses the WOA algorithm to iteratively optimize the parameters of the HKELM model to find the optimal parameter combination and improve the model's predictive performance. For example, in an embodiment of the present invention, the Hybrid Kernel Extreme Learning Machine (HKELM) utilizes a hybrid Poly kernel function and an RBF kernel function. By combining the linear feature extraction capabilities of the Poly kernel function with the nonlinear mapping capabilities of the RBF kernel function, it achieves more accurate predictions for complex and variable casting process data. This approach can capture linear relationships in the data while effectively handling nonlinear dynamic changes, improving the accuracy and robustness of the multivariate prediction model.

[0087] The preset maximum number of training times is a key parameter that determines how well the model learns from the training data. If it is set too low, the model may not converge, making accurate detection impossible. If it is set too high, computational cost and time consumption may increase, leading to overfitting. In this embodiment, the maximum number of training times is set to 200. Depending on the application scenario and user needs, the maximum number of training times can also be set to 100, 300, etc., which is not limited by the present invention.

[0088] In step S132, based on the initial multivariate prediction model, the parameter combination of the HKELM model is iterated by the whale optimization algorithm to perform model training, including:

[0089] Based on the initial multivariate prediction model, obtaining parameters of the whale optimization algorithm;

[0090] Based on the parameters of the whale optimization algorithm, randomly generate parameter combinations of the HKELM model;

[0091] Based on the parameter combination, fitness calculation is performed to obtain the fitness of each parameter combination;

[0092] Based on the fitness, whale optimization iteration is performed, and when the number of iterations is greater than or equal to the preset maximum number of iterations, the optimal parameter combination of the HKELM model is obtained;

[0093] Among them, the parameters of the whale optimization algorithm include population size and maximum number of iterations; the parameter combination of the HKELM model includes regularization parameter, kernel function parameter and kernel function weight.

[0094] It should be noted that the whale optimization algorithm is a meta-heuristic optimization algorithm based on the behavior of whale groups. It mainly searches for the global optimal solution by simulating the encirclement hunting behavior, spiral position update and random search strategy during the humpback whale's hunting process. In an embodiment of the present invention, the whale optimization algorithm is used to optimize the parameter combination of the HKELM model. By adjusting the regularization parameters, kernel function parameters and kernel function weights, the HKELM model can achieve the best prediction effect when processing multivariate time series data in the casting process. Specifically, the whale optimization algorithm gradually approaches the optimal parameter combination through steps such as population initialization, fitness calculation, position update and iterative optimization to improve the prediction accuracy and generalization ability of the model. Finally, after multiple rounds of iterations, the whale optimization algorithm can output the optimal parameter combination of the HKELM model to ensure the optimal performance of the model in practical applications.

[0095] The population size represents the number of whales in the whale optimization algorithm used to search for the optimal solution, that is, the number of solutions simultaneously participating in the optimization process. The population size determines the algorithm's search capability. A larger population size improves the algorithm's diversity and global search capabilities, but also increases computational cost. A smaller population size helps accelerate convergence but may lead to local optima. In this embodiment, the population size is set to 100, meaning that the whale optimization algorithm simultaneously searches for 100 solutions for the optimal parameter combination. Of course, the population size can also be set to 50, 200, and so on, and this is not limited in this invention.

[0096] It is worth noting that the maximum number of iterations represents the maximum number of loops allowed by the algorithm during the optimization process. In each iteration, the whale group will continuously update its position, gradually approaching the global optimal solution. When the maximum number of iterations is reached, regardless of whether the algorithm has converged to the optimal solution, the optimization process will stop and the best parameter combination for the current iteration will be output. The setting of the maximum number of iterations needs to strike a balance between the convergence speed of the algorithm and the calculation time. This parameter determines the maximum number of loops that the algorithm can perform during the optimization process, ensuring that it converges to the optimal solution within a reasonable time. Exemplarily, in the present invention, the maximum number of iterations is set to 300. Depending on the application scenario and user needs, the maximum number of iterations can also be set to 200, 400, etc., and the present invention is not limited to this.

[0097] During the model training process, the fitness calculation is performed based on the parameter combination to obtain the fitness of each parameter combination, including:

[0098] ;

[0099] in, Indicates parameter combination Adaptability; represents the regularization parameter; Represents the RBF kernel function parameters; and Represents the parameters of the Poly kernel function; represents the kernel function weight; Indicates the amount of sample data; Indicates the multivariate time series data, including the casting environment data, the power consumption data, the fuel consumption data, and the emission data; Indicates the The predicted value of multivariate time series data; Indicates the The actual value of the multivariate time series data.

[0100] It is worth mentioning that fitness calculation is a key step in model training and directly affects the effect of parameter optimization. In this process, by calculating the objective function , measures the performance of each parameter combination in the multivariate time series data prediction task. Objective function Indicates parameter combination The fitness of the model is the prediction error of the model under a given parameter combination. Specifically, the objective function is calculated by the prediction value of n multivariate time series data. and actual value The fitness is calculated by summing up the mean squared errors and taking the average value. A smaller fitness value means that the parameter combination performs better in the prediction task, with smaller errors and higher prediction accuracy of the model. By optimizing this objective function, the whale optimization algorithm can continuously adjust the parameter combination. , in order to minimize the prediction error and thus obtain the optimal parameter settings of the model.

[0101] In step S14, sequential Bayesian analysis is performed based on the prediction residuals to obtain the abnormal event probability of each variable, including:

[0102] ,

[0103] ,

[0104] in, Indicates at time The residuals of multivariate time series data are classified as outliers, and the probability of abnormal events is updated; Indicates at time The residuals of multivariate time series data are classified as normal values, and the probability of abnormal events is updated; It represents the probability that the system correctly detects an abnormal event, that is, the ratio of the system to successfully detect an abnormality when an abnormality actually occurs; It represents the probability of the system falsely detecting an abnormal event, that is, the ratio of the system falsely reporting an abnormality when no abnormality actually occurs; It represents the prior probability of an abnormal event occurring at time point t-1.

[0105] It should be noted that sequential Bayesian analysis dynamically monitors multivariate time series data by continuously updating the probability of abnormal events. At each time point t, the probability of abnormal events is updated based on whether the residual is classified as an outlier or a normal value. Make adjustments. When classified as an outlier, the formula Update the probability of abnormal events. At this time, the model is more inclined to believe that there is an abnormality in the system, so according to the previous moment The probability of abnormal events , combined with the system's detection capabilities and false alarm rate To calculate the new abnormal probability. When classified as normal, use the formula Update the probability of abnormal events. At this time, the model is more inclined to believe that the system is normal, so it is also based on the previous moment , combined with the probability of system detection errors and correct detection rate , to adjust the probability of abnormal events. This probability update mechanism based on sequential Bayesian analysis can capture possible abnormal changes in the casting process in real time, ensuring that early warning signals are issued in time when potential hidden dangers may exist, thereby effectively achieving energy-saving and environmental protection early warning control.

[0106] In step S15, the probability fusion calculation and comprehensive evaluation are performed based on the abnormal event probability to obtain an abnormality score, including:

[0107] According to the abnormal event probability of each variable, the Bayesian network method is used to perform probability fusion calculation to obtain the fused comprehensive probability of the abnormal event;

[0108] Comparing the fused comprehensive probability of abnormal events with a preset probability threshold, performing a comprehensive evaluation, and obtaining an abnormality score;

[0109] The calculation formula for the comprehensive probability of abnormal events is as follows:

[0110] ;

[0111] in, represents the comprehensive probability of abnormal events after fusion; It represents the joint probability of abnormal events of each variable under given multivariate time series data; represents the abnormal event of the nth variable at time t; Indicates the total number of variables; Represents the multivariate time series data for the current observation.

[0112] It should be noted that the abnormal event probability of each variable reflects the possibility of abnormality of each variable at the current time t. For example, in an embodiment of the present invention, the abnormal event probability of each variable represents the possibility of abnormality in the casting environment, power consumption, fuel consumption, emissions, etc. during the casting process. In the sequential Bayesian analysis, the prediction residual and the detection capability of the system (such as the correct detection rate RD and the false alarm rate FAR) are used to gradually update the abnormal event probability of each variable to obtain these characteristic values ​​reflecting the system state at the current moment. The main function is to perform probability fusion calculation through the Bayesian network method to obtain a comprehensive probability of abnormal events. The comprehensive probability of abnormal events after fusion It is used to comprehensively evaluate the system and generate an anomaly score by comparing it with a preset probability threshold. The anomaly score directly affects the system's early warning response, ensuring timely warning signals when potential problems arise, guiding casting equipment to adjust operating parameters and achieve energy-saving and environmentally friendly control.

[0113] The preset probability threshold is a value set in advance to measure the comprehensive probability of abnormal events in the system. Whether it exceeds the normal range. This defines the maximum acceptable comprehensive probability of abnormal events under the current system state, and is usually set through statistical analysis of historical data or expert experience. For example, in this embodiment of the present invention, the preset probability threshold is set to 0.8. Depending on the application scenario and user needs, the preset probability threshold can also be set to 0.7, 0.9, etc., which is not limited by the present invention.

[0114] In step S16, when it is determined that the abnormality score is greater than a preset abnormality threshold, a warning signal is generated, and the warning signal is sent to the casting equipment, so that the casting equipment automatically adjusts the operating parameters according to the warning signal, including:

[0115] Obtaining an anomaly score obtained from the multivariate prediction model;

[0116] Comparing the abnormality score with a preset abnormality threshold to determine whether the abnormality score is higher than the preset abnormality threshold;

[0117] When it is determined that the abnormality score is higher than the preset abnormality threshold, an early warning signal is generated and sent to the casting equipment through a pre-configured communication channel, so that the casting equipment automatically adjusts the operating parameters according to the early warning signal;

[0118] Iteratively updating the multivariate prediction model based on the warning signal and the current multivariate time series data as input data to obtain an updated multivariate prediction model;

[0119] The operating parameters include melting temperature, gas flow, pouring speed and cooling water flow.

[0120] It should be noted that operating parameters refer to a set of key control variables used by casting equipment during the casting process, including melting temperature, gas flow rate, pouring speed, and cooling water flow rate. These parameters directly affect the efficiency of the casting process, product quality, energy consumption, and environmental impact. The setting of operating parameters is generally based on the casting process requirements and the equipment's operating manual, and is optimized through field trials and accumulated experience. These parameters need to be appropriately adjusted at different production stages to ensure the stability of the casting process and environmental protection and energy saving effects. Operating parameters are used to control and regulate the operation of the casting equipment. When the system detects an abnormality and generates an early warning signal, these operating parameters are automatically adjusted based on the early warning signal to correct possible deviations or failures, ensuring that the casting process is carried out under optimal conditions, thereby achieving energy conservation and environmental protection goals and improving production efficiency and product quality.

[0121] A preset anomaly threshold is a standard or limit set within the casting process monitoring system to determine whether the anomaly score obtained from the multivariate prediction model indicates potential energy consumption anomaly issues or risks in the current casting process. This threshold is determined based on a comprehensive consideration of historical data analysis, industry standards, expert experience, and product quality requirements. Properly setting the preset anomaly threshold is crucial for casting process monitoring. It not only helps promptly detect and address anomalies, preventing product quality issues, but also reduces unnecessary interventions and production interruptions, thereby improving production efficiency and economic benefits. In practice, the anomaly score ranges from 0 to 100, with 0 indicating a completely normal state with no signs of anomaly; 100 indicating the highest degree of anomaly, indicating a severe anomaly; and intermediate scores representing varying degrees of anomaly, with higher scores indicating greater severity. In this embodiment, the preset anomaly threshold is set to 80 based on historical data analysis, expert experience, and multiple experiments. When the anomaly score exceeds this threshold, the system automatically generates an early warning signal. Alternatively, the preset anomaly threshold can be set to 70, 90, or other values. Depending on the specific application scenario, one or more thresholds can be set. Specifically, a lower threshold value of 60 points can be set to identify minor abnormalities, and a higher threshold value of 90 points can be set to identify serious abnormalities. The present invention is not limited to this.

[0122] Reference Figure 3 The second embodiment of the present invention provides an energy-saving early warning control system in a casting process, comprising:

[0123] A data acquisition module is used to acquire casting environment data collected by sensor equipment, power consumption data and fuel consumption data collected by energy consumption metering equipment, and emission data collected by emission detection equipment;

[0124] a data cleaning module, configured to perform data cleaning and normalization operations based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain pre-processed multivariate time series data;

[0125] A model training module is used to construct an initial multivariate prediction model based on historical multivariate time series data as input data, and train the initial multivariate prediction model to obtain a multivariate prediction model;

[0126] The residual calculation module is used to input multivariate time series data into the pre-trained multivariate prediction model, obtain the predicted value of each variable and compare it with the actual measured value to obtain the prediction residual;

[0127] A probability calculation module is used to perform sequential Bayesian analysis based on the prediction residuals to obtain the probability of abnormal events for each variable;

[0128] A score calculation module is used to perform probability fusion calculation and comprehensive evaluation based on the probability of the abnormal event to obtain an abnormality score;

[0129] The early warning processing module is used to generate an early warning signal when it is determined that the abnormality score is greater than a preset abnormality threshold, and send the early warning signal to the casting equipment so that the casting equipment automatically adjusts the operating parameters according to the early warning signal.

[0130] It should be noted that the energy-saving early warning control system in the casting process provided by an embodiment of the present invention is used to execute all the process steps of the energy-saving early warning control method in the casting process of the above embodiment. The working principles and beneficial effects of the two correspond one to one, so they will not be repeated here.

[0131] In summary, the present invention provides an energy-saving early warning control method in a casting process, which is executed by a server and includes: obtaining casting environment data collected by a sensor device, power consumption data and fuel consumption data collected by an energy consumption metering device, and emission data collected by an emission detection device; performing data cleaning and normalization operations based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain preprocessed multivariate time series data; inputting the multivariate time series data into a pre-trained multivariate prediction model to obtain a predicted value of each variable and performing a comparison calculation with the actual measured value to obtain a prediction residual; performing a sequential Bayesian analysis based on the prediction residual to obtain the abnormal event probability of each variable; performing a probability fusion calculation and a comprehensive evaluation based on the abnormal event probability to obtain an abnormal score; when it is determined that the abnormal score is greater than a preset abnormal threshold, generating a warning signal, and sending the warning signal to the casting equipment, so that the casting equipment automatically adjusts the operating parameters according to the warning signal; wherein, the multivariate prediction model is based on historical multivariate time series data as input data, constructs an initial multivariate prediction model, and trains the initial multivariate prediction model to obtain a multivariate prediction model. The method collects environmental, energy consumption, and emissions data from the casting process, constructs and trains a multivariate prediction model, performs predictions and residual calculations on real-time monitoring data, calculates the probability of abnormal events through sequential Bayesian analysis, and then performs probability fusion and comprehensive evaluation to generate an anomaly score. When the anomaly score exceeds a preset threshold, the system generates a warning signal and sends it to the casting equipment, automatically adjusting operating parameters, thereby achieving real-time and accurate monitoring and early warning of abnormal emissions and energy consumption events.

[0132] The embodiment of the present invention further provides a terminal device. The terminal device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as an energy-saving early warning control program for a casting process. When the processor executes the computer program, the steps of the energy-saving early warning control method embodiments in the casting process described above are implemented, such as Figure 1 Alternatively, the processor implements the functions of the modules / units in the above-mentioned system embodiments when executing the computer program.

[0133] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0134] The terminal device may be a computing device such as a desktop computer, laptop, PDA, or smart tablet. The terminal device may include, but is not limited to, a processor and memory. Those skilled in the art will appreciate that the aforementioned components are merely examples of terminal devices and do not constitute a limitation of the terminal device. The terminal device may include more or fewer components than those described above, or a combination of certain components or different components. For example, the terminal device may also include input / output devices, network access devices, buses, and the like.

[0135] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor. The processor is the control center of the terminal device and connects various parts of the entire terminal device using various interfaces and lines.

[0136] The memory can be used to store the computer programs and / or modules. The processor implements the various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory and accessing the data stored in the memory. The memory may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function (such as a sound playback function or an image playback function); the data storage area may store data generated based on the use of the mobile phone (such as audio data, a phone book, etc.). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0137] If the module / unit integrated into the terminal device is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the process steps in the above-mentioned method embodiments by using a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include: any entity or system capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. It should be noted that the content of the computer-readable medium can be appropriately increased or decreased based on the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, based on legislation and patent practice, computer-readable media does not include electric carrier signals and telecommunication signals.

[0138] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement the present invention without inventive work.

[0139] The specific embodiments described above further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. An energy-saving early warning control method in a casting process, characterized in that: Executed by the server, including: Obtain casting environment data, power consumption data, fuel consumption data, and emission data; performing data cleaning and normalization operations on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain pre-processed multivariate time series data; Input the multivariate time series data into a pre-trained multivariate prediction model to obtain a predicted value for each variable, and compare the predicted value with the actual measured value to obtain a prediction residual; Based on the prediction residuals, sequential Bayesian analysis is performed to obtain the probability of abnormal events for each variable; Based on the probability of the abnormal event, a probability fusion calculation is performed and a comprehensive evaluation is performed to obtain an abnormality score; When it is determined that the abnormality score is greater than a preset abnormality threshold, a warning signal is generated, and the warning signal is sent to the casting equipment, so that the casting equipment automatically adjusts the operating parameters according to the warning signal; The configuration process of the multivariate prediction model includes: Taking historical multivariate time series data as input data, an initial multivariate prediction model is constructed based on the WOA-HKELM model; Based on the initial multivariate prediction model, the parameter combination of the HKELM model is iterated through the whale optimization algorithm to perform model training; When the number of training times is greater than or equal to the preset maximum number of training times, the training is determined to be completed and a multivariate prediction model is obtained; The model training is performed by iterating the parameter combination of the HKELM model through the whale optimization algorithm based on the initial multivariate prediction model, including: Based on the initial multivariate prediction model, obtaining parameters of the whale optimization algorithm; Based on the parameters of the whale optimization algorithm, randomly generate parameter combinations of the HKELM model; Based on the parameter combination, fitness calculation is performed to obtain the fitness of each parameter combination; Based on the fitness, whale optimization iteration is performed, and when the number of iterations is greater than or equal to the preset maximum number of iterations, the optimal parameter combination of the HKELM model is obtained; Among them, the parameters of the whale optimization algorithm include population size and maximum number of iterations; the parameter combination of the HKELM model includes regularization parameter, kernel function parameter and kernel function weight; The method of performing probability fusion calculation and comprehensive evaluation based on the probability of the abnormal event to obtain an abnormality score includes: According to the abnormal event probability of each variable, the Bayesian network method is used to perform probability fusion calculation to obtain the fused comprehensive probability of the abnormal event; Comparing the fused comprehensive probability of abnormal events with a preset probability threshold, performing a comprehensive evaluation, and obtaining an abnormality score; The calculation formula for the comprehensive probability of abnormal events is as follows: ； in, represents the comprehensive probability of abnormal events after fusion; It represents the joint probability of abnormal events of each variable under given multivariate time series data; represents the abnormal event of the nth variable at time t; Indicates the total number of variables; Represents the multivariate time series data for the current observation.

2. The energy-saving early warning control method in the casting process according to claim 1 is characterized in that: The step of performing data cleaning and normalization operations based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain pre-processed multivariate time series data includes: performing outlier detection and processing based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain noise reduction data; interpolating the missing data in the noise-reduced data to obtain clean data; Performing normalization processing on the clean data to obtain normalized data; A time alignment operation is performed based on the normalized data to obtain preprocessed multivariate time series data.

3. The energy-saving early warning control method in the casting process according to claim 1 is characterized in that: The fitness calculation is performed based on the parameter combination to obtain the fitness of each parameter combination, including: ； in, Indicates parameter combination Adaptability; represents the regularization parameter; Represents the RBF kernel function parameters; and Represents the Poly kernel function parameters; represents the kernel function weight; Indicates the amount of sample data; Indicates the multivariate time series data, including the casting environment data, the power consumption data, the fuel consumption data, and the emission data; Indicates the The predicted value of multivariate time series data; Indicates the The actual value of the multivariate time series data.

4. The energy-saving early warning control method in the casting process according to claim 1 is characterized in that: The sequential Bayesian analysis is performed based on the prediction residuals to obtain the abnormal event probability of each variable, including: ， ； in, Indicates at time The residuals of multivariate time series data are classified as outliers, and the probability of abnormal events is updated; Indicates at time The residuals of multivariate time series data are classified as normal values, and the probability of abnormal events is updated; It represents the probability that the system correctly detects an abnormal event, that is, the ratio of the system to successfully detect an abnormality when an abnormality actually occurs; It represents the probability of the system falsely detecting an abnormal event, that is, the ratio of the system falsely reporting an abnormality when no abnormality actually occurs; It represents the prior probability of an abnormal event occurring at time point t-1.

5. The energy-saving early warning control method in the casting process according to claim 1 is characterized in that: When it is determined that the abnormality score is greater than a preset abnormality threshold, a warning signal is generated, and the warning signal is sent to the casting equipment, so that the casting equipment automatically adjusts the operating parameters according to the warning signal, including: Obtaining an anomaly score obtained from the multivariate prediction model; Comparing the abnormality score with a preset abnormality threshold to determine whether the abnormality score is higher than the preset abnormality threshold; When it is determined that the abnormality score is higher than the preset abnormality threshold, an early warning signal is generated and sent to the casting equipment through a pre-configured communication channel, so that the casting equipment automatically adjusts the operating parameters according to the early warning signal; Iteratively updating the multivariate prediction model based on the warning signal and the current multivariate time series data as input data to obtain an updated multivariate prediction model; The operating parameters include melting temperature, gas flow, pouring speed and cooling water flow.

6. An energy-saving early warning control system in a casting process, characterized in that: A method for implementing an energy-saving early warning control method in a casting process as claimed in any one of claims 1 to 5, comprising: A data acquisition module is used to acquire casting environment data collected by sensor equipment, power consumption data and fuel consumption data collected by energy consumption metering equipment, and emission data collected by emission detection equipment; a data cleaning module, configured to perform data cleaning and normalization operations based on the casting environment data, the power consumption data, the fuel consumption data, and the emission data to obtain pre-processed multivariate time series data; A model training module is used to construct an initial multivariate prediction model based on historical multivariate time series data as input data, and train the initial multivariate prediction model to obtain a multivariate prediction model; The residual calculation module is used to input multivariate time series data into the pre-trained multivariate prediction model, obtain the predicted value of each variable and compare it with the actual measured value to obtain the prediction residual; A probability calculation module is used to perform sequential Bayesian analysis based on the prediction residuals to obtain the probability of abnormal events for each variable; A score calculation module is used to perform probability fusion calculation and comprehensive evaluation based on the probability of the abnormal event to obtain an abnormality score; The early warning processing module is used to generate an early warning signal when it is determined that the abnormality score is greater than a preset abnormality threshold, and send the early warning signal to the casting equipment so that the casting equipment automatically adjusts the operating parameters according to the early warning signal.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the energy-saving early warning control method in the casting process according to any one of claims 1 to 5.

Images