Intelligent accounting method for industrial water consumption

Abstract

The invention discloses an industrial water consumption intelligent accounting method, and relates to the technical field of industrial water consumption monitoring and intelligent accounting, and the method comprises the following steps: 1, receiving and analyzing a water consumption accounting demand, respectively obtaining historical data and real-time data from a historical data warehouse and a real-time data collection system, carrying out the preprocessing of the historical data and the real-time data, and carrying out the calculation of the real-time data; and generating a preprocessed data set. According to the industrial water consumption intelligent accounting method, real-time adaptive adjustment of time-varying factors in industrial water consumption accounting is realized by adopting multi-level data fusion, an improved deep learning network, data sliding window online learning, feedback closed-loop control and a layered increment updating strategy; the prediction accuracy and the system response speed of the nonlinear water consumption behavior are obviously improved, so that the defects of static offline training of an accounting model in the prior art are effectively overcome.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial water consumption monitoring and intelligent accounting, and specifically provides an intelligent accounting method for industrial water consumption. Background Art

[0002] In the existing industrial water consumption accounting methods, due to the complexity of industrial scenarios, such as multiple devices, multiple processes, and multiple water use types, the system needs to process multi-source heterogeneous data from sensors, production logs, environmental monitoring devices, etc. However, the existing technologies have the following key problems. That is, the industrial water use scenario has strong time-varying characteristics, such as production plan adjustment, equipment start-stop, and seasonal changes. Most of the existing accounting models are static or offline trained based on historical data, lacking real-time adaptability. For example, the non-linear relationship of cooling water consumption at high and low loads of equipment cannot be accurately captured by a linear model, resulting in the accumulation of water consumption prediction errors. Summary of the Invention

[0003] (1) Technical Problems to be Solved

[0004] In view of the deficiencies of the prior art, the present invention provides an intelligent accounting method for industrial water consumption, which solves the problems that most of the existing accounting models are static or offline trained based on historical data, lacking real-time adaptability, resulting in the accumulation of water consumption prediction errors.

[0005] (2) Technical Solutions

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: An intelligent accounting method for industrial water consumption, comprising the following steps:

[0007] Step 1: Receive and parse the water consumption accounting requirements, and respectively obtain historical data and real-time data from the historical data warehouse and the real-time data acquisition system; preprocess the obtained historical data and real-time data respectively, including data cleaning, outlier removal, missing data compensation, and multi-source data time synchronization, to generate a preprocessed data set;

[0008] Step 2: Select or construct an accounting model including a baseline offline training module and a real-time adaptive adjustment module from a preset dynamic accounting model library; use the accounting model to perform a preliminary prediction on the preprocessed data set to obtain a preliminary water consumption calculation value;

[0009] Step 3: Compare the preliminary calculated value with a preset standard value or real-time monitoring result to determine the prediction deviation; when the prediction deviation exceeds a predetermined tolerance, start an online incremental learning mechanism to adjust the key parameters of the accounting model in real time to capture the non-linear changes of cooling water under high and low load states;

[0010] Step 4: Recalculate the preprocessed data set based on the updated accounting model to generate the corrected industrial water consumption; output the corrected industrial water consumption, and store the latest model parameters in the dynamic model library for subsequent adaptive optimization.

[0011] Preferably, the accounting model consists of two main parts:

[0012] A baseline offline training module for establishing a preliminary prediction model using historical data;

[0013] A real-time adaptive adjustment module for dynamically correcting the preliminary prediction model according to online feedback data, so as to achieve real-time response to equipment start / stop, production plan adjustment and seasonal changes in industrial production.

[0014] Preferably, in the data preprocessing process of Step 1, a multi-level data fusion algorithm is introduced, and the algorithm includes:

[0015] Multi-channel data synchronization processing;

[0016] A feature reconstruction module for feature extraction and reconstruction of heterogeneous data from sensors, production logs and environmental monitoring devices;

[0017] A noise suppression unit for reducing data interference, so as to provide higher-quality input data for the subsequent accounting model.

[0018] Preferably, the real-time adaptive adjustment module adopts an improved deep learning network, and the network includes:

[0019] An online prediction sub-module based on the long short-term memory network (LSTM) for initially capturing temporal changes;

[0020] An adaptive fuzzy logic correction sub-module for real-time fuzzy processing and correction of prediction errors to achieve accurate modeling of the non-linear behavior of cooling water;

[0021] The process is as follows: Obtain the current water consumption data of the industrial site from the real-time data acquisition system, and preprocess the data, including cleaning, standardization and time series construction, to form a time series data set suitable for input to the deep learning model; and input it into the online prediction sub-module based on the long short-term memory network (LSTM); the LSTM module uses its temporal memory ability to capture the time-varying characteristics of industrial water consumption and generate a preliminary prediction value, which reflects the dynamic changes in the current and recent production processes; compare the preliminary prediction value output by the LSTM module with the actual water consumption data collected in real time, and calculate the prediction error, which is used to measure the deviation between the model prediction and the actual situation on site;

[0022] The calculated prediction error is passed as input to the adaptive fuzzy logic correction sub-module; this sub-module first performs fuzzy processing on the prediction error, mapping the error value to a fuzzy membership degree for a soft determination of the error. The fuzzy membership degrees include "low error", "medium error", and "high error"; according to preset adaptive fuzzy rules (the rules can be dynamically adjusted based on historical feedback data), the corresponding correction factor or correction amount is determined to form a real-time correction value; the correction value obtained from the adaptive fuzzy logic correction sub-module is combined with the preliminary prediction value of the LSTM module to generate a corrected final prediction result;

[0023] The final prediction result comprehensively considers the time series characteristics and real-time error correction, achieving an instant response to factors such as equipment start / stop, production plan adjustment, and seasonal changes;

[0024] The corrected prediction result is compared with the subsequent real-time collected data to form a new error feedback; using this feedback information, the weight parameters of the LSTM module in the deep learning network and the fuzzy logic correction rules are updated online to achieve the adaptive optimization of the model.

[0025] Preferably, the online incremental learning mechanism adopts a data sliding window strategy, combined with a hybrid historical and real-time data learning method, to perform phased and granular parameter updates on the accounting model, thereby ensuring the continuous stability and real-time adaptability of model prediction in the context of the dynamic changes in the industrial production environment;

[0026] The process of the online incremental learning mechanism performing phased and granular parameter updates on the accounting model includes: continuously collecting industrial water-related data from various real-time data sources (such as sensors, production logs, environmental monitoring devices); preprocessing the collected real-time data, including cleaning, outlier removal, missing data filling, and time synchronization, to ensure data quality and timeliness; at the same time, extracting historical data similar to the current industrial production environment from the historical data warehouse and performing the same preprocessing on it; then, according to the set time range or data volume, a data sliding window is established, which contains the real-time data within the latest period of time; the window slides forward continuously, and when new data enters, the old data is removed accordingly, ensuring that the model training is always based on the recent data dynamic changes; the real-time data within the sliding window is mixed with the historical data, and the historical data that best matches the current situation is selected according to production plan, equipment status, and seasonal factor indicators; a mixed data set is formed, which not only retains the long-term statistical characteristics obtained during the initial training of the model but also reflects the latest industrial water dynamics;

[0027] Coarse-tune the overall model parameters using the mixed dataset, update the global prediction model, and capture the overall trend of water consumption; construct sub-models for each water usage category respectively, and independently adjust their respective parameters to more accurately reflect the non-linear time-varying characteristics of each category; adopt an online incremental learning algorithm (such as online gradient descent or other online optimization methods) to perform fine-grained updates on parameters at different levels; use a smaller learning rate for the global model parameters to ensure overall stability; use a larger adjustment step for key sub-models or local parameters to quickly respond to real-time data fluctuations; the update process is carried out step by step within each time window to ensure that the model always keeps up with the latest data changes; the updated model is verified in real-time within a sliding window, and by comparing the model prediction values with the actual collected data, the prediction error is calculated; if the error still exceeds the preset tolerance, further refinement adjustments are triggered until the model prediction accuracy meets the requirements; the verification process forms a closed-loop feedback mechanism to provide a basis for the next parameter update; after online incremental learning and feedback regulation, the finally updated model parameters are stored in the dynamic model library; the next accounting task will directly call the latest model parameters to achieve continuous adaptive optimization of industrial water consumption accounting.

[0028] Preferably, the intelligent accounting method for industrial water consumption establishes a feedback closed-loop control mechanism. After the model outputs the prediction value, it is compared with the on-site actual collected data in real-time to form a dynamic feedback signal. This signal is used to automatically trigger the adjustment of model parameters, and through a feedback regulator, hierarchical control of the prediction errors of each water usage type is achieved.

[0029] The process of the feedback closed-loop control mechanism is as follows:

[0030] Input the calculated prediction error into the feedback signal generator, classify the error according to the size and distribution of the error, including slight, medium, and severe errors; generate a dynamic feedback signal, which contains the error level information of different water usage types and clearly reflects the deviation degree between each prediction component and the on-site actual value.

[0031] When the dynamic feedback signal shows that the prediction error of the water usage type exceeds the preset tolerance, the feedback regulator automatically activates the online incremental learning mechanism; the system selects corresponding adjustment strategies according to the error level information in the feedback signal and performs automatic fine-tuning on the key parameters in the model to narrow the gap between the prediction value and the actual data.

[0032] Preferably, the real-time adaptive adjustment module implements a hierarchical incremental update strategy, where different categories of water usage correspond to independent real-time prediction sub-models respectively. Each sub-model independently performs online parameter updates, and through an inter-layer cooperation mechanism, the global prediction performance is optimized.

[0033] The process of the real-time adaptive adjustment module implementing the hierarchical incremental update strategy:

[0034] From the pre - processed real - time data, according to the production process and on - site monitoring information, the data is automatically identified and split according to different water - using categories, where the different water - using categories include cooling water, process water, and auxiliary water; the data of each category separately enters the corresponding sub - data channel to ensure that the data received by each sub - model is targeted and highly relevant; for each water - using category, an independent real - time prediction sub - model is constructed, and each sub - model uses an online learning algorithm suitable for the characteristics of this category to capture the time - series dynamic characteristics of this category; each sub - model implements an independent online parameter update mechanism, and continuously adjusts its own weight through the data sliding window strategy to ensure that the predicted value is dynamically updated with the real - time data;

[0035] Subsequently, a hierarchical incremental update mechanism is implemented:

[0036] The first layer (local update): Each real - time prediction sub - model independently executes online parameter update using its own split data, introduces the latest data into the model training process, and completes the incremental update of the local model;

[0037] The second layer (global collaboration): After the local update is completed, an inter - layer collaboration mechanism is introduced, and the prediction results of each sub - model are fused through a global integration module to generate a global predicted value of the overall industrial water consumption; this inter - layer collaboration mechanism uses strategies such as dynamic weighted average, decision fusion, or neural network fusion to evaluate and correct the prediction results of each sub - model to ensure the optimization of the overall prediction performance;

[0038] The global predicted result is compared with the on - site actual collected data in real time to calculate the overall prediction error; then, according to the error analysis result, adjustment information is transmitted to each sub - model and the global integration module simultaneously through a feedback signal, so that while the local sub - models maintain independent updates, they can be further refined and adjusted based on the global collaboration feedback;

[0039] This closed - loop feedback mechanism ensures that the sub - models of different water - using types can "complement each other", fully consider the mutual influence between categories during adjustment, and improve the overall prediction accuracy;

[0040] After the sub - models are updated through hierarchical incremental update and feedback closed - loop regulation, their updated parameters are stored in the dynamic model library as the basis for subsequent accounting tasks; the global integration module outputs the final predicted result of industrial water consumption after collaborative correction and transmits the result to the upper - layer decision - making and monitoring system to achieve continuous optimization of the global prediction performance.

[0041] Preferably, the intelligent accounting system for industrial water consumption includes:

[0042] A data acquisition unit for real-time collecting industrial water-related data from multiple data sources (sensors, production logs, environmental monitoring devices);

[0043] A data preprocessing unit for cleaning, fusing, time-synchronizing, and feature-extracting the collected multi-source data;

[0044] A dynamic accounting model unit for making a preliminary water consumption prediction on the preprocessed data according to a preset dynamic accounting model;

[0045] A real-time adaptive update unit for realizing real-time adjustment of the accounting model parameters based on online feedback;

[0046] A feedback closed-loop control unit for generating a feedback signal after comparing the model prediction result with the actual data to trigger model update;

[0047] A result output unit for transmitting the corrected industrial water consumption accounting result to a management server for archiving.

[0048] Preferably, the real-time adaptive update unit includes:

[0049] A deep neural network adjustment module for predicting the error of the model based on online collected data;

[0050] A data sliding window online learning module for updating the model parameters batch by batch and time period by time period;

[0051] The two modules cooperate to form a real-time error compensation system to achieve fast response to data fluctuations and dynamic optimization of the model.

[0052] A computer-readable storage medium storing a computer program, which can implement all or part of the steps of the intelligent industrial water consumption accounting method when the computer program is executed by a processor.

[0053] (III) Advantageous Effects

[0054] The present invention provides an intelligent industrial water consumption accounting method, having the following advantageous effects:

[0055] (I). By adopting multi-level data fusion, an improved deep learning network, data sliding window online learning, feedback closed-loop control, and a hierarchical incremental update strategy, the intelligent industrial water consumption accounting method realizes real-time adaptive adjustment of time-varying factors in industrial water consumption accounting, significantly improves the prediction accuracy of non-linear water consumption behaviors and the system response speed, and thus effectively overcomes the deficiencies of static offline training of the accounting model in the prior art.

[0056] (2). The intelligent accounting method for industrial water consumption realizes independent online prediction and hierarchical incremental update of different types of water consumption, and optimizes the global prediction results through an inter-layer collaboration mechanism, so as to ensure that the system can be adaptively adjusted in real time and accurately account for the water consumption of various types when dealing with the start and stop of industrial on-site equipment, production plan adjustment and seasonal changes. Description of the Drawings

[0057] Figure 1 It is a schematic flow chart of the intelligent accounting method for industrial water consumption of the present invention;

[0058] Figure 2 It is a schematic framework diagram of the intelligent accounting system for industrial water consumption of the present invention. Detailed Embodiments

[0059] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0060] Embodiment 1, please refer to Figure 1 , the present invention provides a technical solution: an intelligent accounting method for industrial water consumption, including the following steps:

[0061] Step 1: Receive and parse the water consumption accounting requirements, and respectively obtain historical data and real-time data from the historical data warehouse and the real-time data acquisition system; perform preprocessing on the obtained historical data and real-time data respectively, including data cleaning, outlier removal, missing data compensation, and multi-source data time synchronization, to generate a preprocessed data set;

[0062] Step 2: Select or construct an accounting model including a baseline offline training module and a real-time adaptive adjustment module from the preset dynamic accounting model library; use the accounting model to perform a preliminary prediction on the preprocessed data set to obtain a preliminary water consumption calculation value;

[0063] Step 3: Compare the preliminary calculated value with the preset standard value or the real-time monitoring result to determine the prediction deviation; when the prediction deviation exceeds the predetermined tolerance, start the online incremental learning mechanism to adjust the key parameters of the accounting model in real time to capture the nonlinear changes of the cooling water under high-load and low-load states;

[0064] Step 4: Recalculate the preprocessed data set based on the updated accounting model to generate the corrected industrial water consumption; output the corrected industrial water consumption, and store the latest model parameters in the dynamic model library for subsequent adaptive optimization use.

[0065] The accounting model consists of two main parts:

[0066] A baseline offline training module for establishing a preliminary prediction model using historical data;

[0067] A real-time adaptive adjustment module for dynamically correcting the preliminary prediction model according to online feedback data, so as to achieve real-time response to equipment start / stop, production plan adjustment, and seasonal changes in industrial production;

[0068] The implementation process of the baseline offline training module for establishing a preliminary prediction model using historical data is as follows:

[0069] Extract historical water consumption data related to the industrial site from the industrial water consumption historical data warehouse, including various production logs, equipment operation records, and previously collected sensor data; perform data cleaning on the collected historical data to eliminate outliers and incorrect data; perform missing value filling and time synchronization processing to ensure data continuity and temporal consistency; through data standardization and normalization processing, eliminate the dimensional differences between different data sources; based on historical data, extract key features related to industrial water consumption (such as daily water consumption, equipment operation status, production cycle, seasonal changes, etc.); segment or cluster the data to construct a feature dataset that meets the prediction requirements, and label the data to provide a basis for subsequent training;

[0070] According to the preprocessed feature dataset, select a suitable prediction model (such as a regression model, neural network model, or decision tree model); use an offline training algorithm to optimize the model parameters and construct a preliminary industrial water consumption prediction model to capture the long-term trends and patterns in historical data; use cross-validation and the holdout method to evaluate the accuracy of the trained model; according to the evaluation results, perform model parameter tuning and structure optimization to ensure that the fitting effect of the model on historical data meets the expected requirements;

[0071] Store the verified and optimized baseline prediction model and its parameters in the dynamic model library; this baseline model serves as the initial reference for the subsequent real-time adaptive adjustment module, providing a solid starting point for real-time accounting and online incremental learning;

[0072] Through the baseline offline training module, a preliminary industrial water consumption prediction model can be established using historical data, providing a reliable basis for the subsequent real-time adaptive mechanism, thus effectively solving the problem of insufficient real-time adaptability caused by the static and offline training of existing accounting models;

[0073] The process of the real-time adaptive adjustment module dynamically correcting the preliminary prediction model based on online feedback data: The system continuously collects the latest actual industrial water consumption data from various real-time data sources (such as on-site sensors, production logs, environmental monitoring equipment, etc.); preprocesses the collected data (such as data cleaning, time synchronization, etc.) to ensure the accuracy and timeliness of the feedback data; integrates the processed real-time feedback data with the output data of the preliminary prediction model to provide a basis for subsequent error analysis; compares the water consumption prediction value of the preliminary prediction model with the actually collected feedback data to calculate the prediction errors for each key time period and water consumption category; sets a predetermined error tolerance, and when the error exceeds this tolerance, it is considered that the current model cannot accurately reflect the on-site dynamic changes and needs to be corrected;

[0074] When it is detected that the prediction error exceeds the tolerance, an online incremental learning mechanism is automatically triggered; a data sliding window strategy is adopted to fuse the real-time feedback data within the latest period of time with some historical data to form a mixed data set for model update; this step ensures that the model correction can be based on the latest production situation and can capture the dynamic impacts caused by equipment start-stop, production plan adjustment, and seasonal changes in real time; uses an online learning algorithm (such as online gradient descent, online recursive neural network update, or other incremental learning methods suitable for real-time data processing) to update the key parameters of the preliminary prediction model;

[0075] During the parameter update process, a phased and granularity-based strategy is adopted:

[0076] Global parameter adjustment stage: First, coarsely adjust the overall model to ensure the correct capture of the overall trend;

[0077] Local refinement update stage: Carry out refined parameter optimization for different water consumption types (such as cooling water, process water, etc.) respectively to ensure the accurate response to non-linear and time-varying characteristics;

[0078] The updated model re-predicts the current data and compares it with the real-time feedback data again; further adjusts the model parameters according to the new error value until the prediction error drops within the predetermined tolerance, forming a continuous closed-loop feedback correction process; this process ensures that the system can continuously adaptively update when the industrial site changes and achieve real-time response; stores the updated model parameters obtained through online incremental learning in the dynamic model library as the initial model for subsequent accounting tasks; outputs the updated corrected prediction value for real-time guidance of industrial water consumption management and scheduling, and at the same time records the model update log for subsequent performance evaluation and further optimization; through the real-time adaptive adjustment module, the preliminary prediction model can be dynamically corrected based on online feedback data, so as to achieve real-time response and accurate prediction of the water consumption fluctuations caused by factors such as equipment start-stop, production plan adjustment, and seasonal changes in the industrial site.

[0079] In the data preprocessing process of Step 1, a multi-level data fusion algorithm is introduced, which includes: multi-channel data synchronization processing;

[0080] A feature reconstruction module for extracting and reconstructing features from heterogeneous data from sensors, production logs, and environmental monitoring devices;

[0081] A noise suppression unit for reducing data interference, thereby providing higher-quality input data for the subsequent accounting model.

[0082] The real-time adaptive adjustment module adopts an improved deep learning network, which includes:

[0083] An online prediction sub-module based on the Long Short-Term Memory network (LSTM) for initially capturing temporal changes;

[0084] An adaptive fuzzy logic correction sub-module for performing real-time fuzzification and correction on the prediction error to achieve accurate modeling of the non-linear behavior of the cooling water;

[0085] The process is as follows: Obtain the current water consumption data of the industrial site from the real-time data acquisition system, and preprocess the data, including cleaning, standardization, and time series construction, to form a time series data set suitable for input to the deep learning model; and input it into the online prediction sub-module based on the Long Short-Term Memory network (LSTM); the LSTM module uses its temporal memory ability to capture the time-varying characteristics of industrial water consumption and generates a preliminary prediction value, which reflects the dynamic changes in the current and recent production processes; compare the preliminary prediction value output by the LSTM module with the actual water consumption data collected in real time, and calculate the prediction error, which is used to measure the deviation between the model prediction and the actual situation on site;

[0086] Take the calculated prediction error as input and pass it to the adaptive fuzzy logic correction sub-module; this sub-module first performs fuzzification on the prediction error, maps the error value to a fuzzy membership degree for soft determination of the error, and the fuzzy membership degree includes "low error", "medium error", and "high error"; according to the preset adaptive fuzzy rules (the rules can be dynamically adjusted according to historical feedback data), determine the corresponding correction factor or correction amount to form a real-time correction value; combine the correction value obtained by the adaptive fuzzy logic correction sub-module with the preliminary prediction value of the LSTM module to generate a corrected final prediction result;

[0087] It should be further noted that in the specific implementation process, the final prediction result comprehensively considers the time series characteristics and real-time error correction, achieving an instant response to factors such as equipment startup and shutdown, production plan adjustment, and seasonal changes; comparing the corrected prediction result with the subsequent real-time collected data to form a new error feedback; using this feedback information to online update the weight parameters of the LSTM module and the fuzzy logic correction rules in the deep learning network to achieve the adaptive optimization of the model; storing the updated model parameters and correction rules in the dynamic model library for continuous use in subsequent predictions;

[0088] Through the above process, the online prediction sub-module based on the long short-term memory network (LSTM) in the improved deep learning network is responsible for initially capturing the time series changes of industrial water use, while the adaptive fuzzy logic correction sub-module performs real-time fuzzification processing and correction on the prediction error. The two work together to achieve a real-time adaptive response to the dynamic fluctuations in water consumption caused by complex factors such as equipment startup and shutdown, production plan adjustment, and seasonal changes in industrial production.

[0089] The online incremental learning mechanism adopts a data sliding window strategy and combines a hybrid historical and real-time data learning method to update the parameters of the accounting model in stages and with different granularities, so as to ensure the continuous stability and real-time adaptability of model prediction in the context of the dynamic changes in the industrial production environment;

[0090] The process of the online incremental learning mechanism updating the parameters of the accounting model in stages and with different granularities includes: continuously collecting industrial water use-related data from various real-time data sources (such as sensors, production logs, environmental monitoring equipment); preprocessing the collected real-time data, including cleaning, outlier removal, missing data filling, and time synchronization, to ensure the data quality and timeliness; at the same time, extracting historical data similar to the current industrial production environment from the historical data warehouse and performing the same preprocessing on it; then establishing a data sliding window according to the set time range or data volume, and this window contains the real-time data within the latest period; the window slides forward continuously, and when new data enters, the old data is removed accordingly to ensure that the model training is always based on the recent data dynamic changes; mixing the real-time data within the sliding window with the historical data and selecting the historical data that best matches the current situation according to the production plan, equipment status, and seasonal factor indicators; forming a hybrid data set that not only retains the long-term statistical characteristics obtained during the initial training of the model but also can reflect the latest industrial water use dynamics;

[0091] Coarse-tune the overall model parameters using a mixed dataset, update the global prediction model, and capture the overall trend of water consumption; construct sub-models for each water use category respectively, and independently adjust their respective parameters to more accurately reflect the non-linear time-varying characteristics of each category; adopt an online incremental learning algorithm (such as online gradient descent or other online optimization methods) to update the parameters at different levels in a granular manner; use a smaller learning rate for the global model parameters to ensure overall stability; use a larger adjustment step size for the key sub-models or local parameters to quickly respond to real-time data fluctuations; the update process is carried out step by step within each time window to ensure that the model always keeps up with the latest data changes; the updated model is verified in real time within a sliding window, and by comparing the model prediction values with the actual collected data, the prediction error is calculated; if the error still exceeds the preset tolerance, further refinement adjustments are triggered until the model prediction accuracy meets the requirements; the verification process forms a closed-loop feedback mechanism to provide a basis for the next parameter update; after online incremental learning and feedback regulation, the finally updated model parameters are stored in the dynamic model library; the next accounting task will directly call the latest model parameters to achieve continuous adaptive optimization of industrial water consumption accounting;

[0092] Through the above steps, the online incremental learning mechanism can use the data sliding window strategy and the mixed historical and real-time data learning method to achieve phased and granular parameter updates of the accounting model, thus effectively solving the problems of traditional model static offline training and lack of real-time adaptive ability.

[0093] The intelligent accounting method for industrial water consumption establishes a feedback closed-loop control mechanism. After the model outputs the prediction value, it is compared with the on-site actual collected data in real time to form a dynamic feedback signal. This signal is used to automatically trigger the adjustment of the model parameters, and through the feedback regulator, hierarchical control of the prediction errors of each water use type is achieved;

[0094] The process of the feedback closed-loop control mechanism is as follows:

[0095] Input the calculated prediction error into the feedback signal generator, classify the error according to the error magnitude and distribution, including slight, medium, and severe errors; generate a dynamic feedback signal, which contains the error level information of different water use types and clearly reflects the deviation degree between each prediction component and the on-site actual value;

[0096] When the dynamic feedback signal shows that the prediction error of a water use type exceeds the preset tolerance, the feedback regulator automatically activates the online incremental learning mechanism; the system selects the corresponding adjustment strategy according to the error level information in the feedback signal and automatically fine-tunes the key parameters in the model to narrow the gap between the prediction value and the actual data;

[0097] It should be further noted that in the specific implementation process, a hierarchical control strategy is adopted for different water use types through a feedback regulator, that is, different correction amplitudes and learning rates are set according to different error levels to achieve hierarchical control of the global model and local sub-models; the updated model parameters are immediately put into the prediction of the next cycle, and the new prediction results are compared with the on-site actual data again, thus forming a continuous closed-loop feedback process to continuously and adaptively optimize the model performance;

[0098] After closed-loop regulation, the updated model parameters and corresponding feedback signal data are stored in the dynamic model library as the basis for subsequent prediction and optimization; the system regularly evaluates the effect of the feedback closed-loop regulation and further optimizes the regulation algorithm according to the long-term feedback to ensure that the model always maintains a high adaptability to the dynamic changes in the industrial production site;

[0099] Through the above steps, the feedback closed-loop control mechanism can compare the model prediction value with the on-site actual data in real time, form a dynamic feedback signal, and automatically trigger the hierarchical control of the model parameters, so as to achieve accurate compensation and real-time adaptive adjustment of the prediction errors of various water use types.

[0100] The real-time adaptive adjustment module implements a hierarchical incremental update strategy, corresponding different categories of water use to independent real-time prediction sub-models respectively. Each sub-model independently performs online parameter update and realizes the optimization of the global prediction performance through an inter-layer cooperation mechanism;

[0101] The process of the real-time adaptive adjustment module implementing the hierarchical incremental update strategy:

[0102] From the preprocessed real-time data, according to the production process and on-site monitoring information, the data is automatically identified and shunted according to different water use categories. Among them, different water use categories include cooling water, process water, and auxiliary water; the data of each category enters the corresponding sub-data channel separately to ensure that the data received by each sub-model is targeted and highly relevant; for each water use category, an independent real-time prediction sub-model is constructed, and each sub-model adopts an online learning algorithm suitable for the characteristics of this category to capture the time-series dynamic characteristics of this category; each sub-model realizes an independent online parameter update mechanism and continuously adjusts its own weight through the data sliding window strategy to ensure that the prediction value is dynamically updated with the real-time data;

[0103] Subsequently, the hierarchical incremental update mechanism is implemented:

[0104] The first layer (local update): each real-time prediction sub-model independently performs online parameter update by using the shunted data respectively, introduces the latest data into the model training process, and completes the incremental update of the local model;

[0105] The second layer (global collaboration): After the local updates are completed, an inter-layer collaboration mechanism is introduced. The prediction results of each sub-model are fused through a global integration module to generate a global prediction value for the overall industrial water consumption. This inter-layer collaboration mechanism adopts strategies such as dynamic weighted average, decision fusion, or neural network fusion to evaluate and correct the prediction results of each sub-model to ensure the optimization of the overall prediction performance.

[0106] The global prediction result is compared with the on-site actual collected data in real time to calculate the overall prediction error. Then, according to the error analysis result, adjustment information is transmitted to each sub-model and the global integration module simultaneously through a feedback signal, enabling the local sub-models to be further refined and adjusted based on the global collaboration feedback while maintaining independent updates.

[0107] This closed-loop feedback mechanism ensures that the sub-models of different water use types can "complement each other", fully considering the mutual influence between categories during adjustment, and improving the overall prediction accuracy.

[0108] Each sub-model after hierarchical incremental update and feedback closed-loop adjustment stores its updated parameters in the dynamic model library as the basis for subsequent accounting tasks. The global integration module outputs the final industrial water consumption prediction result after collaborative correction and transmits the result to the upper-level decision-making and monitoring system to achieve the continuous optimization of the global prediction performance.

[0109] Through the above process, independent online prediction and hierarchical incremental update of different categories of water use are achieved, and the global prediction result is optimized through the inter-layer collaboration mechanism, thereby ensuring that the system can adapt in real time and accurately account for the water consumption of various categories when dealing with the start and stop of industrial site equipment, production plan adjustments, and seasonal changes.

[0110] Example 2, please refer to Figure 2 This intelligent industrial water consumption accounting system includes:

[0111] A data acquisition unit for real-time acquisition of industrial water-related data from multiple data sources (sensors, production logs, environmental monitoring devices).

[0112] A data preprocessing unit for cleaning, fusing, time synchronization, and feature extraction of the collected multi-source data.

[0113] A dynamic accounting model unit for making a preliminary water consumption prediction on the preprocessed data according to a preset dynamic accounting model.

[0114] A real-time adaptive update unit for real-time adjustment of the accounting model parameters based on online feedback.

[0115] A feedback closed-loop control unit for generating a feedback signal after comparing the model prediction result with the actual data to trigger model update.

[0116] A result output unit transmits the corrected industrial water consumption accounting result to a management server for archiving.

[0117] The real-time adaptive update unit includes:

[0118] A deep neural network adjustment module for predicting errors in the model based on online collected data;

[0119] A data sliding window online learning module for updating model parameters in batches and by time periods;

[0120] The two modules cooperate to form a real-time error compensation system to achieve fast response to data fluctuations and dynamic optimization of the model.

[0121] A computer-readable storage medium stores a computer program which, when executed by a processor, can implement all or part of the steps of the intelligent accounting method for industrial water consumption;

[0122] The above technologies adopt multi-level data fusion, an improved deep learning network, data sliding window online learning, feedback closed-loop control, and a hierarchical incremental update strategy, achieving real-time adaptive adjustment of time-varying factors in industrial water consumption accounting, significantly improving the prediction accuracy of non-linear water use behaviors and the system response speed, thus effectively overcoming the deficiencies of static offline training of accounting models in the prior art.

[0123] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, article or device comprising the element.

[0124] Although embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent calculation method for industrial water consumption, characterized in that: The steps include: Step 1: Receive and analyze water consumption accounting requirements, and simultaneously obtain historical data and real-time data from the historical data warehouse and real-time data collection system, respectively, and pre-process them to generate a pre-processed data set; Step 2: Select or construct a calculation model including a baseline offline training module and a real-time adaptive adjustment module from a preset dynamic calculation model library; use the calculation model to make a preliminary prediction on the preprocessed data set to obtain a preliminary water consumption calculation value; Step 3: Compare the preliminary calculated value with the preset standard value or real-time monitoring result to determine the prediction deviation; when the prediction deviation exceeds the preset tolerance, start the online incremental learning mechanism to adjust the key parameters of the accounting model in real time to capture the nonlinear changes of cooling water under high load and low load conditions; Step 4: Recalculate the preprocessed data set based on the updated accounting model to generate the corrected industrial water consumption; The corrected industrial water consumption is output, and the latest model parameters are stored in a dynamic model library for subsequent adaptive optimization.

2. According to claim 1, the intelligent calculation method for industrial water consumption is characterized by: The calculation model consists of two main parts: A baseline offline training module, used to build a preliminary prediction model using historical data; The real-time adaptive adjustment module is used to dynamically modify the preliminary prediction model according to online feedback data, thereby achieving real-time response to equipment start-up and shutdown, production plan adjustment and seasonal changes in industrial production.

3. The method for intelligent calculation of industrial water consumption according to claim 2 is characterized by: In the data preprocessing process of step 1, a multi-level data fusion algorithm is introduced, which includes: Multi-channel data synchronous processing; Feature reconstruction module, used to extract and reconstruct features from heterogeneous data from sensors, production logs, and environmental monitoring equipment; The noise suppression unit is used to reduce data interference, thereby providing higher quality input data for subsequent accounting models.

4. The method for intelligent calculation of industrial water consumption according to claim 3 is characterized by: The real-time adaptive adjustment module adopts an improved deep learning network, which includes: An online prediction submodule based on the long short-term memory network (LSTM) is used to initially capture the temporal changes; Adaptive fuzzy logic correction submodule, used for real-time fuzzification processing and correction of prediction errors; Among them, the use process of the real-time adaptive adjustment module is as follows: the current water consumption data of the industrial site is obtained from the real-time data acquisition system, and the data is preprocessed to form a time series data set; and input into the online prediction submodule based on the long short-term memory network (LSTM); the LSTM module uses its time series memory capability to capture the time-varying characteristics of industrial water consumption and generate a preliminary prediction value; the output preliminary prediction value is compared with the actual water consumption data collected in real time, and the prediction error is calculated; the prediction error is passed as input to the adaptive fuzzy logic correction submodule; the submodule first fuzzifies the prediction error and maps the error value to fuzzy membership, which includes low error, medium error, and high error; according to the preset adaptive fuzzy rules, the corresponding correction factor or correction amount is determined to form a real-time correction value; the correction value is combined with the preliminary prediction value of the LSTM module to generate the corrected final prediction result.

5. The method for intelligent calculation of industrial water consumption according to claim 4 is characterized by: The online incremental learning mechanism adopts a data sliding window strategy, combined with a hybrid historical and real-time data learning method, to update the parameters of the accounting model; The process of updating the parameters of the accounting model in stages and granularity by the online incremental learning mechanism includes: continuously collecting industrial water-related data from various real-time data sources; preprocessing the collected real-time data, and at the same time, extracting historical data similar to the current industrial production environment from the historical data warehouse and performing the same preprocessing on it; then establishing a data sliding window according to the set time range or data volume, the window contains the real-time data within the latest period of time; mixing the real-time data in the sliding window with the historical data, and selecting the historical data that best matches the current situation according to the production plan, equipment status, and seasonal factor indicators; forming a mixed data set; Use the mixed data set to make a rough adjustment to the overall model parameters, update the global prediction model, and capture the overall trend of water consumption; build sub-models for each water use category and adjust their parameters independently; use an online incremental learning algorithm (such as online gradient descent or other online optimization methods) to update parameters at different levels in a granular manner; the updated model is verified in real time within the sliding window, and the prediction error is calculated by comparing the model prediction value with the actual collected data; after online incremental learning and feedback adjustment, the final updated model parameters are stored in the dynamic model library.

6. The method for intelligent calculation of industrial water consumption according to claim 5 is characterized by: The intelligent calculation method for industrial water consumption establishes a feedback closed-loop control mechanism. After the model outputs the predicted value, it is compared with the actual collected data on site in real time to form a dynamic feedback signal. The signal is used to automatically trigger the adjustment of model parameters and realize the hierarchical regulation of the prediction error of each water type through the feedback regulator. The process of the feedback closed-loop control mechanism is as follows: The calculated prediction error is input into the feedback signal generator, and the error is graded according to the error size and distribution, including slight, medium and severe errors; a dynamic feedback signal is generated; When the dynamic feedback signal shows that the prediction error of the water use type exceeds the preset tolerance, the feedback regulator automatically activates the online incremental learning mechanism; the system selects the corresponding adjustment strategy based on the error level information in the feedback signal, and automatically fine-tunes the key parameters in the model to narrow the gap between the predicted value and the actual data.

7. The method for intelligent calculation of industrial water consumption according to claim 6 is characterized by: The real-time adaptive adjustment module implements a hierarchical incremental update strategy, corresponding different types of water use to independent real-time prediction sub-models, each sub-model independently performs online parameter updates, and optimizes the global prediction performance through an inter-layer coordination mechanism; The process of implementing the hierarchical incremental update strategy by the real-time adaptive adjustment module: From the pre-processed real-time data, according to the production process and on-site monitoring information, the data is automatically identified and divided according to different water use categories, including cooling water, process water, and auxiliary water; the data of each category enters the corresponding sub-data channel separately; for each water use category, an independent real-time prediction sub-model is constructed, and each sub-model adopts an online learning algorithm suitable for the characteristics of the category to capture the time series dynamic characteristics of the category; Then the hierarchical incremental update mechanism is implemented: First layer: Each real-time prediction sub-model independently performs online parameter updates using its own diverted data, introduces the latest data into the model training process, and completes the incremental update of the local model; Second layer: After the local update is completed, an inter-layer coordination mechanism is introduced. The prediction results of each sub-model are integrated through the global integration module to generate a global prediction value of the overall industrial water consumption. The inter-layer coordination mechanism uses dynamic weighted average, decision fusion or neural network fusion strategy to evaluate and correct the prediction results of each sub-model. The global prediction results are compared with the actual data collected on site in real time to calculate the overall prediction error. Then, based on the error analysis results, adjustment information is transmitted to each sub-model and the global integration module through feedback signals, so that the local sub-models can be further refined and adjusted based on the global collaborative feedback while maintaining independent updates. After hierarchical incremental updating and feedback closed-loop adjustment, each sub-model stores its updated parameters in the dynamic model library as the basis for subsequent accounting tasks; the global integration module outputs the final industrial water consumption forecast results after collaborative correction and passes the results to the upper-level decision-making and monitoring system.

8. An intelligent calculation system for industrial water consumption, characterized in that: The industrial water consumption intelligent accounting system includes: A data collection unit, used for collecting industrial water-related data from multiple data sources in real time; Data preprocessing unit, which cleans, fuses, synchronizes time and extracts features from the collected multi-source data; The dynamic accounting model unit makes a preliminary water consumption forecast for the preprocessed data according to the preset dynamic accounting model; Real-time adaptive update unit, which realizes real-time adjustment of accounting model parameters based on online feedback; The feedback closed-loop control unit compares the model prediction results with the actual data to generate a feedback signal, triggering the model update; The result output unit transmits the corrected industrial water consumption accounting results to the management server and archives them.

9. The intelligent industrial water consumption calculation system according to claim 8 is characterized by: The real-time adaptive updating unit comprises: A deep neural network adjustment module is used to predict the model error based on online collected data; Data sliding window online learning module, used to update model parameters in batches and time periods; The two modules work together to form a real-time error compensation system to achieve rapid response to data fluctuations and dynamic optimization of the model.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, all or part of the steps of the intelligent calculation method for industrial water consumption described in any one of claims 1 to 7 can be implemented.

Images