Visual management method and system for energy consumption fluctuation of intelligent mold factory

Abstract

The invention discloses a visual management method and system for energy consumption fluctuation of an intelligent mold factory, and the method comprises the steps: obtaining multi-source data from production equipment, and enabling the multi-source data to comprise sensor data, power monitoring data and production management data; analyzing the multi-source data by adopting a pre-established protocol conversion module to obtain standardized data; executing a data cleaning algorithm on the standardized data to generate a real-time energy consumption data set; according to the real-time energy consumption data set, a time sequence decomposition method is adopted to extract periodic features and trend features; modeling the periodic features, the trend features and the production parameters by adopting a support vector machine algorithm to obtain an energy consumption fluctuation prediction model; whether the output value of the energy consumption fluctuation prediction model exceeds a preset threshold value or not is judged, and if the output value of the energy consumption fluctuation prediction model exceeds the preset threshold value, an anomaly detection algorithm is triggered to locate the high-energy-consumption process. According to the method, energy consumption abnormity can be found in time, corresponding measures can be taken, and the production efficiency and the energy-saving and emission-reducing effects are improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular discloses a method and system for visually managing energy consumption fluctuations in an intelligent mold factory. Background Art

[0002] Energy consumption management in smart mold factories is an important research direction in the field of industrial manufacturing, and is of key significance for achieving energy conservation and emission reduction, improving production efficiency and meeting green compliance requirements.

[0003] With the advancement of Industry 4.0, mold factories are facing an increasingly urgent need for refined energy management, necessitating intelligent approaches to optimize energy efficiency. However, existing energy management methods suffer from widespread limitations, primarily manifested in incomplete data collection, a lack of dynamic adaptability in analytical models, and a lack of closed-loop optimization in management processes. These limitations make it difficult to accurately identify and effectively control energy consumption fluctuations. In particular, traditional methods often rely on manual inspections or static data analysis, making it difficult to capture multi-dimensional energy consumption changes in real time and unable to dynamically integrate with production scheduling to address energy price fluctuations.

[0004] Energy management in smart mold factories faces the following core technical challenges: First, the real-time collection and integration of heterogeneous multi-source data is difficult to ensure in terms of comprehensiveness and real-time performance due to complex equipment protocols and poor compatibility with legacy systems. Second, the precise modeling and prediction of energy consumption fluctuations is limited by the complexity of dynamic production parameters, resulting in inaccurate timing analysis. Finally, the automated generation and closed-loop tuning of optimization strategies is difficult due to the lack of a real-time feedback mechanism, making it difficult to continuously iterate energy-saving solutions. Unresolved technical challenges such as these make it difficult to provide early warnings of energy consumption anomalies, accurately locate high-energy-consuming processes, and dynamically adapt energy-saving strategies to production needs, creating unique technical challenges for energy management in smart mold factories.

[0005] Therefore, how to build a real-time and accurate energy consumption fluctuation visualization management system through efficient multi-source data collection, dynamic modeling and closed-loop optimization to achieve refined control and continuous optimization of mold factory energy consumption has become a key issue in this study. Summary of the Invention

[0006] The present invention provides a method and system for visual management of energy consumption fluctuations in an intelligent mold factory, aiming to solve at least one of the above-mentioned defects of existing energy consumption management methods in intelligent mold factories.

[0007] One aspect of the present invention relates to a method for visually managing energy consumption fluctuations in an intelligent mold factory, comprising the following steps:

[0008] Acquire multi-source data from production equipment, including sensor data, power monitoring data, and production management data;

[0009] Use pre-established protocol conversion modules to parse multi-source data and obtain standardized data;

[0010] Execute data cleaning algorithms on standardized data to generate real-time energy consumption datasets;

[0011] Based on the real-time energy consumption dataset, the time series decomposition method is used to extract periodic and trend features;

[0012] The support vector machine algorithm is used to model the periodic characteristics, trend characteristics and production parameters to obtain the energy consumption fluctuation prediction model;

[0013] Determine whether the output value of the energy consumption fluctuation prediction model exceeds the preset threshold. If the output value of the energy consumption fluctuation prediction model exceeds the preset threshold, the anomaly detection algorithm is triggered to locate the high-energy-consuming process.

[0014] Furthermore, the step of acquiring multi-source data from production equipment includes:

[0015] Acquire multi-source data from sensor devices, power monitoring equipment, and production management systems;

[0016] The multi-source data are formatted using the data acquisition protocol, and a multi-source data set with consistent format is generated by timestamp alignment to obtain a multi-source data set.

[0017] Furthermore, the steps of parsing the multi-source data using a pre-established protocol conversion module to obtain standardized data include:

[0018] Parsing the multi-source data using a pre-established protocol conversion module to generate parsed data and obtain the parsed data;

[0019] For the parsed data, use data verification tools to check the completeness and accuracy. If there are any missing or errors in the parsed data, fix them using the preset completion rules to obtain the verified data;

[0020] Based on the verified data, a data conversion tool is used to map it into a predefined standard data structure, and standardized data is generated through field alignment to determine the standardized data.

[0021] Furthermore, the steps of performing a data cleaning algorithm on the standardized data to generate a real-time energy consumption data set include:

[0022] Obtain the timestamp and energy consumption value records in the standardized data, and use the outlier detection tool to determine whether the energy consumption value exceeds the preset threshold. If it exceeds, it is marked as noise data to obtain the marked data set;

[0023] Based on the labeled data set, the missing energy consumption values ​​are filled using data completion rules, and the time fields are calibrated using the timestamp alignment tool to generate the completed data set.

[0024] For the completed data set, a deduplication algorithm is used to merge duplicate records, and the data is merged according to timestamps and device identifiers to determine the cleaned data set;

[0025] The cleaned data set is converted into a predefined structure through the field mapping tool, and the energy consumption values ​​are merged using data aggregation rules to generate a real-time energy consumption data set.

[0026] Furthermore, based on the real-time energy consumption dataset, the steps of extracting periodic features and trend features using a time series decomposition method include:

[0027] Obtain time series from the real-time energy consumption dataset, segment the time series using data granularity adjustment rules, and obtain an aligned time series set;

[0028] For the aligned time series set, the moving average algorithm is used to calculate the trend characteristics;

[0029] For the aligned time series set, Fourier transform tools are used to extract periodic features.

[0030] Furthermore, the support vector machine algorithm is used to model the periodic characteristics, trend characteristics and production parameters, and the steps of obtaining the energy consumption fluctuation prediction model include:

[0031] Obtain periodic characteristics, trend characteristics and production parameters from the data set, use data cleaning tools to remove outliers, and obtain a cleaned feature data set;

[0032] For the cleaned feature dataset, the support vector machine algorithm was used to build a model, and the hyperparameters were adjusted using the grid search tool to obtain the initial energy consumption fluctuation prediction model.

[0033] If the error of the initial energy consumption fluctuation prediction model exceeds the preset threshold, the validation data set is used for evaluation, and the model parameters are optimized through the cross-validation tool to obtain the optimized energy consumption fluctuation prediction model;

[0034] According to the optimized energy consumption fluctuation prediction model, the real-time production parameters and feature data sets are obtained, and the prediction values ​​are calculated through batch processing tools to obtain the energy consumption fluctuation prediction results.

[0035] Furthermore, determining whether the output value of the energy consumption fluctuation prediction model exceeds a preset threshold value, and if the output value of the energy consumption fluctuation prediction model exceeds the preset threshold value, triggering the anomaly detection algorithm to locate the high energy consumption process includes the following steps:

[0036] If the energy consumption fluctuation prediction result output by the energy consumption fluctuation prediction model exceeds the preset threshold, the energy consumption fluctuation prediction result is analyzed by the anomaly detection algorithm to locate the high energy consumption process and obtain a process identification list;

[0037] According to the process identification list, an alarm trigger tool is used to generate alarm information and process identification, and the alarm information and process identification are saved through a log recording tool to obtain an exception handling record.

[0038] Another aspect of the present invention relates to a visual management system for energy consumption fluctuations in an intelligent mold factory, which is used to implement the above-mentioned visual management method for energy consumption fluctuations in an intelligent mold factory. The visual management system for energy consumption fluctuations in an intelligent mold factory includes:

[0039] A first acquisition module is used to acquire multi-source data from production equipment, where the multi-source data includes sensor data, power monitoring data, and production management data;

[0040] The second acquisition module is used to parse the multi-source data using a pre-established protocol conversion module to obtain standardized data;

[0041] A generation module is used to perform data cleaning algorithms on standardized data to generate real-time energy consumption data sets;

[0042] An extraction module is used to extract periodic features and trend features based on the real-time energy consumption dataset using a time series decomposition method;

[0043] The third acquisition module is used to use the support vector machine algorithm to model the periodic characteristics, trend characteristics and production parameters to obtain an energy consumption fluctuation prediction model;

[0044] The judgment module is used to judge whether the output value of the energy consumption fluctuation prediction model exceeds the preset threshold. If the output value of the energy consumption fluctuation prediction model exceeds the preset threshold, the anomaly detection algorithm is triggered to locate the high-energy-consuming process.

[0045] Furthermore, the first acquisition module includes:

[0046] A first acquisition unit is used to acquire multi-source data from sensor equipment, power monitoring equipment and production management system;

[0047] The second acquisition unit is used to format the multi-source data using a data acquisition protocol, generate a multi-source data set with a consistent format through timestamp alignment, and obtain a multi-source data set.

[0048] Furthermore, the second acquisition module includes:

[0049] a third acquiring unit, configured to parse the multi-source data using a pre-established protocol conversion module to generate parsed data, and obtain the parsed data;

[0050] The fourth acquisition unit is used to check the integrity and accuracy of the parsed data using a data verification tool. If the parsed data is missing or erroneous, it is repaired using a preset completion rule to obtain verified data;

[0051] The determination unit is used to map the verified data into a predefined standard data structure using a data conversion tool, generate standardized data through field alignment, and determine the standardized data.

[0052] The beneficial effects achieved by the present invention are:

[0053] The present invention provides a method and system for visually managing energy consumption fluctuations in intelligent mold factories. This method uses a protocol conversion module to parse and standardize multi-source data from production equipment, and employs a data cleaning algorithm to generate a real-time energy consumption dataset. A time series decomposition method is then used to extract periodic and trend features, and a support vector machine algorithm is combined to establish an energy consumption fluctuation prediction model. When the output value of the prediction model exceeds a preset threshold, an anomaly detection algorithm is triggered to locate high-energy-consuming processes. This invention enables real-time analysis, prediction, and anomaly detection of energy consumption data in industrial production processes, facilitating the timely identification of energy consumption anomalies and the implementation of corresponding measures, thereby improving production efficiency and energy conservation and emission reduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flow chart of an embodiment of a method for visually managing energy consumption fluctuations in an intelligent mold factory according to the present invention. DETAILED DESCRIPTION

[0055] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0056] like Figure 1 As shown, the first embodiment of the present invention proposes a method for visually managing energy consumption fluctuations in an intelligent mold factory, comprising the following steps:

[0057] Step S100: Acquire multi-source data from production equipment, where the multi-source data includes sensor data, power monitoring data, and production management data.

[0058] Multi-source data refers to diverse data sets from different collection channels, storage media, or application scenarios. In this embodiment, multi-source data refers to sensor data, power monitoring data, and production management data collected from production equipment.

[0059] Step S200: parse multi-source data using a pre-established protocol conversion module to obtain standardized data.

[0060] A protocol conversion module is a hardware or software device used to enable data exchange between different communication protocols. Its core function is to resolve protocol compatibility issues across heterogeneous networks or systems. By parsing and reconstructing data formats and transmission rules, the protocol conversion module converts data from one protocol into a form recognizable by another, thus enabling cross-protocol communication.

[0061] Standardized data refers to the process of converting raw data into a unified format, scale or distribution form through normalization, aiming to eliminate dimensional differences or structural incompatibilities between data and improve the comparability and application efficiency of data.

[0062] Step S300: Execute a data cleaning algorithm on the standardized data to generate a real-time energy consumption data set.

[0063] Data cleaning algorithms are a set of techniques used to identify, correct, or remove errors, redundancies, inconsistencies, and invalid information from datasets. The core goal of data cleaning algorithms is to improve data quality through normalization, ensuring its reliability and suitability for subsequent analysis, modeling, and decision-making.

[0064] Real-time energy consumption datasets are dynamically updated sets of energy consumption data. Using sensors, smart metering devices, and IoT technologies, they collect and integrate real-time consumption information on electricity, gas, water, and other energy sources. This information is used to monitor, analyze, and optimize energy efficiency. The core characteristics of real-time energy consumption datasets are timeliness, continuity, and multimodality. They support data updates in seconds or minutes, facilitating real-time energy management decisions in industries such as industry, construction, and transportation.

[0065] Step S400: extracting periodic features and trend features based on the real-time energy consumption dataset using a time series decomposition method.

[0066] Time series decomposition is a statistical analysis method that breaks down time series data into trend, seasonality, and residual terms. It aims to reveal the long-term variation patterns, periodic fluctuations, and random interference components implicit in the data, providing a basis for prediction, anomaly detection, and pattern recognition.

[0067] Periodicity is a recurring, regular pattern of fluctuations in time series or data. Its magnitude and time intervals exhibit a fixed or predictable correlation, often driven by the external environment, human activity, or inherent system mechanisms. In time series analysis, periodicity is a core component, distinct from trend terms (long-term variation) and residual terms (random noise). It is used to reveal the inherent stability of data fluctuations and support tasks such as forecasting, optimization, and anomaly detection.

[0068] Trend characteristics are regular patterns in time series that reflect the direction of long-term changes in data. They manifest as continuous increases, decreases, or relative stability. They are driven by systemic factors (such as technological progress, economic policies, or natural laws) and are a core component of time series decomposition. The core role of trend characteristics is to reveal the macro-regimes behind the data, providing directional evidence for forecasting, decision-making, and anomaly detection.

[0069] Step S500: Using a support vector machine algorithm to model periodic characteristics, trend characteristics and production parameters, and obtaining an energy consumption fluctuation prediction model.

[0070] A support vector machine (SVM) is a supervised learning algorithm that classifies or regresses data by constructing a hyperplane (or set of hyperplanes). Its core goal is to maximize the classification margin to achieve high generalization performance. It is particularly adept at handling high-dimensional data and nonlinear separable problems. The core idea of ​​a support vector machine is to map data into a high-dimensional space and find the optimal decision boundary to minimize model complexity and prediction error.

[0071] The energy consumption fluctuation prediction model is a mathematical tool that uses historical and real-time data to quantitatively analyze energy consumption variations and predict future fluctuations. It aims to identify periodic, trend-based, and random fluctuations to optimize energy scheduling and equipment management strategies. The core of the energy consumption fluctuation prediction model is to construct a dynamic mapping relationship by integrating statistical methods, physical mechanisms, and machine learning techniques to capture the drivers of energy consumption changes (such as equipment performance degradation, ambient temperature, and production schedules).

[0072] Step S600: Determine whether the output value of the energy consumption fluctuation prediction model exceeds a preset threshold. If the output value of the energy consumption fluctuation prediction model exceeds the preset threshold, trigger the anomaly detection algorithm to locate the high energy consumption process.

[0073] Anomaly detection algorithms are computational methods that identify observations in data that significantly deviate from common patterns or distributions. They aim to uncover rare events, errors, or potential risks in the data and are widely used in fields such as fault diagnosis, fraud detection, and quality control. The core of anomaly detection algorithms is to model the "normal" behavior of the data, quantify the degree of abnormality of the observations (such as deviation probability, distance, or density difference), and then determine anomalies based on thresholds or statistical tests.

[0074] High-energy-consuming processes are specific links in the production process where the energy consumption intensity is significantly higher than the industry benchmark level. Their quantitative energy consumption indicators (such as energy consumption per unit product, energy cost ratio, etc.) exceed the limit standards stipulated by the state or industry, and energy consumption needs to be reduced through technological transformation or management optimization.

[0075] Furthermore, the method for visual management of energy consumption fluctuations in a smart mold factory provided in this embodiment includes step S100:

[0076] Step S110: Acquire multi-source data from sensor devices, power monitoring equipment, and production management systems.

[0077] Multi-source data collection and formatting are the basis for equipment monitoring in industrial production.

[0078] For example, temperature and vibration data are obtained from sensor devices, voltage and current data are obtained from power monitoring equipment, and information such as production plans and equipment status are obtained from production management systems.

[0079] The Modbus protocol is used to format sensor data into a unified JSON structure, and the OPCUA protocol is used to process power monitoring data to ensure consistency of data fields.

[0080] Step S120 : formatting the multi-source data using a data acquisition protocol, generating a multi-source data set with a consistent format by aligning timestamps, and obtaining a multi-source data set.

[0081] By aligning timestamps, for example, with second-level accuracy (e.g., 2025-04-25 10:00:00), we can merge multi-source data into a unified dataset, generating a multi-source dataset containing data for temperature 25°C, vibration 0.5 mm / s, and voltage 220 V. This approach ensures consistent data formatting, facilitating subsequent fusion and analysis.

[0082] Furthermore, the method for visual management of energy consumption fluctuations in a smart mold factory provided in this embodiment includes step S200:

[0083] Step S210: parse the multi-source data using a pre-established protocol conversion module to generate parsed data.

[0084] In industrial production, acquiring raw data from multiple source devices and parsing, verifying, and standardizing it are the keys to achieving equipment monitoring and data analysis.

[0085] The following analysis and examples focus on technical topics such as protocol conversion, data verification, and data standardization, focusing on industrial production scenarios to ensure that the content is logically rigorous and mutually supportive.

[0086] For example, the implementation of the protocol conversion module depends on the diversity of device communication protocols. In industrial scenarios, devices may output data using Modbus, Profibus, or custom protocols.

[0087] In one possible implementation, a pre-established protocol conversion module uses a parser to unify raw data from different protocols into an intermediate format. For example, a temperature sensor outputs register values ​​using the Modbus protocol, which the module parses into key-value pairs, such as "Temperature: 26°C." Meanwhile, a current monitoring device using the Profibus protocol extracts the current field from the message and generates "Current: 10A." This approach ensures that raw data from different protocols can be parsed uniformly, facilitating subsequent processing.

[0088] Step S220: Use a data verification tool to check the integrity and accuracy of the parsed data. If there are any missing or errors in the parsed data, fix them using preset completion rules to obtain verified data.

[0089] It should be noted that data validation tools are crucial in ensuring data integrity and accuracy.

[0090] Specifically, validation tools identify problems by checking data packets for missing fields or unusual values.

[0091] In one embodiment, if a temperature sensor data packet lacks a timestamp, the verification tool will mark it as incomplete; if the current value is outside a reasonable range, such as "current: 1000A", it will be determined as an error.

[0092] For missing data, pre-set completion rules might use the mean of historical data to fill in missing values, for example, completing a missing temperature value with 25°C from the previous minute. For erroneous data, outliers are removed using the device calibration range. This verification mechanism ensures data reliability.

[0093] Step S230: Based on the verified data, a data conversion tool is used to map the data into a predefined standard data structure, and standardized data is generated by field alignment to determine the standardized data.

[0094] Preferably, the data conversion tool maps the verified data into a predefined standard data structure to further improve data consistency. For example, the standard data structure may require that all data contain device ID, timestamp, and measurement value fields. In one embodiment, the verified temperature data "temperature: 26°C" and the current data "current: 10A" are generated into a unified format through field mapping, such as {"device ID": "T001", "timestamp": "2025-04-2510:00:01", "temperature": 26°C} and {"device ID": "C001", "timestamp": "2025-04-2510:00:01", "current": 10A}. Through field alignment, standardized data ensures that all field formats are consistent, which is convenient for storage and analysis. It can be understood that each link of the above process supports each other.

[0095] The protocol conversion module addresses the heterogeneity of device communications, providing a unified data foundation for verification. Data verification tools improve data reliability through completion and error correction, providing high-quality input for standardization. Standardized data, through a unified structure, facilitates subsequent data fusion and analysis. For example, standardized temperature and current data can be directly used for device status monitoring or production optimization. This layered logic ensures the integrity of the data from raw to standardized.

[0096] In one possible implementation, the protocol conversion module can support dynamic protocol adaptation to accommodate expansion scenarios. For example, if a new device uses the MQTT protocol, the module automatically loads parsing rules from the configuration file and generates an intermediate format consistent with the existing data. This extensibility ensures the system's adaptability.

[0097] Similarly, verification tools can integrate machine learning algorithms to dynamically learn the normal range of device data, further improving the accuracy of anomaly detection. These extensions enrich the applicability of core processes. For example, in a specific production line application, the protocol conversion module processes 100,000 sensor data points daily. The verification tool removes approximately 0.5% of abnormal data, and the standardized data supports real-time monitoring dashboard updates. This multi-step collaborative approach significantly improves the efficiency and reliability of data processing, providing a solid foundation for intelligent industrial production.

[0098] Furthermore, the method for visual management of energy consumption fluctuations in a smart mold factory provided in this embodiment includes step S300:

[0099] Step S310: Obtain the timestamp and energy consumption value records in the standardized data, and use an outlier detection tool to determine whether the energy consumption value exceeds a preset threshold. If it exceeds, mark it as noise data to obtain a marked data set.

[0100] For example, in industrial production scenarios, obtaining timestamps and energy consumption values ​​from standardized data is essential for real-time monitoring of equipment operating status. Standardized data typically includes fields such as device identification, timestamp, and energy consumption value. For example, data for a compressor might be recorded as {"Device ID":"M001","Timestamp":"2025-04-2508:00:00","Energy Consumption":500kWh}. The timestamp records the moment the data was generated, and the energy consumption value reflects the power consumption of the equipment. When obtaining these fields, ensure that the data format is consistent, for example, timestamps must be in the "YYYY-MM-DDHH:MM:SS" format to avoid parsing errors.

[0101] In one possible implementation, the outlier detection tool uses a preset threshold to determine whether energy consumption values ​​are reasonable. For example, suppose the normal energy consumption of compressors on a production line ranges from 400 to 600 kWh. The detection tool scans each record. If a record's energy consumption value is 800 kWh, exceeding the threshold, it is marked as noise data, generating a labeled data set such as {"Device ID":"M001","Timestamp":"2025-04-25 08:01:00","Energy Consumption":800 kWh","Flag":"Noise"}. This labeling facilitates subsequent processing of abnormal data.

[0102] Step S320: Based on the marked data set, the missing energy consumption values ​​are filled using data completion rules, and the time field is calibrated using a timestamp alignment tool to generate a completed data set.

[0103] It should be noted that the data completion rule is used to fill in the missing energy consumption values.

[0104] Specifically, if the energy consumption value corresponding to a timestamp "2025-04-2508:02:00" is missing, the energy consumption value of the previous moment can be used to fill it in, for example, using 500kWh at "08:01:00" to fill it in.

[0105] The timestamp alignment tool further calibrates the time fields, ensuring that all record timestamps are evenly spaced, for example, one record per minute. If a timestamp deviation is found, such as "08:02:30," it is corrected to "08:03:00." The completed data set maintains data continuity.

[0106] Step S330: For the completed data set, a deduplication algorithm is used to merge duplicate records, and the data is merged according to the timestamp and device identifier to determine the cleaned data set.

[0107] Preferably, a deduplication algorithm processes the completed data set, merging duplicate records. In one embodiment, if two records appear at "2025-04-25 08:00:00," namely {"Device ID": "M001," "Energy Consumption": 500kWh} and {"Energy Consumption": 510kWh}, the deduplication algorithm takes the average value, generating {"Energy Consumption": 505kWh}. The data is then merged by timestamp and device ID, ensuring that each device has only one record at each time point, forming the cleaned data set.

[0108] Step S340: Convert the cleaned data set into a predefined structure using a field mapping tool, merge the energy consumption values ​​using data aggregation rules, and generate a real-time energy consumption data set.

[0109] For example, the field mapping tool converts cleaned data into a predefined structure. Assuming the standard structure requires the inclusion of "Device ID," "Time," and "Energy Consumption Value" fields, the cleaned record {"Device ID":"M001,""Timestamp":"2025-04-25 08:00:00,""Energy Consumption":505kWh} is mapped to a consistent format. Data aggregation rules further consolidate energy consumption values, for example, summarizing the total energy consumption of each device by hour, generating a real-time energy consumption dataset such as {"Device ID":"M001,""Time":"2025-04-25 08:00:00,""Total Energy Consumption":3000kWh}.

[0110] As you can see, each of these steps supports each other. Outlier detection ensures data reliability, completion and alignment provide continuity, deduplication and merging improve data simplicity, and mapping and aggregation generate datasets that are easy to analyze. This logically rigorous process supports energy consumption monitoring needs in industrial production.

[0111] In one possible expansion, the outlier detection tool could incorporate trend analysis, combining historical data to determine whether a sudden change in energy consumption is reasonable. For example, if energy consumption suddenly increases from 500kWh to 700kWh, but a similar trend existed in the previous few days, it would not be marked as noise. This dynamic detection improves the accuracy of anomaly detection and adapts to complex production scenarios.

[0112] Furthermore, the method for visual management of energy consumption fluctuations in a smart mold factory provided in this embodiment includes step S400:

[0113] Step S410: Obtain a time series from the real-time energy consumption dataset, and segment the time series using a data granularity adjustment rule to obtain an aligned time series set.

[0114] For example, in industrial production scenarios, obtaining time series from real-time energy consumption datasets is the basis for analyzing equipment operating modes. Real-time energy consumption datasets typically contain fields such as device identification, timestamp, and energy consumption value. For example, a compressor record is {device ID: M001, timestamp: 2025-04-25 08:00:00, energy consumption: 500kWh}. A time series consists of a timestamp and energy consumption value, and the device identification is used to distinguish different devices. When obtaining this data, it is necessary to ensure that the timestamp format is unified, such as YYYY-MM-DDHH:MM:SS, to avoid errors caused by inconsistent formats during subsequent processing. This unified format facilitates continuity analysis of time series.

[0115] In one possible implementation, a timestamp alignment tool is used to calibrate the time field to ensure the regularity of the time series.

[0116] Specifically, if timestamps in a dataset deviate, such as a record with the time 2025-04-25 08:02:30, the timestamp alignment tool will calibrate it to 2025-04-25 08:03:00 to ensure uniform intervals of one record per minute. For example, in data records from a compressor, the timestamps may be irregular from 08:01:00 to 08:02:30. After alignment, the timestamps will be aligned to generate a sequence of 08:01:00, 08:02:00, and 08:03:00. This alignment provides a standardized time basis for subsequent data segmentation and feature extraction.

[0117] It should be noted that the data granularity adjustment rules are used to segment the time series and generate segments suitable for analysis.

[0118] Data granularity adjustment changes the time interval according to analysis requirements, such as aggregating minute-level data to hour-level data.

[0119] To analyze a device's energy consumption trends over a day, preferably, minute-by-minute records can be aggregated into one hourly record. For example, 60 records for device M001 from 08:00:00 to 08:59:00, with original energy consumption values ​​of 500kWh, 510kWh, and so on, can be aggregated to create a single record: {Device ID: M001, Time: 2025-04-25 08:00:00, Total Energy Consumption: 30500kWh}. This aggregation facilitates observing energy consumption changes over a longer period.

[0120] Step S420: Calculate trend characteristics using a moving average algorithm for the aligned time series set.

[0121] Specifically, for the aligned time series set, the moving average algorithm is used to calculate trend characteristics.

[0122] Moving average highlights long-term trends by smoothing data. For example, a 5-minute sliding window is set and the average energy consumption value within the window is calculated.

[0123] In one embodiment, the energy consumption sequence of the M001 device is 500 kWh, 510 kWh, 520 kWh, 530 kWh, 540 kWh, and the 5-minute moving average is 520 kWh.

[0124] This smoothing process helps identify rising or falling trends in energy consumption and facilitates prediction of equipment operating status.

[0125] Step S430: For the aligned time series set, use Fourier transform tools to extract periodic features.

[0126] For example, the Fourier transform tool is used to extract the periodic characteristics of time series. The Fourier transform decomposes the time series into sine waves of different frequencies, revealing the periodic pattern of energy consumption.

[0127] In one possible implementation, analyzing the daily energy consumption data for equipment M001 reveals peaks every four hours, such as 550 kWh and 560 kWh at 8:00:00 and 12:00:00, respectively. A Fourier transform identifies this four-hour cycle and generates periodic signature data. This periodic information can be used to optimize equipment scheduling.

[0128] It can be understood that the above steps form a logically rigorous process.

[0129] Timestamp alignment ensures data consistency, granularity is adjusted to suit analysis needs, moving averages reveal trends, and Fourier transforms mine periodic features. These features complement each other, providing comprehensive data support for energy consumption monitoring. For example, trend features can predict future energy consumption, while periodic features can optimize production plans. This multi-dimensional analysis enhances data-driven decision-making in industrial scenarios.

[0130] Furthermore, the method for visual management of energy consumption fluctuations in a smart mold factory provided in this embodiment includes step S500:

[0131] Step S510: Obtain periodic features, trend features, and production parameters from the data set, and use a data cleaning tool to remove outliers to obtain a cleaned feature data set.

[0132] For example, in an industrial production scenario, periodic characteristics, trend characteristics, and production parameters are obtained from a real-time energy consumption dataset, and data quality must be ensured.

[0133] Data cleaning tools are used to remove outliers. For example, a compressor dataset typically records energy consumption values ​​between 400 and 600 kWh, but one record shows 2000 kWh, significantly deviating from the normal range. The cleaning tool uses a set threshold, such as three times the standard deviation, to remove these outliers and generate a cleaned feature dataset. This cleaning ensures the accuracy of subsequent modeling.

[0134] In one possible implementation, periodic features are extracted by analyzing the cyclical patterns in energy consumption data. For example, the energy consumption data for a particular compressor peaks at 08:00:00 and 16:00:00 daily, reaching 580 kWh and 590 kWh, respectively. Using a time series analysis tool, this 8-hour periodic pattern is identified and a periodic feature is generated. This feature reflects the regularity of equipment operation and facilitates subsequent forecasting.

[0135] It should be noted that trend features are extracted through long-term energy consumption changes. For example, the energy consumption of a device gradually increases from 500kWh on Monday to 550kWh on Friday within a week.

[0136] Use smoothing tools, such as exponential smoothing, to extract upward trends and generate trend-like features that help capture long-term changes in equipment operation.

[0137] Specifically, production parameters include temperature, pressure, etc. For example, when the compressor is running, the temperature is 25°C and the pressure is 5 bar. These parameters are closely related to energy consumption.

[0138] The cleaning tool ensures that parameter values ​​are reasonable, such as eliminating abnormal records with a temperature of -10°C, to generate a reliable production parameter data set.

[0139] Step S520: For the cleaned feature data set, a support vector machine algorithm is used to perform modeling, and hyperparameters are adjusted through a grid search tool to obtain an initial energy consumption fluctuation prediction model.

[0140] Preferably, a support vector machine algorithm is used for modeling.

[0141] The cleaned feature dataset is fed into the model, and the grid search tool adjusts hyperparameters, such as the kernel function and penalty coefficient. For example, linear and radial basis kernels are tested to find the optimal combination and generate an initial energy consumption fluctuation prediction model. This adjustment improves the model's generalization capabilities.

[0142] Step S530: If the error of the initial energy consumption fluctuation prediction model exceeds a preset threshold, a validation data set is used for evaluation, and the model parameters are optimized by a cross-validation tool to obtain an optimized energy consumption fluctuation prediction model.

[0143] In one embodiment, if the initial model prediction error exceeds a preset threshold, such as a mean square error greater than 10%, a validation dataset is used for evaluation.

[0144] The cross-validation tool divides the data into five parts, alternating between training and testing to optimize model parameters. For example, adjusting the regularization parameters reduced the error to 5%, resulting in an optimized energy consumption fluctuation prediction model. This optimization ensures the reliability of the predictions.

[0145] Step S540: According to the optimized energy consumption fluctuation prediction model, real-time production parameters and feature data sets are obtained, and prediction values ​​are calculated by a batch processing tool to obtain energy consumption fluctuation prediction results.

[0146] As you can see, the optimized model combines real-time production parameters and feature data sets, and calculates predictions through batch processing tools. For example, if we input the temperature of 26°C, pressure of 5.2 bar, and historical feature data at 08:00:00, the model predicts energy consumption of 585 kWh.

[0147] Batch processing tools predict a day's worth of data and generate energy consumption fluctuation forecasts. This forecast supports production scheduling optimization. For example, the above process, from data cleaning to model optimization, forms a complete chain.

[0148] Cleaning ensures data quality, feature extraction provides multi-dimensional input, model training and optimization improve prediction accuracy, and batch processing enables real-time application. Each link supports each other and jointly improves the efficiency of energy management.

[0149] Furthermore, the method for visual management of energy consumption fluctuations in a smart mold factory provided in this embodiment includes step S600:

[0150] Step S610: If the energy consumption fluctuation prediction result output by the energy consumption fluctuation prediction model exceeds a preset threshold, the energy consumption fluctuation prediction result is analyzed by an anomaly detection algorithm to locate high energy consumption processes and obtain a process identification list.

[0151] For example, in an industrial production scenario, after the energy consumption fluctuation prediction model outputs the prediction result, if the predicted value exceeds the preset threshold, such as energy consumption exceeds 600kWh, the result needs to be analyzed by an anomaly detection algorithm to locate the high-energy-consuming process.

[0152] The anomaly detection algorithm can use the isolation forest algorithm, which works by randomly segmenting data features to identify anomalies.

[0153] Isolation Forest breaks down energy consumption data into multiple feature dimensions, such as time and process parameters, to quickly isolate outliers. For example, the predicted energy consumption of a compressor process at 08:00:00 is 650kWh, exceeding the threshold. After algorithmic analysis, the process is identified as high energy consuming and labeled "Compressor A - Process 1."

[0154] In one possible implementation, anomaly detection not only focuses on energy consumption values ​​but also incorporates production parameters. For example, when a compressor is running, the temperature is 28°C and the pressure is 5.5 bar, both within the normal range, but the energy consumption is abnormal.

[0155] By comparing historical data, the algorithm discovered that energy consumption under similar parameters is typically 580 kWh, confirming that the anomaly stems from the process itself, rather than external parameters. This multi-dimensional analysis improves location accuracy and ensures targeted follow-up actions.

[0156] It's important to note that after locating high-energy-consuming processes, a process ID list must be generated. This list records detailed information about the abnormal process, such as process name, timestamp, and energy consumption. For example, the process ID list might include "Compressor A - Process 1, 08:00:00, 650 kWh" and "Compressor B - Process 2, 09:00:00, 620 kWh." The process ID list is stored in a structured format for easy access and analysis. This approach ensures clarity and traceability of abnormality information.

[0157] Step S620: Based on the process identification list, an alarm trigger tool is used to generate an alarm message and process identification. The alarm message and process identification are saved through a logging tool to obtain an exception handling record. Specifically, the alarm trigger tool generates an alarm message based on the process identification list. The tool can be configured with multiple alarm methods, such as SMS or system pop-up windows. For example, when an abnormality is detected in "Compressor A - Process 1", the tool generates an alarm message "08:00:00 Compressor A - Process 1 energy consumption abnormality, predicted value 650kWh" and attaches the process identification to the management personnel. This real-time notification facilitates quick response and reduces the duration of the abnormality.

[0158] Preferably, alarm information and process identifiers are saved in a logging tool to form a record of exception handling. The logging tool stores data in a time series format, including the alarm time, process identifier, energy consumption value, and handling status. For example, a log entry might read "08:00:00, Compressor A - Process 1, 650kWh, pending." Management personnel can then mark it as "Equipment parameters adjusted" to create a complete record. This record supports post-analysis and production process optimization.

[0159] In one embodiment, the exception handling records can be further used for tracing and improvement. For example, if the log analysis reveals that "Compressor A - Process 1" frequently fails at 08:00:00 every Monday, it may be related to the equipment startup parameters.

[0160] Managers adjust startup settings accordingly to reduce the incidence of abnormal energy consumption. This closed-loop mechanism from recording to optimization improves production efficiency.

[0161] It's easy to understand that the above process, from anomaly detection to logging, forms a complete chain. Anomaly detection accurately locates the problem, alarm triggering ensures a timely response, and logging provides data support for subsequent optimization. Each link supports each other, collectively improving the response speed and decision-making quality of energy management.

[0162] The present invention relates to a visual management system for energy consumption fluctuations in an intelligent mold factory, which is used to implement the above-mentioned visual management method for energy consumption fluctuations in an intelligent mold factory. The visual management system for energy consumption fluctuations in an intelligent mold factory includes a first acquisition module, a second acquisition module, a generation module, an extraction module, a third acquisition module and a judgment module, wherein the first acquisition module is used to acquire multi-source data from production equipment, and the multi-source data includes sensor data, power monitoring data and production management data; the second acquisition module is used to parse the multi-source data using a pre-established protocol conversion module to obtain standardized data; the generation module is used to execute a data cleaning algorithm on the standardized data to generate a real-time energy consumption data set; the extraction module is used to extract periodic features and trend features based on the real-time energy consumption data set using a time series decomposition method; the third acquisition module is used to model the periodic features, trend features and production parameters using a support vector machine algorithm to obtain an energy consumption fluctuation prediction model; the judgment module is used to trigger an anomaly detection algorithm to locate a high-energy-consuming process if the output value of the energy consumption fluctuation prediction model exceeds a preset threshold.

[0163] Furthermore, the present embodiment provides a visualization management system for energy consumption fluctuations in intelligent mold factories, wherein the first acquisition module includes a first acquisition unit and a second acquisition unit, wherein the first acquisition unit is used to acquire multi-source data from sensor equipment, power monitoring equipment, and production management systems; the second acquisition unit is used to format the multi-source data using a data acquisition protocol, and generate a multi-source data set with a consistent format through timestamp alignment to obtain a multi-source data set.

[0164] Preferably, in the energy consumption fluctuation visualization management system for intelligent mold factories provided in this embodiment, the second acquisition module includes a third acquisition unit, a fourth acquisition unit and a determination unit, wherein the third acquisition unit is used to parse multi-source data using a pre-established protocol conversion module to generate parsed data and obtain parsed data; the fourth acquisition unit is used to check the integrity and accuracy of the parsed data using a data verification tool, and if the parsed data is missing or erroneous, it is repaired by preset completion rules to obtain verified data; the determination unit is used to map the verified data into a predefined standard data structure using a data conversion tool, generate standardized data through field alignment, and determine the standardized data.

[0165] Compared with the existing technology, the method and system for visual management of energy consumption fluctuations in intelligent mold factories provided in this embodiment parse and standardize the multi-source data of production equipment through a protocol conversion module, and use a data cleaning algorithm to generate a real-time energy consumption data set. Then, the time series decomposition method is used to extract periodic and trend features, and the support vector machine algorithm is combined to establish an energy consumption fluctuation prediction model. When the output value of the prediction model exceeds the preset threshold, the anomaly detection algorithm is triggered to locate high-energy-consuming processes. This embodiment realizes real-time analysis, prediction and anomaly detection of energy consumption data in industrial production processes, which helps to promptly detect energy consumption anomalies and take corresponding measures to improve production efficiency and energy conservation and emission reduction effects.

[0166] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they are aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the invention. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the invention. Thus, the present invention is intended to include such changes and modifications as fall within the scope of the claims and their equivalents.

Claims

1. A method for visual management of energy consumption fluctuations in intelligent mold factories, characterized in that: The following steps are involved: Acquire multi-source data from production equipment, wherein the multi-source data includes sensor data, power monitoring data, and production management data; Parsing the multi-source data using a pre-established protocol conversion module to obtain standardized data; executing a data cleaning algorithm on the standardized data to generate a real-time energy consumption data set; According to the real-time energy consumption data set, a time series decomposition method is used to extract periodic features and trend features; A support vector machine algorithm is used to model the periodic characteristics, trend characteristics and production parameters to obtain an energy consumption fluctuation prediction model; It is determined whether the output value of the energy consumption fluctuation prediction model exceeds a preset threshold. If the output value of the energy consumption fluctuation prediction model exceeds the preset threshold, an abnormality detection algorithm is triggered to locate a high energy consumption process.

2. The method for visual management of energy consumption fluctuations in an intelligent mold factory according to claim 1, characterized in that: The step of acquiring multi-source data from production equipment includes: Acquire multi-source data from sensor devices, power monitoring equipment, and production management systems; The multi-source data is formatted using a data acquisition protocol, and a multi-source data set with a consistent format is generated by aligning timestamps to obtain the multi-source data set.

3. The method for visual management of energy consumption fluctuations in an intelligent mold factory according to claim 1, characterized in that: The step of using a pre-established protocol conversion module to parse the multi-source data to obtain standardized data includes: Parsing the multi-source data using a pre-established protocol conversion module to generate parsed data to obtain the parsed data; The parsed data is checked for completeness and accuracy using a data verification tool. If the parsed data is missing or erroneous, it is repaired using preset completion rules to obtain verified data. According to the verified data, a data conversion tool is used to map the data into a predefined standard data structure, and standardized data is generated by field alignment to determine the standardized data.

4. The method for visual management of energy consumption fluctuations in an intelligent mold factory according to claim 1, characterized in that: The steps of executing a data cleaning algorithm on the standardized data to generate a real-time energy consumption data set include: Obtaining the timestamp and energy consumption value records in the standardized data, using an outlier detection tool to determine whether the energy consumption value exceeds a preset threshold, and if so, marking it as noise data to obtain a marked data set; According to the marked data set, the missing energy consumption values ​​are filled using a data completion rule, and the time field is calibrated using a timestamp alignment tool to generate a completed data set; For the completed data set, a deduplication algorithm is used to merge duplicate records, and the data is merged according to timestamps and device identifiers to determine a cleaned data set; The cleaned data set is converted into a predefined structure through a field mapping tool, and energy consumption values ​​are merged using data aggregation rules to generate the real-time energy consumption data set.

5. The method for visual management of energy consumption fluctuations in an intelligent mold factory according to claim 1, characterized in that: According to the real-time energy consumption data set, the steps of extracting periodic features and trend features using a time series decomposition method include: Acquire a time series from the real-time energy consumption dataset, and segment the time series using a data granularity adjustment rule to obtain an aligned time series set; For the aligned time series set, a moving average algorithm is used to calculate trend characteristics; For the aligned time series set, a Fourier transform tool is used to extract periodic features.

6. The method for visual management of energy consumption fluctuations in an intelligent mold factory according to claim 1, characterized in that: The step of using the support vector machine algorithm to model the periodic characteristics, trend characteristics and production parameters to obtain an energy consumption fluctuation prediction model includes: Obtain periodic characteristics, trend characteristics and production parameters from the data set, use data cleaning tools to remove outliers, and obtain a cleaned feature data set; For the cleaned feature data set, a support vector machine algorithm is used to build a model, and hyperparameters are adjusted through a grid search tool to obtain an initial energy consumption fluctuation prediction model; If the error of the initial energy consumption fluctuation prediction model exceeds a preset threshold, a validation data set is used for evaluation, and the model parameters are optimized by a cross-validation tool to obtain an optimized energy consumption fluctuation prediction model; According to the optimized energy consumption fluctuation prediction model, real-time production parameters and the characteristic data set are obtained, and prediction values ​​are calculated through a batch processing tool to obtain energy consumption fluctuation prediction results.

7. The method for visual management of energy consumption fluctuations in an intelligent mold factory according to claim 1, characterized in that: The step of determining whether the output value of the energy consumption fluctuation prediction model exceeds a preset threshold, and triggering an anomaly detection algorithm to locate a high-energy-consuming process if the output value of the energy consumption fluctuation prediction model exceeds the preset threshold, includes: If the energy consumption fluctuation prediction result output by the energy consumption fluctuation prediction model exceeds a preset threshold, the energy consumption fluctuation prediction result is analyzed by an anomaly detection algorithm to locate high energy consumption processes and obtain a process identification list; According to the process identification list, an alarm trigger tool is used to generate alarm information and process identification, and the alarm information and process identification are saved through a log recording tool to obtain an exception handling record.

8. A system for visualizing energy consumption fluctuation management in a smart mold factory, for implementing the method for visualizing energy consumption fluctuation management in a smart mold factory as claimed in any one of claims 1 to 7, characterized in that: The energy consumption fluctuation visualization management system for the intelligent mold factory includes: A first acquisition module is used to acquire multi-source data from production equipment, wherein the multi-source data includes sensor data, power monitoring data and production management data; A second acquisition module is used to parse the multi-source data using a pre-established protocol conversion module to obtain standardized data; A generation module, configured to execute a data cleaning algorithm on the standardized data to generate a real-time energy consumption data set; An extraction module, configured to extract periodic features and trend features based on the real-time energy consumption dataset using a time series decomposition method; A third acquisition module is used to model the periodic characteristics, trend characteristics and production parameters using a support vector machine algorithm to obtain an energy consumption fluctuation prediction model; The judgment module is used to trigger an anomaly detection algorithm to locate a high-energy-consuming process if the output value of the energy consumption fluctuation prediction model exceeds a preset threshold.

9. The energy consumption fluctuation visualization management system for an intelligent mold factory according to claim 8, characterized in that: The first acquisition module includes: A first acquisition unit is used to acquire multi-source data from sensor equipment, power monitoring equipment and production management system; The second acquisition unit is configured to format the multi-source data using a data acquisition protocol, generate a multi-source data set with a consistent format by aligning timestamps, and obtain the multi-source data set.

10. The energy consumption fluctuation visualization management system for an intelligent mold factory according to claim 8, characterized in that: The second acquisition module includes: a third acquiring unit, configured to parse the multi-source data using a pre-established protocol conversion module to generate parsed data; a fourth acquiring unit, configured to use a data verification tool to check the integrity and accuracy of the parsed data, and to repair any missing or erroneous data using a preset completion rule to obtain verified data; The determination unit is configured to map the verified data into a predefined standard data structure using a data conversion tool, generate standardized data through field alignment, and determine the standardized data.

Images