Color coating production line energy consumption monitoring and optimizing system based on multi-source data fusion

Abstract

The invention discloses a color coating production line energy consumption monitoring and optimizing system based on multi-source data fusion, and particularly relates to the technical field of production energy consumption management, which comprises a sensing layer, a data layer, an analysis layer and an application layer, the sensing layer collects sub-item energy consumption, a production process, an equipment state and environment data through various sensors; the data layer receives and stores multi-source data to form a database; the analysis layer constructs a reference energy consumption model based on historical data, calculates deviation between actual energy consumption and predicted energy consumption through real-time rolling comparison, and starts an analysis program and updates the model regularly when the deviation exceeds a threshold value; and the application layer realizes energy consumption monitoring, energy efficiency abnormity warning, optimization strategy recommendation and carbon emission accounting based on the analysis result. Through multi-source data fusion and dynamic model correction, accurate monitoring and intelligent optimization of energy consumption of the color coating production line are realized, energy efficiency is effectively improved, carbon emission is reduced, and data support is provided for green production.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy consumption management, and specifically discloses a color coating production line energy consumption monitoring and optimization system based on multi-source data fusion. BACKGROUND

[0002] In the development process of the color coating production industry, the requirement for fine energy consumption management is increasing; the color coating production line includes multiple high-energy consumption links such as pretreatment, roller coating, and curing, and the energy consumption of electric energy and fuel gas accounts for more than 30% of the production cost, so the demand for energy consumption monitoring and optimization by enterprises is increasingly urgent. The traditional scheme adopts manual meter reading statistics, single-machine energy consumption recorders, and simple threshold alarm methods; among them, manual meter reading records the data of each device electric meter and flowmeter daily, and forms an energy consumption report by summarizing; the single-machine recorder installs independent metering devices on key equipment to display the energy consumption of a single device in real time; and the simple threshold alarm system presets a fixed upper limit of energy consumption, and triggers an alarm when it is exceeded, thereby realizing basic energy consumption management.

[0003] However, the traditional energy consumption management method has many problems: on the one hand, the data collection is not comprehensive, manual meter reading has a lag and is prone to errors, and single-machine recording is difficult to realize multi-device data linkage analysis, resulting in prominent energy consumption "black box" phenomenon; on the other hand, the energy consumption evaluation lacks dynamic model support, and the fixed threshold does not consider the influence of production processes such as substrate thickness, production line speed, and environment such as temperature and humidity changes, so false positives or false negatives often occur, for example, the energy consumption of the same device in a high-temperature environment in summer should be higher than in winter, but the fixed threshold cannot be adjusted adaptively; in addition, the optimization strategy relies on experience and lacks data-driven accurate recommendations, such as setting the temperature of the curing furnace according to operating habits rather than the most energy-efficient value, resulting in serious energy waste; at the same time, carbon emission accounting relies on monthly summary data and cannot reflect the dynamic carbon emission in the production process in real time, making it difficult to meet the real-time monitoring requirements of green factories. Therefore, it is very important to develop a color coating production line energy consumption monitoring and optimization system that can realize multi-source data fusion, dynamic model correction, and intelligent strategy recommendation. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a color coating production line energy consumption monitoring and optimization system based on multi-source data fusion.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a color coating production line energy consumption monitoring and optimization system based on multi-source data fusion, comprising: a perception layer comprising a plurality of sensors arranged on key devices of the color coating production line, for collecting multi-source data including sub-energy consumption data, production process data, device state data, and environmental data; a data layer in communication connection with the perception layer, for receiving and storing the multi-source data collected by the perception layer to form a corresponding database; The analysis layer is communicatively connected with the data layer, and is configured to construct a benchmark energy consumption model based on historical data, to perform a rolling comparison between actual energy consumption and predicted energy consumption of the benchmark energy consumption model and to calculate a deviation in real-time monitoring, and to start an analysis program when the deviation exceeds a threshold value, and to update the benchmark energy consumption model based on new data at regular intervals; The application layer is communicatively connected with the analysis layer, and is configured to implement energy consumption monitoring, energy efficiency anomaly alarm, optimization strategy recommendation and carbon emission accounting based on output results of the analysis layer.

[0006] Preferably, the sub-energy consumption data includes electric energy consumption data, gas consumption data, compressed air and water consumption data; the production process data includes product information, process parameters and production plans; the equipment state data includes equipment start-stop signals, standby time and load rate; and the environmental data includes environmental temperature and humidity.

[0007] Preferably, the electric energy consumption data is collected by an intelligent electric meter installed on a key equipment circuit; the gas consumption data is collected by a flow meter installed on a gas main pipe of a curing furnace; and the compressed air and water consumption data is collected by flow meters installed on corresponding pipelines, respectively.

[0008] Preferably, the product information includes substrate material, thickness, width and color; the process parameters include production line speed, curing furnace temperature, roller pressure and fan speed.

[0009] Preferably, the benchmark energy consumption model adopts a machine learning algorithm, and is expressed as a function: , wherein E is predicted energy consumption, P is a production process data vector, Env is an environmental data vector, F is a nonlinear mapping function, and the model obtains a parameter set through training and learning.

[0010] Preferably, the analysis layer selects historical data of a specific time period as a training set to construct an initial benchmark energy consumption model, and obtains an initial model parameter set .

[0011] Preferably, the rolling comparison includes: inputting the production process data vector P and the environmental data vector Env of a monitoring time period into the benchmark energy consumption model to obtain predicted energy consumption , collecting actual energy consumption of the monitoring time period, and calculating a deviation between the predicted energy consumption and the actual energy consumption, with a calculation formula being: .

[0012] Preferably, the operation of the analysis layer includes first-level correction and second-level correction: The first-level correction is that when the deviation value When the set threshold is exceeded, the key variables causing the deviation are analyzed and optimization suggestions are generated; The secondary correction is to periodically add the newly added production data to the model training set, and minimize the loss function through the optimization algorithm: to re-solve the model parameter set , and obtain the updated benchmark energy consumption model.

[0013] Preferably, in the primary correction, the method of analyzing the input variables includes calculating the partial derivative of the benchmark energy consumption model F with respect to each input variable , wherein is the i-th variable in the vector , and identifying the variable with the large absolute value of the partial derivative as the main influencing factor.

[0014] Preferably, the optimization strategy recommendation includes scheduling optimization, parameter optimization, and maintenance reminder; the scheduling optimization is to suggest concentrating orders of the same color and the same thickness; the parameter optimization is to recommend adjusting the process parameters in the production process data vector P under the premise of ensuring quality; and the maintenance reminder is to prompt equipment maintenance when the equipment state data indicates performance degradation.

[0015] Technical effects and advantages of the present application: 1. The present application realizes full-dimensional data collection from "total electricity bill" to "sub-equipment, sub-product, and sub-process" through the multi-sensor deployment of the perception layer, constructs a fine energy consumption management system, accurately depicts the energy consumption portrait, and makes the energy consumption data transparent and traceable; 2. The present application overcomes the limitations of traditional static energy consumption benchmarks through the benchmark energy consumption model of the analysis layer, establishes an "intelligent benchmark" that dynamically adjusts with production process data and environmental data, makes energy efficiency evaluation more in line with actual production conditions, and improves the scientificity and fairness of evaluation; 3. The present application can automatically compare the deviation between actual energy consumption and predicted energy consumption through the deviation analysis program of the analysis layer, locate the key influencing factors of energy efficiency anomalies through partial derivative calculation, and refine the problem from "high energy consumption" to specific equipment and working conditions, significantly enhancing the guidance of diagnosis; 4. The present application enables the system to continuously absorb new data to optimize the parameter set through the model periodic updating mechanism of the analysis layer, adapts to changes such as equipment aging and process adjustment in long-term production line operation, avoids benchmark model failure, and ensures the long-term accuracy of monitoring and optimization suggestions; 5. The present application provides data support for production scheduling, process adjustment, equipment maintenance, and other decisions through the optimization strategy recommendation function, promotes the transformation from the "experience-based energy saving" mode relying on experience to the "intelligent energy saving" mode based on data, and improves the accuracy and effectiveness of energy saving measures. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to make the technical solutions of the embodiments of the present application clearer, the drawings needed for the description of the embodiments will be briefly introduced as follows. Obviously, the drawings described below are only some of the embodiments of the present application, and for those skilled in the basic field, other drawings can also be obtained according to these drawings without creative labor.

[0017] Figure 1 It is a system overall architecture diagram.

[0018] Figure 2 It is an analysis layer workflow diagram.

[0019] Figure 3 It is an application layer function schematic diagram. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0021] Referring to Figure 1 As shown in the figure, the system overall architecture adopts a layered design and is divided into four layers. The perception layer is the basis, and various sensors deployed on key equipment in the production line collect sub-item energy consumption, production process, equipment state and environmental data. The data layer is responsible for receiving and storing the multi-source heterogeneous data uploaded by the perception layer, and constructs a time series database supporting the upper layer application. The analysis layer is the core, and a dynamically updated baseline energy consumption model is constructed based on historical data. The deviation between the predicted value and the actual value is compared in real time to trigger analysis and update the model parameters regularly. The application layer is the top layer, directly facing the user, and provides four functions of energy consumption monitoring, abnormal alarm, optimization strategy recommendation and carbon emission accounting based on the output of the analysis layer, realizing the final value of the system.

[0022] The perception layer includes various sensors arranged on key equipment in the color coating production line, which are used to collect multi-source data, including sub-item energy consumption data, production process data, equipment state data and environmental data. The sub-item energy consumption data includes electric energy consumption data, gas consumption data, compressed air and water consumption data; the production process data includes product information, process parameters and production plan; the equipment state data includes equipment start-stop signal, standby time and load rate; and the environmental data includes environmental temperature and humidity.

[0023] The electric energy consumption data is collected by the smart electric meter installed on the key equipment circuit; the gas consumption data is collected by the flow meter installed on the curing furnace gas main pipe; the compressed air and water consumption data is collected by the flow meter installed on the corresponding pipeline.

[0024] The product information includes the substrate material, thickness, width and color; the process parameters include the production line speed, curing furnace temperature, coating roller pressure and fan speed.

[0025] It needs to be specifically explained in this embodiment that the perception layer is composed of multiple sensors installed on each key equipment of the color coating production line, including but not limited to: smart electric meters installed on the key equipment circuit of the pretreatment equipment, roller coater, curing furnace, fan, water pump and air compressor, used to collect electric energy consumption data, including active power, reactive power and electric quantity; flow meters installed on the curing furnace gas main pipe, used to collect gas consumption data, including instantaneous flow and cumulative consumption; flow meters installed on the pneumatic unit and purging point, used to collect compressed air consumption data; flow meters installed on the water washing section and cooling section, used to collect water consumption data. In addition, production process data is collected through the production line PLC and MES system, including product information such as substrate material, thickness, width and color, process parameters such as production line speed, curing furnace temperature setting and actual temperature of each temperature zone, coating roller pressure and fan speed, and production plan such as order number and product specification switching time point. Device state data is collected through the device control system, including device start-stop signal, standby time and load rate; environmental data is collected through environmental sensors, including environmental temperature and humidity. All collected data is synchronized with IEEE1588 precise time protocol for unified time stamping, ensuring the time consistency of multi-source data.

[0026] The data layer is in communication connection with the perception layer, used to receive and store the multi-source data collected by the perception layer, forming a corresponding database; It needs to be specifically explained in this embodiment that the data layer receives multi-source data with time stamp from the perception layer, and stores them in different databases according to data type: the energy consumption time series database stores sub-energy consumption data, including electric energy consumption data, gas consumption data, compressed air and water consumption data; the production information database stores production process data, including product information (substrate material, thickness, width and color), process parameters (production line speed, curing furnace temperature setting and actual temperature of each temperature zone, coating roller pressure and fan speed) and production plan (order number and product specification switching time point); the device state database stores device state data, including device start-stop signal, standby time and load rate. Each database adopts time series database architecture, supports high-speed writing and mass data storage, and provides data support for the upper layer analysis.

[0027] Reference Figure 2As shown, the analysis layer workflow is a closed-loop optimization process. Its core is to build a benchmark energy consumption model and use it for real-time monitoring. The process begins with training an initial model using historical data, and then enters a real-time monitoring loop: input real-time collected production and environmental data into the model to get the predicted energy consumption, compare it with the measured value to calculate the deviation. If the deviation is not beyond the threshold, continue monitoring; if it is beyond the threshold, immediately start the first-level correction, locate the key influencing factors through sensitivity analysis and generate optimization suggestions. At the same time, the system regularly performs second-level correction, adds new data to the training set to re-optimize model parameters, and realizes iterative updating of the model, thereby ensuring long-term accuracy.

[0028] The analysis layer is in communication connection with the data layer, and is used to build a benchmark energy consumption model based on historical data, to perform rolling comparison between actual energy consumption and predicted energy consumption of the benchmark energy consumption model in real-time monitoring and to calculate the deviation, to start the analysis program when the deviation exceeds the threshold, and to update the benchmark energy consumption model based on new data at regular intervals; The analysis layer selects historical data of a specific time period as a training set to build an initial benchmark energy consumption model, and obtains an initial model parameter set .

[0029] The rolling comparison includes: inputting the production process data vector P and the environmental data vector Env of the monitoring time period into the benchmark energy consumption model to obtain the predicted energy consumption , collecting the actual energy consumption of the time period , and calculating the deviation between the predicted energy consumption and the actual energy consumption. The deviation formula is: .

[0030] The operation of the analysis layer includes first-level correction and second-level correction: The first-level correction is to analyze the key variables causing the deviation and generate optimization suggestions when the deviation value exceeds the set threshold; The second-level correction is to periodically add new production data to the model training set, and to re-solve the model parameter set by minimizing the loss function: to obtain the updated benchmark energy consumption model.

[0031] In the first-level correction, the method of analyzing input variables includes calculating the partial derivative of the benchmark energy consumption model F with respect to each input variable , where is the i-th variable in the vector , and identifying the variable with a large absolute value of the partial derivative as the main influencing factor.

[0032] ​The embodiment needs to be specifically explained that the production process data vector P is constituted by the production line speed v, the unit is m / min, the curing furnace first zone temperature t1, the unit is ℃, the curing furnace second zone temperature t2, the unit is ℃, the coating roller pressure p, the unit is MPa, and the fan rotating speed n, the unit is r / min. For example, the P of a certain steady production period can be expressed as [55, 230, 250, 0.25, 1400]. The environmental data vector Env is constituted by the environmental temperature T and the environmental humidity H, for example, the Env at a certain moment is [24, 55].

[0033] In the embodiment, the reference energy consumption model F adopts the gradient boosting tree algorithm as a nonlinear mapping function, and the model parameter set including the decision tree depth, the learning rate, the leaf node number and the like. The stable production data of 20 consecutive days is selected as the reference period (in the period, the equipment is fault-free, and the product qualified rate is greater than or equal to 99%), 96 groups of data are collected every day at an interval of 15 minutes, and 1920 samples are accumulated. Each sample contains the input vector [P, Env] and the corresponding actual energy consumption value E_actual (unit: kWh). The model training is performed by minimizing the initial loss function , and the initial parameter set is finally obtained. For example, in the , the decision tree depth is set to 8, and the learning rate is 0.05. At this time, the average prediction error of the model on the reference period data is less than or equal to 3%.

[0034] The specific construction process of the reference energy consumption model F is as follows: S1, data preprocessing stage: 1920 samples in the reference period are standardized, and the input variables are mapped to the [0, 1] interval. For example, the original range of the production line speed v is 40-70 m / min, the standardization formula is v'= (v-40) / (70-40), and v=55 m / min is standardized to v'= (55-40) / 30=0.5. The original range of the environmental temperature T is 15-35 ℃, the standardization formula is T'= (T-15) / (35-15), and T=24 ℃ is standardized to T'= (24-15) / 20=0.45. At the same time, 3 groups of abnormal values, that is, the samples whose actual energy consumption exceeds 3 times the standard deviation of the mean value, are removed, and 5 groups of missing data are supplemented by using the mean value of adjacent periods.

[0035] S2, model training iteration process: S21, the actual energy consumption mean value μ=250 kWh in the reference period is taken as the initial prediction value F0, that is, F0=μ.

[0036] S22, the initial residual error is calculated, (i=1, 2,..., 1917), for example, a certain sample =260 kWh, then =260-250=10kWh. The first tree is fitted by CART algorithm to relate the residual to the input variable , which includes the splitting rule: if >0.6, go to the left child, where The normalized curing oven zone temperature, and the leaf node output value is the residual mean of this zone.

[0037] S23, the prediction value of the mth round (m=2, 3,..., 100) is: , where the learning rate α=0.05, is the decision tree fitted to the (m-1)th round residual, and the (m-1)th round residual is: .

[0038] For example, the 2nd round residual of a sample =8kWh, The output of this sample is 150, so F2=250+0.05x10+0.05x150=258kWh.

[0039] S24, the final model is: , where contains the splitting threshold, leaf node value and other parameters of 100 trees.

[0040] For input P=[55, 230, 250, 0.25, 1400], after the above standardization, P'=[0.5, 0.67, 0.83, 0.5, 0.67], Env=[24, 55], and the normalized Env'=[0.45, 0.63], the model calculation process is: The output of the first tree is =10, the output of the second tree is =8,..., and the output of the 100th tree is =2, so .

[0041] In the production process, real-time rolling monitoring is carried out, and the deviation between the actual energy consumption and the model predicted energy consumption is continuously calculated ; It is specifically pointed out in this embodiment that the rolling comparison period is set to 15 minutes, that is, data acquisition and deviation calculation are completed every 15 minutes. For example, in a certain monitoring period: the real-time collected production process data P=[56, 228, 248, 0.26, 1380], and the environment data Env=[25, 56] are input into the benchmark energy consumption model, and the predicted energy consumption = 268 kWh; at the same time, the actual energy consumption in this period is obtained through the data layer = 291 kWh, and the deviation AE = 291-268 = 23 kWh is calculated. The system preset deviation threshold is ±20 kWh, since 23 kWh > 20 kWh, the analysis program is triggered.

[0042] When the AE exceeds the threshold, a first-level correction is performed, and a second-level correction is performed every 30 days to update the model parameter set ; It is specifically pointed out in this embodiment that in the first-level correction, a sensitivity analysis method is used to calculate the partial derivative of the model output to each input variable : it is calculated that the influence of the temperature of the curing furnace zone on the energy consumption is = 4.8 kWh / ℃, the influence of the production line speed on the energy consumption is = 2.1 kWh / (m / min), and the influence of the ambient temperature on the energy consumption is = 0.5 kWh / ℃, wherein the partial derivative of the temperature of the curing furnace zone has the largest absolute value, and is identified as the main influencing factor, and an optimization suggestion “check the temperature control system of the curing furnace zone and calibrate the temperature sensor” is generated.

[0043] In the second-level correction, 7200 new data groups (240 groups per day) in the past 30 days are added to the training set, and an adaptive gradient descent algorithm is used to minimize the loss function . For example, the loss function value before correction is 15600, after iterative optimization, the new parameter set θ makes the loss function decrease to 9800, and the model prediction accuracy is improved to an average error ≤2.5%, and the update of the baseline energy consumption model is completed.

[0044] Referring to Figure 3 , the application layer receives the processing result of the analysis layer, and four functions are specifically implemented. The energy consumption monitoring function visualizes the data, displays the real-time data, historical trend curve and equipment energy consumption ranking of the sub-energy consumption. The energy efficiency abnormality alarm function triggers a first-level alarm, a second-level alarm or a pre-alarm according to the size of the deviation value AE, and notifies the relevant personnel through various ways. The optimization strategy recommendation function intelligently generates specific optimization suggestions including order production scheduling, process parameter adjustment and equipment maintenance reminder based on the analysis conclusion. The carbon emission accounting function converts and counts various types of energy consumption in real time according to the preset factors, generates periodical reports and compares with the industry benchmark, and quantifies the energy saving and carbon reduction achievements.

[0045] The application layer is in communication connection with the analysis layer, and is used to realize energy consumption monitoring, energy efficiency abnormality alarm, optimization strategy recommendation and carbon emission accounting based on the output result of the analysis layer.

[0046] The optimization strategy recommendation includes scheduling optimization, parameter optimization and maintenance reminder; the scheduling optimization is to suggest concentrating production of orders of the same color and the same thickness; the parameter optimization is to recommend adjusting process parameters in the production process data vector P under the premise of ensuring quality; the maintenance reminder is to prompt equipment maintenance when the equipment state data indicates performance degradation.

[0047] The application layer energy consumption monitoring module is in communication connection with the analysis layer, and based on the real-time energy consumption data, the predicted energy consumption value and the deviation ΔE output by the analysis layer, a visual monitoring interface is constructed to realize dynamic monitoring of energy consumption; The monitoring content includes: sub-item energy consumption real-time data: instantaneous value and cumulative value of electric energy, unit kWh, gas, unit m³, compressed air, unit m³, water, unit m³, such as displaying "the current electric energy consumption of the curing furnace: 120 kWh / h, today's cumulative value: 2880 kWh"; Energy consumption trend curve: draw the 24-hour energy consumption change curve with an interval of 15 minutes, superimpose the predicted energy consumption curve and the actual energy consumption curve of the analysis layer, for example, the actual energy consumption is 268 kWh and the predicted energy consumption is 255 kWh in the 10:00-10:15 period, and the curve presents a slight upward deviation; Device energy consumption ranking: sort the unit product energy consumption of key devices such as pretreatment device, roller coater and curing furnace, and update the TOP3 high energy consumption devices in real time.

[0048] Energy efficiency abnormality alarm is based on the deviation ΔE and the main influencing factors output by the analysis layer, and a multi-level alarm mechanism is triggered when ΔE exceeds the threshold value; It needs to be specifically explained in this embodiment that the alarm system is composed of a local audible and light alarm (decibel ≥ 85 dB), a PLC control system and a mobile terminal APP, and the alarm levels are divided into: First-level alarm (ΔE>20kWh): trigger red audible and light alarm, and push alarm information to the mobile APP of the production supervisor, including "the curing furnace energy consumption deviation is 23 kWh in the 14:00-14:15 period, the main influencing factor is the temperature of the curing furnace area (partial derivative 4.8 kWh / ℃)", and the real-time energy consumption data screenshot is attached; Second-level alarm (10kWh<ΔE≤20kWh): trigger yellow audible and light alarm and push information to the terminal of the team leader; Early warning (5kWh<ΔE≤10kWh): only display an orange early warning icon on the monitoring interface, without triggering audible and light alarm.

[0049] The alarm records are automatically stored in the database for more than 1 year, and the history alarm can be queried according to time period and device type.

[0050] The optimization strategy recommendation is based on the deviation reason and the model sensitivity analysis result output by the analysis layer, and three types of strategies of scheduling optimization, parameter optimization and maintenance reminder are generated; The embodiment needs to be specifically explained that the strategy generation adopts the combination of rule engine and case reasoning, and the output form is editable PDF report and system pop-up prompt: Dispatch optimization: when the analysis layer identifies that multiple orders are frequently switched, it recommends "concentrating the production of orders with the same color and thickness", for example, a batch contains 3 galvanized sheet orders (thickness 0.3mm, color gray-white) and 2 cold-rolled sheet orders (thickness 0.5mm, color dark blue), the system suggests adjusting to "first continuously produce 3 gray-white galvanized sheet orders, expected to reduce color change downtime by 2 times, save energy 150kWh, then produce dark blue cold-rolled sheet orders"; Parameter optimization: based on the partial derivative analysis, the recommended process parameter adjustment range is "the temperature of the curing oven zone 1 can be reduced from 220°C to 215°C, the deviation affects 4.8kWh / °C x 5°C = 24kWh, while keeping the production line speed at 55m / min, it is expected to save 24kWh of electricity per hour", and it is marked "the adhesion of the coating must be ≥5N / cm²"; Maintenance reminder: when the equipment state data shows that "the load rate of the air compressor has been below 70% for 8 hours, and the rated load rate is 85%", push "suggest checking the cleanliness of the air compressor filter, expected to increase the load rate to more than 80% after maintenance, reduce unit energy consumption by 12%";

[0051] Carbon emission accounting is based on the sub-item energy consumption data output by the analysis layer and the preset carbon emission factor to realize real-time calculation and statistics of carbon emissions; The embodiment needs to be specifically explained that the carbon emission factor is: electricity 0.610tCO2 / MWh, gas 0.194tCO2 / m³, compressed air 0.120tCO2 / m³, water 0.001tCO2 / m³.

[0052] Further need to be explained is that the carbon emission calculation method is: , For example: the carbon emission in a certain period is: The carbon emission calculated from the electricity consumption of 2880kWh is 1.7568tCO2; The carbon emission calculated from the gas consumption of 1200m³ is 0.2328tCO2; The total carbon emission is 1.9896tCO2.

[0053] The accounting result generates carbon emission report by day, week and month, supports comparison with industry benchmark value, such as 3.2tCO2 / ton of product, and displays "the current unit product carbon emission is 2.8tCO2 / ton, which is 12.5% better than the industry benchmark".

[0054] Secondly: the embodiment of the present application discloses only the structure related to the embodiment of the present application, other structures can refer to the general design, and in the case of no conflict, the same embodiment and different embodiments of the present application can be combined with each other; Finally: the above is only the preferred embodiment of the present application, and is not used to limit the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A system for monitoring and optimizing energy consumption in a color coating production line based on multi-source data fusion, characterized in that, include: The perception layer includes various sensors installed on key equipment in the color coating production line to collect multi-source data, including energy consumption data, production process data, equipment status data, and environmental data. The data layer, which is communicatively connected to the perception layer, is used to receive and store multi-source data collected by the perception layer to form a corresponding database. The analysis layer, which is connected to the data layer, is used to build a benchmark energy consumption model based on historical data, to perform rolling comparisons between the actual energy consumption and the predicted energy consumption of the benchmark energy consumption model in real-time monitoring and to calculate the deviation, to start the analysis program when the deviation exceeds a threshold, and to periodically update the benchmark energy consumption model based on new data. The application layer, which communicates with the analysis layer, is used to monitor energy consumption, issue energy efficiency alarms, recommend optimization strategies, and calculate carbon emissions based on the output of the analysis layer.

2. The energy consumption monitoring and optimization system for a color coating production line based on multi-source data fusion as described in claim 1, characterized in that, The energy consumption data includes electricity consumption data, gas consumption data, compressed air and water consumption data; the production process data includes product information, process parameters and production plan; the equipment status data includes equipment start / stop signals, standby time and load rate; the environmental data includes ambient temperature and humidity.

3. The energy consumption monitoring and optimization system for a color coating production line based on multi-source data fusion as described in claim 2, characterized in that, The product information includes substrate material, thickness, width, and color; the process parameters include production line speed, curing oven temperature, coating roller pressure, and fan speed.

4. The energy consumption monitoring and optimization system for a color coating production line based on multi-source data fusion as described in claim 1, characterized in that, The baseline energy consumption model employs a machine learning algorithm and is represented as a function: ， Where E represents predicted energy consumption, P is the production process data vector, Env is the environmental data vector, and F is the nonlinear mapping function. The parameter set is obtained by the model through training.

5. The energy consumption monitoring and optimization system for a color coating production line based on multi-source data fusion as described in claim 4, characterized in that, The analysis layer selects historical data from a specific time period as a training set to construct the initial baseline energy consumption model and obtain the initial model parameter set. .

6. The energy consumption monitoring and optimization system for a color coating production line based on multi-source data fusion as described in claim 1, characterized in that, The rolling comparison includes: within the monitoring period, inputting the production process data vector P and the environmental data vector Env for that period into the baseline energy consumption model to obtain predicted energy consumption. Collect the actual energy consumption during this time period. And calculate the deviation between the predicted energy consumption and the actual energy consumption. The formula for calculating the deviation is: 。 7. The energy consumption monitoring and optimization system for a color coating production line based on multi-source data fusion as described in claim 6, characterized in that, The operations of the analysis layer include first-level correction and second-level correction: The first-level correction is when the deviation value When the set threshold is exceeded, the key variables causing the deviation are analyzed and optimization suggestions are generated; The secondary correction involves periodically adding newly added production data to the model training set and minimizing the loss function through an optimization algorithm. To resolve the model parameter set This yields the updated baseline energy consumption model.

8. The energy consumption monitoring and optimization system for a color coating production line based on multi-source data fusion as described in claim 7, characterized in that, In the first-level correction, the method for analyzing the input variables includes calculating the partial derivatives of the baseline energy consumption model F with respect to each input variable. ,in For vectors The i-th variable in the partial derivative is identified, and the variable with the largest absolute value of the partial derivative is identified as the main influencing factor.

9. The energy consumption monitoring and optimization system for a color coating production line based on multi-source data fusion as described in claim 1, characterized in that, The optimization strategy recommendations include scheduling optimization, parameter optimization, and maintenance reminders; the scheduling optimization suggests concentrating production of orders with the same color and thickness; the parameter optimization suggests adjusting the process parameters in the production process data vector P while ensuring quality; the maintenance reminders prompt equipment maintenance when equipment status data indicates performance degradation.

Images