Energy consumption monitoring and analysis system of smart factory

Abstract

The invention relates to the field of energy consumption monitoring and analysis, and particularly discloses an energy consumption monitoring and analysis system of a smart factory. According to the invention, various sensor meters are installed in the factory to detect the energy consumption, then the collected data are summarized and analyzed, and then the data are displayed in a report form, so that workers can easily know the energy consumption in the factory through the chart, and the prediction efficiency is greatly improved when the energy consumption of the factory is predicted. The method comprises the following steps: performing quantitative prediction on an energy consumption result through a time sequence prediction method and a regression method, performing relatively accurate future energy consumption prediction on the basis of previous energy consumption data of a factory, performing subjective qualitative prediction, and correcting and perfecting the quantitative prediction result through opinions and experiences of professionals in the field, so as to achieve the purpose of energy consumption prediction. According to a prediction result obtained through a time sequence, regression analysis and qualitative analysis technology, deterministic factors and uncertain factors are comprehensively considered, the accuracy is higher, a factory can be effectively helped to predict future factory energy consumption, and factory energy can be adjusted and distributed in time.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy consumption monitoring and analysis, and specifically relates to an energy consumption monitoring and analysis system for a smart factory. Background Art

[0002] The advancement of industry is inseparable from the promotion of energy. Most modern industries rely on electricity. With the pressure of global warming and increasingly scarce resources, more and more factories and enterprises have put forward higher requirements for energy conservation and emission reduction. Energy conservation and emission reduction also means reduced operating costs. Therefore, how to effectively monitor and manage energy consumption has become a problem to be solved by many industrial enterprises.

[0003] In the Chinese patent with publication number CN117908484A, a smart factory energy consumption detection and analysis system is mentioned, which includes an acquisition module, a communication module, a data management module and a detection and analysis platform mounted in a cloud server. The data end of the acquisition module transmits data bidirectionally with the data management module through the communication module. The port of the detection and analysis platform is connected to the data management module. The detection and analysis platform is internally built with a data integration module, an energy consumption comparison module, an energy consumption analysis module, an energy consumption prediction module, an AI correction module and an AI decision-making module. On the one hand, this energy consumption monitoring and analysis system can perform AI intelligent correction on the energy consumption prediction model according to the terminal conditions of the factory equipment, thereby improving the prediction accuracy of the energy consumption detection and prediction model. On the other hand, after detecting and analyzing the energy consumption, this detection and analysis system can intelligently calculate the abnormal energy consumption depreciation of a certain equipment.

[0004] During the application process, the applicant discovered that the above application had defects during use. Although the application could calculate the total predicted energy consumption within the expected work plan based on the work plan parameters of a certain equipment in the factory and the matching power of the equipment, the equipment energy consumption is only a part of the factory energy consumption, and the equipment energy consumption is not always linearly related to the set production plan. Some factors outside the equipment will also cause changes in weekly and monthly equipment energy consumption values, such as holidays, increased equipment operating time due to increased orders, etc. AI cannot effectively judge the energy consumption caused by these factors. Summary of the invention

[0005] The purpose of the present invention is to provide an energy consumption monitoring and analysis system for a smart factory to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An energy consumption monitoring and analysis system for a smart factory, specifically comprising:

[0008] A data collection layer, which is used to accurately collect factory energy consumption, including electricity consumption, water consumption and gas consumption;

[0009] The data transmission layer is used to transmit the collected engineering energy consumption data to the data processing center via wired or wireless means;

[0010] Data processing and storage layer, which stores and analyzes the received data, generates analytical data for predicting the energy consumption trend of the factory, and provides a basis for energy-saving diagnosis, analysis, management and transformation of the factory;

[0011] The system management layer serves as the overall control platform for factory energy consumption management and decision-making operations.

[0012] Preferably, the data acquisition layer collects energy consumption data in real time through sensors and metering equipment installed inside the factory, including electricity meters, water meters and gas meters. The data transmission layer is composed of Ethernet and industrial Ethernet wired networks and wireless LAN, cellular network and LoRaWAN wireless network. The data processing and storage layer receives the collected data through the data transmission layer, analyzes and stores the data, and the system management layer performs statistical analysis and display on the processed data, and configures corresponding function management ports.

[0013] Preferably, the system management layer includes:

[0014] Energy consumption monitoring and statistics module, which consists of regional energy consumption submodule, energy consumption report submodule, classification and item submodule, power distribution monitoring submodule, energy consumption warning submodule, energy consumption map submodule, time-sharing energy consumption submodule and instrument management submodule;

[0015] An energy consumption analysis and prediction module, which is composed of a power distribution trend submodule, an energy consumption comparison submodule, an energy consumption ranking submodule, an energy consumption index submodule, a quota management submodule, a loss analysis submodule, an equipment utilization submodule and an energy consumption prediction submodule;

[0016] The energy consumption billing management module is composed of an energy consumption billing authority management submodule, an energy consumption log management submodule, an energy consumption bill management submodule and an energy consumption dashboard.

[0017] Preferably, the energy consumption prediction submodule has a built-in energy consumption prediction algorithm, and uses data processing and storage layer computing power to perform energy consumption prediction, specifically including the following steps:

[0018] S1, divide the factory energy consumption data into three parts: energy consumption of production workshop, air conditioning energy consumption and other energy consumption. Set the other energy consumption as a constant, collect factory energy consumption data, clean and supplement the collected data, correspond working days to factory energy consumption, the dependent variable is factory energy consumption, generate time variables and draw time series graphs;

[0019] S2, use data analysis software to establish a time series model, automatically check the outliers of the time series model and eliminate their influence, establish a model graph display including measured values, predicted values ​​and fitted values, retain the autocorrelation function ACF and partial autocorrelation function PACF of the residual, and derive the corresponding optimal time series model;

[0020] S3, input the time independent variable into the time series model to predict the dependent variable, and predict the energy consumption data of the factory at the corresponding time point;

[0021] S4, using regression analysis to improve forecast results and compensate for the impact of external factors on factory consumption;

[0022] S5, introduces a qualitative prediction mechanism, performs qualitative prediction adjustments on the improved prediction results, and obtains the final prediction results.

[0023] Preferably, in S1, the Lagrange interpolation method is used to supplement the missing plant energy consumption data, which is specifically expressed as:

[0024]

[0025] Among them, x is the independent variable, l j (x) takes the value of 1 at xj and takes the value of 0 at other points xi, i≠j. Substitute the data before and after the missing value into the Lagrange polynomial for calculation to obtain the missing value of the day.

[0026] Preferably, the data analysis software used in S2 is Spss modeler, and outlier detection and influence elimination include:

[0027] Additive: outliers that affect individual observations;

[0028] Horizontal shift: All observations starting from a certain sequence point shift a constant outlier;

[0029] Innovation: An outlier attached to the noise term from a specific sequence point. For a stationary series, an innovation outlier will affect multiple observations, and for a non-stationary series, it will affect every observation starting at a specific sequence point.

[0030] Transient: outliers whose impact decays exponentially to 0;

[0031] Seasonal additivity: an outlier affects a particular observation and all subsequent observations separated from it by one or more seasonal components;

[0032] Local trend: outliers that start a local trend from a specific sequence point;

[0033] Additive patching: a group of two or more consecutive additive outliers;

[0034] The Spss software tests the above items of data for outliers and eliminates their influence.

[0035] Preferably, in S4, using regression analysis to improve the prediction results specifically includes the following steps:

[0036] S41, manually find the regression analysis point, determine the time point of the regression analysis point and the corresponding energy consumption;

[0037] S42, determine the independent variables, find all the independent variables for analysis, discard the independent variables with medium correlation or below, the correlation coefficient R value of ±0.5 to ±0.7 is medium correlation, below ±0.5 is weak correlation, and above ±0.7 is strong correlation. Perform multicollinearity assessment on the remaining independent variables, remove or merge the independent variables with multicollinearity, and obtain the final independent variables;

[0038] S43, solving the multiple regression equation and performing a significance test of the regression equation, and then substituting the relevant data into the multiple regression equation to obtain a prediction result;

[0039] S44, returning to the regression analysis point, and using the multiple linear regression prediction results to correct and improve the preliminary prediction to obtain improved prediction data, the improved prediction data value being the average of the preliminary prediction value and the multiple linear regression prediction value.

[0040] Preferably, in S5, the specific steps of introducing a qualitative prediction mechanism to perform qualitative prediction adjustment include:

[0041] S51, select experts to participate in qualitative adjustment and form a qualitative adjustment expert team;

[0042] S52, the qualitative adjustment experts independently give prediction results, and score each expert's prediction result, using the expert's final score as the weight, and weighted average calculation to obtain the qualitative prediction value;

[0043] S53, adjusting the prediction value after regression improvement with the qualitative prediction value to obtain a final prediction result, and the final prediction result is the average of the qualitative prediction value and the prediction value after regression improvement.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention installs a variety of sensor meters inside the factory to detect the factory's energy consumption such as water, electricity and gas, and then summarizes and analyzes the collected data, and then displays it in the form of a report. The staff can easily understand the energy consumption of each group of equipment, each group of process flows, each workshop or each area in the factory through charts and graphs. At the same time, the factory energy consumption is predicted based on the collected data. The energy consumption results are first quantitatively predicted through time series prediction and regression methods, and a more accurate future energy consumption prediction is made based on the factory's previous energy consumption data. Then, subjective qualitative predictions are added, and the quantitative prediction results are corrected and improved through the opinions and experience of professionals in the field. The prediction results obtained through time series, regression analysis and qualitative analysis techniques comprehensively consider the deterministic factors and uncertain factors of the enterprise's energy consumption, and have higher accuracy, which can effectively help the factory predict future factory energy consumption, so as to timely adjust the allocation of factory energy and complete the target operation with lower energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a block diagram of the energy consumption monitoring and analysis system of the present invention;

[0047] Figure 2 This is a block diagram of the energy consumption prediction algorithm of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] Example:

[0050] See also Figure 1 - Figure 2 As shown, a smart factory energy consumption monitoring and analysis system specifically includes:

[0051] A data collection layer, which is used to accurately collect factory energy consumption, including electricity consumption, water consumption and gas consumption;

[0052] The data transmission layer is used to transmit the collected engineering energy consumption data to the data processing center via wired or wireless means;

[0053] Data processing and storage layer, which stores and analyzes the received data, generates analytical data for predicting the energy consumption trend of the factory, and provides a basis for energy-saving diagnosis, analysis, management and transformation of the factory;

[0054] The system management layer serves as the overall control platform for factory energy consumption management and decision-making operations.

[0055] From the above, we can see that the data collection layer collects energy consumption in various parts of the factory, mainly electricity, water and gas, and then passes the data to the data processing and storage layer through the data transmission layer. The data is analyzed while being saved. Relevant operating personnel can view the summarized collected data and data analysis results through the system management layer, and then take corresponding measures.

[0056] The data acquisition layer collects energy consumption data in real time through sensors and metering equipment installed inside the factory, including electricity meters, water meters and gas meters. If the factory also needs to consume other energy, other sensors can be set up to collect consumption data, or energy consumption can be recorded manually. The data transmission layer is composed of Ethernet and industrial Ethernet wired networks and wireless LAN, cellular network and LoRaWAN wireless network. The data processing and storage layer receives the collected data through the data transmission layer, analyzes and stores the data. The system management layer performs statistical analysis and display on the processed data, and configures the corresponding function management port.

[0057] The system management layer includes:

[0058] Energy consumption monitoring and statistics module, which consists of regional energy consumption submodule, energy consumption report submodule, classification and item submodule, power distribution monitoring submodule, energy consumption warning submodule, energy consumption map submodule, time-sharing energy consumption submodule and instrument management submodule;

[0059] An energy consumption analysis and prediction module, which is composed of a power distribution trend submodule, an energy consumption comparison submodule, an energy consumption ranking submodule, an energy consumption index submodule, a quota management submodule, a loss analysis submodule, an equipment utilization submodule and an energy consumption prediction submodule;

[0060] The energy consumption billing management module is composed of an energy consumption billing authority management submodule, an energy consumption log management submodule, an energy consumption bill management submodule and an energy consumption dashboard.

[0061] In the energy consumption monitoring and statistics module, the regional energy consumption submodule can display the energy consumption of different areas of the factory, the energy consumption report submodule displays the energy consumption in a graphical form, the classification and sub-item submodule is used to display different types of energy consumption, the power distribution monitoring submodule is used to detect the power distribution consumption of each part, the energy consumption warning submodule is used for energy consumption alarms, the energy consumption map submodule is used to display the energy consumption of different module areas in the factory or the energy consumption of different process modules, the time-sharing energy consumption submodule is used to display the energy consumption of equipment in different time periods, and the instrument management submodule is used to manage instruments in various parts of the factory.

[0062] In the energy consumption billing management module, the energy consumption billing authority management submodule is used to manage and set energy consumption billing, so as to facilitate the total cost according to the energy consumption billing standard. The energy consumption log management submodule is used to record energy consumption related information for archiving management. The energy consumption bill management submodule is used to manage energy consumption bills. The energy consumption dashboard displays the energy consumption of each part in the style of a dashboard.

[0063] In the energy consumption analysis and prediction module, the power distribution trend submodule is used to display the power distribution change trend of each part, the energy consumption comparison submodule is used to compare and display the energy consumption of each part, the energy consumption ranking submodule can rank each part according to the energy consumption, the energy consumption index submodule can display the energy consumption index of each part of the factory, the quota management submodule is used to display the energy amount allocated to each part of the factory, the loss analysis submodule is used to analyze the energy loss of each module in the factory, and the analysis method can adopt the common general equipment energy consumption analysis method, the equipment utilization rate submodule is used to display the energy utilization efficiency of the equipment, and the energy consumption prediction submodule is used to predict energy consumption.

[0064] The energy consumption prediction submodule has a built-in energy consumption prediction algorithm, which uses data processing and storage layer computing power to perform energy consumption prediction, specifically including the following steps:

[0065] S1, divide the factory energy consumption data into three parts: energy consumption of production workshop, air conditioning energy consumption and other energy consumption. Set the other energy consumption as a constant, collect factory energy consumption data, clean and supplement the collected data, correspond working days to factory energy consumption, the dependent variable is factory energy consumption, generate time variables and draw time series graphs;

[0066] S2, use data analysis software to establish a time series model, automatically check the outliers of the time series model and eliminate their influence, establish a model graph display including measured values, predicted values ​​and fitted values, retain the autocorrelation function ACF and partial autocorrelation function PACF of the residual, which is helpful to determine whether the residual is white noise and derive the corresponding optimal time series model;

[0067] S3, input the time independent variable into the time series model to predict the dependent variable, and predict the energy consumption data of the factory at the corresponding time point;

[0068] S4, using regression analysis to improve forecast results and compensate for the impact of external factors on factory consumption;

[0069] S5, introduces a qualitative prediction mechanism, performs qualitative prediction adjustments on the improved prediction results, and obtains the final prediction results.

[0070] From the above, we can see that the factory's equipment production energy consumption and air-conditioning energy consumption can be quantitatively predicted through the collected factory's historical energy consumption data, and then the equipment production energy consumption, air-conditioning energy consumption and the quantitative remaining energy consumption are added together to get the total energy consumption of the factory at a certain point in the future. At the same time, the prediction method is divided into multiple steps. First, the time series model is used for prediction, and then regression analysis is used to improve the prediction results and improve the accuracy of the prediction results. Finally, a qualitative prediction mechanism is introduced to add some subjective factors other than data to make the prediction results more in line with reality.

[0071] In S1, the Lagrange interpolation method is used to supplement the missing factory energy consumption data, which is specifically expressed as:

[0072]

[0073] Among them, x is the independent variable, l j (x) takes the value of 1 at xj and takes the value of 0 at other points xi, i≠j. Substitute the data before and after the missing value into the Lagrange polynomial for calculation to obtain the missing value of the day.

[0074] The data analysis software used in S2 is the Spss modeler, which includes the following when performing outlier testing and eliminating influence:

[0075] Additive: outliers that affect individual observations;

[0076] Horizontal shift: All observations starting from a certain sequence point shift a constant outlier;

[0077] Innovation: An outlier attached to the noise term from a specific sequence point. For a stationary series, an innovation outlier will affect multiple observations, and for a non-stationary series, it will affect every observation starting at a specific sequence point.

[0078] Transient: outliers whose impact decays exponentially to 0;

[0079] Seasonal additivity: an outlier affects a particular observation and all subsequent observations separated from it by one or more seasonal components;

[0080] Local trend: outliers that start a local trend from a specific sequence point;

[0081] Additive patching: a group of two or more consecutive additive outliers;

[0082] The Spss software tests the above items of data for outliers and eliminates their influence.

[0083] Spss software can give the most appropriate model based on relevant data, and use this model to predict future data. For example, when predicting quarterly energy consumption, the Winters additive model can be used for calculation. The Spss software calculates the model's goodness of fit statistical stationary R-square value, Young-Box Q statistical value, DF value, significance value and number of outliers. When the significance value is greater than 0.05, the original hypothesis cannot be rejected and it is white noise. Then the exponential smoothing model parameters of the Winters additive model are obtained, including the estimated values ​​of the level parameter, trend parameter and seasonal parameter, standard error value, t and significance. These values ​​can be used to calculate the specific values ​​of various parameters in the Winters additive equation, and then test them. The residual test of white noise is performed, and the collected original data is fitted and predicted through the Winters additive model to obtain energy consumption data at future time points.

[0084] In S4, using regression analysis to improve the prediction results specifically includes the following steps:

[0085] S41. Manually find regression analysis points, determine the time points of the regression analysis points and the corresponding energy consumption, such as time points with large changes in energy consumption, increased air conditioning energy consumption when the temperature is high in summer, or sudden increases in energy consumption caused by increased factory orders, etc.;

[0086] S42, determine the independent variables, find all the independent variables for analysis, discard the independent variables with medium correlation or below, the correlation coefficient R value of ±0.5 to ±0.7 is medium correlation, below ±0.5 is weak correlation, and above ±0.7 is strong correlation. Perform multicollinearity assessment on the remaining independent variables, remove or merge the independent variables with multicollinearity, and obtain the final independent variables;

[0087] S43. Solve the multiple regression equation and perform a significance test on the regression equation. Then substitute the relevant data into the multiple regression equation to obtain the prediction result. When performing the calculation, you can set y as the dependent variable, x1, x2, ...x k is the independent variable, and there is a linear relationship between the independent variable and the dependent variable, and the formula is expressed as:

[0088] y=b0+b1+b2k2+…+b k x k +e

[0089] Among them, b0 is a constant term, b1, b2, … b k is the regression coefficient, b1 is x2, x3…x k When x1 is fixed, the effect of each unit increase in x1 on y is the partial regression coefficient of x1 on y. Similarly, b2 is the effect of x1, x k When x2 is fixed, the effect of each unit increase in x2 on y is the partial regression coefficient of x2 on y. If the two independent variables x1 and x2 are linearly correlated with the same dependent variable y, the binary linear regression model can be described as

[0090] y=b0+b1x1+b2x2+e

[0091] The parameter estimation of the multivariate regression model, like the linear regression equation, also requires the sum of squared errors Σe 2 Under the premise that is the smallest, the least square method is used to solve the parameters, then the standard equation system of regression parameters is expressed as

[0092] Σy=nb0+b1Σx1+b2Σx2

[0093]

[0094] Solving this equation can obtain the values ​​of b0, b1 and b2, and then solve the multiple regression equation to calculate the predicted value;

[0095] S44, returning to the regression analysis point, and using the multiple linear regression prediction results to correct and improve the preliminary prediction to obtain improved prediction data, the improved prediction data value being the average of the preliminary prediction value and the multiple linear regression prediction value.

[0096] In S5, the specific steps of introducing the qualitative prediction mechanism to perform qualitative prediction adjustment include:

[0097] S51, select experts to participate in qualitative adjustment and form a qualitative adjustment expert team;

[0098] S52, the qualitative adjustment experts independently give the forecast results and score each expert's forecast result. The scoring standard is determined by the manufacturer based on the ability and experience of each expert, previous forecast performance and the degree of expert participation. The stronger the expert's ability, the better the past forecast performance and the higher the degree of participation, the higher the score, which means that the forecast result given by the expert occupies a greater proportion in the final forecast result. The final score of the expert is used as the weight, and the weighted average is calculated to obtain the qualitative forecast value;

[0099] S53, adjusting the prediction value after regression improvement with the qualitative prediction value to obtain a final prediction result, and the final prediction result is the average of the qualitative prediction value and the prediction value after regression improvement.

[0100] In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other.

Claims

1. A smart factory energy consumption monitoring and analysis system, characterized in that: Specifically include: A data collection layer, which is used to accurately collect factory energy consumption, including electricity consumption, water consumption and gas consumption; The data transmission layer is used to transmit the collected engineering energy consumption data to the data processing center via wired or wireless means; Data processing and storage layer, which stores and analyzes the received data, generates analytical data for predicting the energy consumption trend of the factory, and provides a basis for energy-saving diagnosis, analysis, management and transformation of the factory; The system management layer serves as the overall control platform for factory energy consumption management and decision-making operations.

2. The energy consumption monitoring and analysis system for a smart factory according to claim 1, characterized in that: The data acquisition layer collects energy consumption data in real time through sensors and metering equipment installed inside the factory, including electricity meters, water meters and gas meters. The data transmission layer is composed of Ethernet and industrial Ethernet wired networks and wireless LAN, cellular network and LoRaWAN wireless network. The data processing and storage layer receives the collected data through the data transmission layer, analyzes and stores the data. The system management layer performs statistical analysis and display on the processed data, and configures the corresponding function management port.

3. The energy consumption monitoring and analysis system for a smart factory according to claim 2 is characterized in that: The system management layer includes: Energy consumption monitoring and statistics module, which consists of regional energy consumption submodule, energy consumption report submodule, classification and item submodule, power distribution monitoring submodule, energy consumption warning submodule, energy consumption map submodule, time-sharing energy consumption submodule and instrument management submodule; An energy consumption analysis and prediction module, which is composed of a power distribution trend submodule, an energy consumption comparison submodule, an energy consumption ranking submodule, an energy consumption index submodule, a quota management submodule, a loss analysis submodule, an equipment utilization submodule and an energy consumption prediction submodule; The energy consumption billing management module is composed of an energy consumption billing authority management submodule, an energy consumption log management submodule, an energy consumption bill management submodule and an energy consumption dashboard.

4. The energy consumption monitoring and analysis system for a smart factory according to claim 3 is characterized in that: The energy consumption prediction submodule has a built-in energy consumption prediction algorithm, which uses data processing and storage layer computing power to perform energy consumption prediction. The following steps are involved: S1, divide the factory energy consumption data into three parts: energy consumption of production workshop, air conditioning energy consumption and other energy consumption. Set the other energy consumption as a constant, collect factory energy consumption data, clean and supplement the collected data, correspond working days to factory energy consumption, the dependent variable is factory energy consumption, generate time variables and draw time series graphs; S2, use data analysis software to establish a time series model, automatically check the outliers of the time series model and eliminate their influence, establish a model graph display including measured values, predicted values ​​and fitted values, retain the autocorrelation function ACF and partial autocorrelation function PACF of the residual, and derive the corresponding optimal time series model; S3, input the time independent variable into the time series model to predict the dependent variable, and predict the energy consumption data of the factory at the corresponding time point; S4, using regression analysis to improve forecast results and compensate for the impact of external factors on factory consumption; S5, introduces a qualitative prediction mechanism, performs qualitative prediction adjustments on the improved prediction results, and obtains the final prediction results.

5. The energy consumption monitoring and analysis system for a smart factory according to claim 4 is characterized in that: In S1, the Lagrange interpolation method is used to supplement the missing factory energy consumption data, which is specifically expressed as: Among them, x is the independent variable, l j (x) takes the value of 1 at xj and takes the value of 0 at other points xi, i≠j. Substitute the data before and after the missing value into the Lagrange polynomial for calculation to obtain the missing value of the day.

6. The energy consumption monitoring and analysis system for a smart factory according to claim 5, characterized in that: The data analysis software used in S2 is the Spss modeler, which includes the following when performing outlier testing and eliminating influence: Additive: outliers that affect individual observations; Horizontal shift: All observations starting from a certain sequence point shift a constant outlier; Innovation: An outlier attached to the noise term from a specific sequence point. For a stationary series, an innovation outlier will affect multiple observations, and for a non-stationary series, it will affect every observation starting at a specific sequence point. Transient: outliers whose impact decays exponentially to 0; Seasonal additivity: an outlier affects a particular observation and all subsequent observations separated from it by one or more seasonal components; Local trend: outliers that start a local trend from a specific sequence point; Additive patching: a group of two or more consecutive additive outliers; The Spss software tests the above items of data for outliers and eliminates their influence.

7. The energy consumption monitoring and analysis system for a smart factory according to claim 6 is characterized in that: In S4, using regression analysis to improve the prediction results specifically includes the following steps: S41, manually find the regression analysis point, determine the time point of the regression analysis point and the corresponding energy consumption; S42, determine the independent variables, find all the independent variables for analysis, discard the independent variables with medium correlation or below, the correlation coefficient R value of ±0.5 to ±0.7 is medium correlation, below ±0.5 is weak correlation, and above ±0.7 is strong correlation. Perform multicollinearity assessment on the remaining independent variables, remove or merge the independent variables with multicollinearity, and obtain the final independent variables; S43, solving the multiple regression equation and performing a significance test of the regression equation, and then substituting the relevant data into the multiple regression equation to obtain a prediction result; S44, returning to the regression analysis point, and using the multiple linear regression prediction results to correct and improve the preliminary prediction to obtain improved prediction data, the improved prediction data value being the average of the preliminary prediction value and the multiple linear regression prediction value.

8. The energy consumption monitoring and analysis system for a smart factory according to claim 7, characterized in that: In S5, the specific steps of introducing the qualitative prediction mechanism to perform qualitative prediction adjustment include: S51, select experts to participate in qualitative adjustment and form a qualitative adjustment expert team; S52, the qualitative adjustment experts independently give prediction results, and score each expert's prediction result, using the expert's final score as the weight, and weighted average calculation to obtain the qualitative prediction value; S53, adjusting the prediction value after regression improvement with the qualitative prediction value to obtain a final prediction result, and the final prediction result is the average of the qualitative prediction value and the prediction value after regression improvement.

Images