TW202236198A - Artificial intelligent manufacturing & production energy-saving system and method thereof - Google Patents

Abstract

An artificial intelligent manufacturing & production energy-saving system is provided, which includes a monitoring module and a prediction module. The monitoring module collects the historical electricity consumption data and the historical temperature data of an area. The prediction module receives the historical electricity consumption data, the historical temperature data and the temperature forecasting data of the area. The prediction module executes a pre-processing process to preprocess the historical electricity consumption data and the historical temperature data. Then, the prediction module performs a training model based on XGboost algorithm to execute a training process according to the temperature forecasting data of a prediction period, all historical electricity consumption data and historical temperature data before the prediction period so as to generate an electricity consumption prediction result of the prediction period. Afterward, the prediction module generating an evaluation result according to an evaluation indicator and the electricity consumption prediction so as to determine whether to adopt the electricity consumption prediction result or recreate the training model.

Description

Artificial intelligence production and manufacturing energy-saving management system and method

This disclosure is about an energy-saving management system, especially an energy-saving management system for artificial intelligence production and manufacturing. The present disclosure also relates to the energy saving management method of the system.

Due to the advancement of technology, the functions of the energy management monitoring system (EMS) are becoming more and more powerful; at present, the energy management monitoring system has been widely used in residential and commercial buildings, hospitals, schools and other fields. However, most of the existing energy management and monitoring systems are designed to meet the load reduction needs of users with high power consumption, but they lack automatic monitoring and management mechanisms and key forecasting technologies, and cannot achieve the function of predicting power consumption in advance.

Therefore, generally speaking, when the power consumption of users exceeds the contracted capacity, most of them can only passively perform instant load shedding or unloading according to the alarm issued by the energy management monitoring system, which may cause managers to be unable to deal with the situation or impact of excessive power consumption in a timely manner. The comfort of air conditioning and so on.

According to an embodiment of the present disclosure, the present disclosure proposes an artificial intelligence manufacturing energy-saving management system, which includes a monitoring module and a forecasting module. The monitoring module collects historical electricity consumption data and historical temperature data of an area. The prediction module receives historical electricity consumption data, historical temperature data and temperature forecast data in this area. Among them, the prediction module executes the pre-processing program to pre-process the historical power consumption data and historical temperature data, and executes the training model based on the extreme gradient boosting algorithm to perform the temperature prediction data of a forecast period and all the previous forecast periods. The historical power consumption data and historical temperature data are trained to generate the power consumption prediction result of this forecast period, and the evaluation result is generated according to the evaluation index and the power consumption prediction result to decide to adopt the power consumption prediction result or re-establish the training model.

In one embodiment, the evaluation index is coefficient of determination.

In one embodiment, when the evaluation result is that the coefficient of determination is greater than 0.9, the prediction module decides to use the power consumption prediction result, and when the evaluation result is that the coefficient of determination is less than 0.9, the prediction module executes a model reconstruction procedure to re-establish the training model.

In one embodiment, the model reconstruction procedure includes adding variables, filling missing values, shortening the prediction period, and normalizing.

In one embodiment, the artificial intelligence production and manufacturing energy-saving management system further includes an analysis module. The analysis module compares the power consumption prediction result with the contracted capacity when the prediction module decides to use the power consumption prediction result, and compares the power consumption prediction result When the contracted capacity is exceeded, one or more of power consumption suggestions and warning signals will be generated.

According to another embodiment of this disclosure, this disclosure proposes an artificial intelligence production and manufacturing energy-saving management method, which includes the following steps: collecting historical power consumption data and historical temperature data of an area through a monitoring module; receiving historical power consumption through a forecasting module Data, historical temperature data and temperature forecast data in this area; the preprocessing program is executed by the forecasting module to preprocess the historical power consumption data and historical temperature data; the training model based on the extreme gradient boosting algorithm is executed by the forecasting module The temperature forecast data of a forecast period, all historical power consumption data and historical temperature data before this forecast period are trained to generate the power consumption forecast result of this forecast period; Evaluate the results to decide whether to use this power usage forecast or rebuild the trained model.

In one embodiment, the evaluation index is coefficient of determination.

In one embodiment, the step of using the prediction module to generate an evaluation result based on the evaluation index and the power consumption prediction result to decide to adopt the power consumption prediction result or to re-establish the training model further includes: the determination coefficient of the evaluation result is greater than 0.9 through the prediction module It is decided to adopt the power consumption prediction result; and through the prediction module, when the evaluation result is that the coefficient of determination is less than 0.9, the model reconstruction program is executed to re-establish the training model.

In one embodiment, the model reconstruction procedure includes adding variables, filling missing values, shortening the prediction period, and normalizing.

In one embodiment, the artificial intelligence production and manufacturing energy-saving management method further includes the following steps: comparing the power consumption prediction result with the contract capacity through the analysis module when the prediction module decides to adopt the power consumption prediction result; One or more of power consumption suggestions and warning signals are generated when the forecast result of power consumption in advance exceeds the contracted capacity.

Based on the above, the artificial intelligence manufacturing energy-saving management system and method thereof according to the present disclosure may have one or more of the following advantages:

(1) In one embodiment of the present disclosure, the artificial intelligence production and manufacturing energy-saving management system can generate power consumption forecast results, compare the power consumption forecast results with the contract capacity, and generate a warning signal when the power consumption forecast results exceed the contract capacity. Alert management personnel, so it is possible to save electricity and avoid additional electricity fines, thereby effectively implementing energy-saving management.

(2) In one embodiment of this disclosure, the artificial intelligence production and manufacturing energy-saving management system can generate power consumption forecast results, compare the power consumption forecast results with the contract capacity, and provide power consumption suggestions when the power consumption forecast results exceed the contract capacity , so that managers can take the most appropriate power-saving measures in real time, so as to implement energy-saving management more effectively.

(3) In one embodiment of this disclosure, the artificial intelligence production and manufacturing energy-saving management system adopts the limit gradient boosting algorithm, and uses the coefficient of determination as the evaluation index, which can more accurately evaluate whether the predicted power consumption result is close to the actual power consumption result , so it can effectively improve the efficiency of the artificial intelligence production and manufacturing energy-saving management system.

(4) In one embodiment of this disclosure, the training model of the artificial intelligence production and manufacturing energy-saving management system integrates the concept of extreme gradient boosting algorithm and time series. The prediction result is more accurate, so it can further improve the efficiency of the artificial intelligence manufacturing energy-saving management system.

(5) In one embodiment of this disclosure, the artificial intelligence production and manufacturing energy-saving management system can not only be applied to the production process or manufacturing process of various products, but also can be applied to electricity consumption prediction in various places, so it is widely used.

The following will refer to the relevant drawings to illustrate the embodiments of the artificial intelligence manufacturing energy-saving management system and its method according to the present disclosure. For the sake of clarity and convenience in the illustration, the dimensions and proportions of the components in the drawings may be changed. Exaggerated or reduced in size. In the following description and/or claims, when it is mentioned that an element is "connected" or "coupled" to another element, it may be directly connected or coupled to the other element or there may be an intervening element; When "directly connected" or "directly coupled" to another element, there are no intervening elements present, and other words used to describe the relationship between elements or layers should be interpreted in the same manner. To facilitate understanding, the same components in the following embodiments are described with the same symbols.

Please refer to FIG. 1, which is a block diagram of an artificial intelligence manufacturing energy-saving management system according to the first embodiment of the present disclosure. As shown in the figure, the artificial intelligence production and manufacturing energy-saving management system 1 includes a monitoring interface entrance 11 , a monitoring module 12 , a prediction module 13 and an analysis module 14 . In one embodiment, the artificial intelligence production and manufacturing energy-saving management system 1 can be a computer host, and the monitoring module 12 , prediction module 13 and analysis module 14 can be a chip integrating all functions or an independent chip.

The monitoring interface entrance 11 is connected with users in a region. Wherein, this area may be but not limited to factories, office buildings, shopping centers and schools.

The monitoring module 12 is connected with the monitoring interface entrance 11, and collects historical electricity consumption data H and historical temperature data T of this area through the monitoring interface entrance 11.

The prediction module 13 is connected with the monitoring module 12, and the monitoring module 12 receives the historical electricity consumption data H and the historical temperature data T, and the air temperature (outside air temperature) prediction data F of this area is obtained by the Meteorological Bureau. Wherein, the prediction module 13 can execute a pre-processing program to perform pre-processing on the historical power consumption data H and the historical temperature data T. The previous processing procedure may include one or more of actively identifying data column types, processing missing values, adjusting collinearity, feature screening, rolling self-regression, and normalization.

Next, the prediction module 13 executes a training model based on the extreme gradient boosting algorithm (XGbbost) to train the temperature prediction data F of a prediction period and all historical electricity consumption data H and historical temperature data T before this prediction period to generate The electricity consumption prediction result R of this prediction period; wherein, the electricity consumption prediction result R includes but is not limited to one or more of the electricity consumption unit, date, label of the electricity consumption monitoring device, and total power. In addition, the aforementioned forecast period can be adjusted according to actual needs; for example, the forecast period can be one week, 15 days, one month or two months, etc. In addition, the prediction module 13 can integrate historical power consumption data H, historical temperature data T and air temperature forecast data F at a preset time interval; for example, the prediction module 13 can (but not limited to) "every minute" as a preset Time interval integration of historical power consumption data H, historical temperature data T and temperature forecast data F. In addition, the forecasting module 13 can acquire the temperature forecast data F at a preset amount and frequency; for example, the forecasting module 13 can (but not limited to) capture two days at a time with an interval of three hours to obtain the temperature forecast data F.

Then, the prediction module 13 generates an evaluation result E according to the evaluation index and the electricity consumption prediction result R; in this embodiment, the evaluation index can be a coefficient of determination (coefficient of determination, R ² ). After the evaluation result E is generated, the prediction module 13 can decide to adopt the electricity consumption prediction result R or re-establish the training model according to the evaluation result E. The evaluation result E can effectively evaluate whether the training model can effectively explain the electricity consumption situation in this area. In this embodiment, when the evaluation result E has a coefficient of determination greater than 0.9, the prediction module 13 decides to adopt the power consumption prediction result R, and simultaneously stores the power consumption prediction result R, and then transmits the power consumption prediction result R to Analysis module 14.

Next, the analysis module 14 compares the power consumption prediction result R with the contracted capacity when the prediction module 13 decides to adopt the power consumption prediction result R. When the power consumption prediction result R exceeds the contracted capacity, the analysis module 14 generates a power consumption suggestion S and/or a warning signal W. Among them, the power consumption suggestion S can propose the most appropriate power-saving measures based on the difference between the power forecast result R and the contracted capacity, so that managers can maintain the normal operation of the facilities in the area as much as possible and reduce enough power consumption. The warning signal W can effectively remind managers that the power consumption in the forecast period may exceed the contracted capacity, so necessary measures must be taken as soon as possible.

On the contrary, when the evaluation result E is that the coefficient of determination is less than 0.9, the prediction module 13 executes a model rebuilding procedure to re-establish the training model. In one embodiment, the model reconstruction process includes adding variables, filling missing values, shortening the forecast period, and normalizing. When the evaluation result E is that the coefficient of determination is less than 0.9, it means that the training model cannot effectively explain the electricity consumption in this area, so the prediction module 13 can re-establish the training model in various ways, so that the training model can effectively explain the electricity consumption in this area situation. For example, the prediction module 13 can actively identify patterns when the data set is added later, or process abnormal or missing fields of the measurement points to improve the correctness and accuracy of the data. For example, the prediction module 13 can remove the variables whose correlation coefficient exceeds 0.7 in the collinear regression model, so as to avoid incorrect theoretical construction of the algorithm. For example, if the area is a school, "club activity events" (such as clubs borrowing classrooms) may be important variables affecting electricity consumption, so the prediction module 13 can add corresponding variables to the training model. Through the above method, the prediction module 13 can re-establish a training model more suitable for this area through the model reconstruction program. For example, if the area is a factory, "order quantity" may be an important variable affecting electricity consumption, so the prediction module 13 can also add the corresponding variable to the training model. In another embodiment, the evaluation index may also be mean square error (MSE), root mean square error (RMSE), or mean absolute error (MAE), etc. After the forecasting module 13 rebuilds the training model through the model rebuilding program, the forecasting module 13 will generate the power consumption prediction result R through the training model again, and generate the evaluation result E according to the evaluation index and the power consumption prediction result R, and according to the evaluation result E decides to adopt the power consumption prediction result R or re-establish the training model.

As mentioned above, the forecasting module 13 executes the training model based on the extreme gradient boosting algorithm (XGbbost) to train all the historical electricity consumption data H before the forecasting period to generate the power consumption forecasting result R of this forecasting period, so the forecasting module 13 The training model has integrated the extreme gradient boosting algorithm and the concept of time series. In other words, as time goes by, there will always be new actual electricity consumption, and this training model will take these actual electricity consumption into consideration in the model in chronological order, and only predict the value of the next forecast period each time; for example: use 2017 / After the actual power consumption in December is generated, use the historical data from 2012/2 to 2017/12 to predict the value in 2018/1. Experiments have proved that the training model that integrates the extreme gradient boosting algorithm and the concept of time series can better meet the needs of electricity consumption forecasting, and the results of electricity consumption forecasting are more accurate, so it can further improve the performance of the artificial intelligence manufacturing energy-saving management system 1 .

As mentioned above, the artificial intelligence production and manufacturing energy-saving management system 1 adopts the training model based on the extreme gradient boosting algorithm (Extreme Gradient Boosting, XGBoost), and uses the coefficient of determination (R ² ) as the evaluation index. The definition of the coefficient of determination represents the ratio of the variation value of the regression model to all yi variations. The larger the coefficient of determination, the greater the proportion of the regression model that can explain the overall yi variation. In this embodiment, whether the coefficient of determination is greater than 0.9 is used as an evaluation index, and it is proved by experiments that it can best meet the demand of power consumption forecast. The extreme gradient boosting algorithm is an iterative decision tree algorithm. It has been proved by experiments that the training model integrating the extreme gradient boosting algorithm and the concept of time series can best meet the needs of electricity consumption forecasting.

Through the above-mentioned mechanism, the artificial intelligence production and manufacturing energy-saving management system 1 can generate the power consumption prediction result R, compare the power consumption prediction result R with the contract capacity, and generate power consumption suggestions S and The warning signal W can save electricity and avoid additional electricity fines, and enable managers to take the most appropriate energy-saving measures in real time so as to effectively implement energy-saving management.

In addition, the artificial intelligence production and manufacturing energy-saving management system 1 can generate power consumption forecast results, compare the power consumption forecast results with the contract capacity, and provide power consumption suggestions when the power consumption forecast results exceed the contract capacity, so that managers can take the most effective measures in real time. Appropriate power-saving measures to perform energy-saving management more effectively.

Of course, the above is just an example, and the various components and their collaborative relationships of the artificial intelligence manufacturing energy-saving management system 1 can be changed according to actual needs, and this disclosure is not limited thereto.

It is worth mentioning that when the user's power consumption exceeds the contracted capacity, the existing energy management monitoring system can only send out an alarm to remind the management personnel to perform instant load shedding or unloading, which may cause the management personnel to be unable to deal with the excessive power consumption immediately. Or affect the comfort of the air conditioner and so on. On the contrary, according to the embodiments of the present disclosure, the artificial intelligence manufacturing energy-saving management system can generate power consumption forecast results, compare the power consumption forecast results with the contract capacity, and generate warning signals to warn when the power consumption forecast results exceed the contract capacity Managers can save electricity and avoid additional electricity fines, thereby effectively implementing energy-saving management.

In addition, according to the embodiments of the present disclosure, the artificial intelligence production and manufacturing energy-saving management system can generate power consumption forecast results, compare the power consumption forecast results with the contract capacity, and provide power consumption suggestions when the power consumption forecast results exceed the contract capacity. Managers can take the most appropriate power-saving measures in real time, thereby implementing energy-saving management more effectively.

In addition, according to the embodiment of the present disclosure, the training model of the artificial intelligence production and manufacturing energy-saving management system integrates the concept of extreme gradient boosting algorithm and time series. It is proved by experiments that this training model can better meet the needs of electricity consumption forecasting, and the use of electricity forecasting results It is more accurate, so it can further improve the efficiency of artificial intelligence manufacturing energy-saving management system. From the above, it can be seen that the artificial intelligence production and manufacturing energy-saving management system disclosed in this disclosure can indeed achieve excellent technical effects.

Please refer to FIG. 2 , which is a flow chart of the artificial intelligence manufacturing energy-saving management method of the first embodiment of the present disclosure. As shown in the figure, the artificial intelligence production and manufacturing energy-saving management method of this embodiment includes the following steps:

Step S21: Collect historical electricity consumption data and historical temperature data of an area through the monitoring module.

Step S22: Receive historical power consumption data, historical temperature data, and air temperature forecast data in this area through the forecasting module.

Step S23: Execute the pre-processing program with the forecasting module to pre-process the historical power consumption data and historical temperature data.

Step S24: Execute the training model based on the extreme gradient boosting algorithm through the forecasting module to train the temperature forecast data of a forecast period, all historical power consumption data and historical temperature data before this forecast period to generate a forecast period of electricity consumption result.

Step S25: Generate an evaluation result through the prediction module according to the evaluation index and the power consumption prediction result to decide to adopt the power consumption prediction result or re-establish the training model.

Please refer to FIG. 3 , which is a flow chart of the artificial intelligence manufacturing energy-saving management method of the second embodiment of the present disclosure. As shown in the figure, this embodiment illustrates more detailed steps of the method for energy-saving management of artificial intelligence production and manufacturing:

Step S31: Collect historical electricity consumption data and historical temperature data of an area through the monitoring module, and proceed to step S32.

Step S32: Receive historical electricity consumption data, historical temperature data, and air temperature forecast data in this area through the forecasting module, and proceed to step S33.

Step S33: Execute the preprocessing program with the forecasting module to preprocess the historical power consumption data and historical temperature data, and proceed to step S34.

Step S34: Execute the training model based on the extreme gradient boosting algorithm through the forecasting module to train the temperature forecast data of a forecast period, all historical power consumption data and historical temperature data before this forecast period to generate a forecast period of electricity consumption result, and go to step S35.

Step S36: Generate an evaluation result through the prediction module according to the evaluation index and the electricity consumption prediction result, and determine whether the evaluation result is a coefficient of determination greater than 0.9? If yes, proceed to step S361; if not, proceed to step S37.

Step S361: Execute the model reconstruction program through the prediction module to re-establish the training model, and return to step S34.

Step S37: Through the analysis module, it is judged whether the prediction result of the pre-consumption exceeds the contracted capacity? If yes, go to step S38; if not, go back to step S31.

Step S38: Generate power consumption suggestions and warning signals through the analysis module, and return to step S31.

Please refer to FIG. 4 , which is a diagram of the experimental results of the second embodiment of the present disclosure. In this embodiment, two different areas of Tamkang University are used as test areas (hereinafter referred to as area A and area B), and the data from 2020/11/23 to 2020/11/29 are used as training data for training. And generate the forecast results of electricity consumption with a forecast period of 2020/11/30-2020/12/06, and show the electricity consumption forecast results of 2020/11/30-2020/12/02 in Figure 4. As shown in the figure, curve A represents the actual power consumption of area A, curve A' represents the forecast result of power consumption in area A; curve B represents the actual power consumption of area B, and curve B' represents the forecast result of power consumption in area B . It can be seen from the figure that the predicted power consumption in region A is in line with the actual power consumption in region A, and the predicted power consumption in region B is also in line with the actual power consumption in region B (R ² is 0.984862342120521, greater than 0.9). From the experimental results, it can be seen that the training model of the artificial intelligence production and manufacturing energy-saving management system integrates the extreme gradient boosting algorithm and the concept of time series, which can better meet the needs of electricity consumption prediction, and the prediction results of electricity use are more accurate, so it can further improve artificial intelligence. Manufacturing efficiency of energy-saving management systems.

To sum up, according to the embodiments of this disclosure, the artificial intelligence manufacturing energy-saving management system can generate power consumption forecast results, compare the power consumption forecast results with the contract capacity, and generate warning signals when the power consumption forecast results exceed the contract capacity To warn management personnel, it can save electricity and avoid additional electricity fines, thereby effectively implementing energy-saving management.

In addition, according to the embodiments of the present disclosure, the artificial intelligence production and manufacturing energy-saving management system can generate power consumption forecast results, compare the power consumption forecast results with the contract capacity, and provide power consumption suggestions when the power consumption forecast results exceed the contract capacity, using Managers can take the most appropriate power-saving measures in real time, thereby implementing energy-saving management more effectively.

In addition, according to the embodiments of the present disclosure, the artificial intelligence production and manufacturing energy-saving management system adopts the limit gradient boosting algorithm, and uses the coefficient of determination as the evaluation index, which can more accurately evaluate whether the power consumption prediction result is close to the actual power consumption result, so It can effectively improve the efficiency of artificial intelligence production and manufacturing energy-saving management system.

In addition, according to the embodiment of the present disclosure, the training model of the artificial intelligence production and manufacturing energy-saving management system integrates the concept of extreme gradient boosting algorithm and time series. It is proved by experiments that this training model can better meet the needs of electricity consumption forecasting, and the use of electricity forecasting results It is more accurate, so it can further improve the efficiency of artificial intelligence manufacturing energy-saving management system.

Furthermore, according to the embodiments of the present disclosure, the artificial intelligence production and manufacturing energy-saving management system can not only be applied to the production process or manufacturing process of various products, but also can be applied to electricity consumption prediction in various places, so it is widely used.

It can be seen that this disclosure has indeed achieved the effect of the desired improvement under the breakthrough of the previous technology, and it is not easy for those who are familiar with the technology to think about it. Its progress and practicability obviously meet the requirements for patent application. ￠I filed a patent application in accordance with the law, and I sincerely ask your bureau to approve this invention patent application to encourage creation, and I am grateful for it.

The above descriptions are illustrative only, not restrictive. Any other equivalent modifications or changes made without departing from the spirit and scope of this disclosure shall be included in the scope of the appended patent application.

1: Artificial intelligence production and manufacturing energy-saving management system 11: Monitoring interface entrance 12: Monitoring module 13: Prediction module 14: Analysis module H: Historical electricity consumption data T: historical temperature data F: temperature forecast data R: Electricity prediction results E: Evaluation Results S: Electricity suggestion W: warning signal A, A’, B, B’: Curves S21~S25, S31~S38: step process

Fig. 1 is a block diagram of an artificial intelligence manufacturing energy-saving management system according to the first embodiment of the present disclosure.

FIG. 2 is a flow chart of the artificial intelligence manufacturing energy-saving management method of the first embodiment of the present disclosure.

FIG. 3 is a flow chart of the artificial intelligence manufacturing energy-saving management method of the second embodiment of the present disclosure.

Fig. 4 is a diagram showing the experimental results of the second embodiment of the present disclosure.

1: Artificial intelligence production and manufacturing energy-saving management system

11: Monitoring interface entrance

12: Monitoring module

13: Prediction module

14: Analysis module

H: Historical electricity consumption data

T: historical temperature data

F: temperature forecast data

R: Electricity prediction results

E: Evaluation Results

S: Electricity suggestion

W: warning signal

Claims (10)

An artificial intelligence production and manufacturing energy-saving management system includes: A monitoring module collects a historical electricity consumption data and a historical temperature data of an area; and A prediction module, which receives the historical electricity consumption data, the historical temperature data and a temperature prediction data of the area; Wherein, the forecasting module executes a preprocessing procedure to preprocess the historical electricity consumption data and the historical temperature data, and executes a training model based on a limit gradient boosting algorithm to the air temperature forecast data, All the historical power consumption data and the historical temperature data before the forecast period are trained to generate a power consumption forecast result for the forecast period, and an evaluation result is generated according to an evaluation index and the power consumption forecast result to decide to use the power consumption forecast. Electrically predict outcomes or rebuild the trained model. The artificial intelligence manufacturing energy-saving management system as described in Claim 1, wherein the evaluation index is a coefficient of determination. The artificial intelligence manufacturing energy-saving management system as described in Claim 2, wherein when the evaluation result is that the coefficient of determination is greater than 0.9, the prediction module decides to adopt the power consumption prediction result, and when the evaluation result is that the coefficient of determination is less than 0.9 , the forecasting module executes a model rebuilding procedure to recreate the training model. The artificial intelligence manufacturing energy-saving management system as described in Claim 3, wherein the model reconstruction procedure includes adding variables, supplementing missing values, shortening the forecast period, and normalizing. The artificial intelligence production and manufacturing energy-saving management system as described in Claim 1 further includes an analysis module, and the analysis module compares the power consumption prediction result with a contract capacity when the prediction module decides to adopt the power consumption prediction result , and generate one or more of an electricity consumption suggestion and a warning signal when the electricity consumption prediction result exceeds the contracted capacity. An artificial intelligence production and manufacturing energy-saving management method, including: Collecting a historical power consumption data and a historical temperature data of an area through a monitoring module; receiving the historical electricity consumption data, the historical temperature data and a temperature prediction data of the area through a forecasting module; Executing a pre-processing program with the forecasting module to pre-process the historical electricity consumption data and the historical temperature data; Executing a training model based on an extreme gradient boosting algorithm through the forecasting module to train the air temperature forecast data for a forecast period, all the historical electricity consumption data and the historical temperature data before the forecast period to generate the forecast period An electricity consumption forecast result of ; and Through the prediction module, an evaluation result is generated according to an evaluation index and the power consumption prediction result to decide to adopt the power consumption prediction result or to re-establish the training model. The method for energy-saving management of artificial intelligence production and manufacturing as described in Claim 6, wherein the evaluation index is a coefficient of determination. The artificial intelligence production and manufacturing energy-saving management method as described in claim item 7, wherein the evaluation result is generated by the prediction module based on the evaluation index and the power consumption prediction result to decide to adopt the power consumption prediction result or re-establish the training model The steps further include: When the evaluation result is that the coefficient of determination is greater than 0.9 through the forecasting module, it is decided to adopt the forecasted result of electricity consumption; and Through the prediction module, when the evaluation result is that the determination coefficient is less than 0.9, a model reconstruction procedure is executed to re-establish the training model. The artificial intelligence production and manufacturing energy-saving management method as described in Claim 8, wherein the model reconstruction procedure includes adding variables, supplementing missing values, shortening the forecast period, and normalizing. The artificial intelligence production and manufacturing energy-saving management method described in claim 6 further includes: comparing the electricity usage forecast with a contracted capacity through an analysis module when the forecasting module decides to use the electricity usage forecast; and One or more of an electricity consumption suggestion and a warning signal are generated when the pre-consumption prediction result exceeds the contract capacity through the analysis module.

Images