TWI815202B - Method and apparatus for determining efficiency influencing factors - Google Patents

Abstract

A method and an apparatus for determining efficiency influencing factors are provided. In the method, plural operation parameters and an operation efficiency of an equipment currently operating is retrieved and input as current operation information to a machine learning model created based on the plural operation parameters, so as to generate a predictive efficiency curve. The current operation information to be input to the machine learning model is adjusted according to a comparison between the predictive efficiency curve and a current efficiency curve generated based on the operation efficiency, so as to fit the predictive efficiency curve generated by the machine learning model to the current efficiency curve. At least one influencing factor that influences the operation efficiency of the equipment is determined from the operation parameters according to an adjustment ratio of each of the operation parameters in the current operation information.

Description

Efficiency influencing factor determination method and device

The present disclosure relates to a method and system for determining efficiency impact factors.

The quality of equipment efficiency affects equipment energy consumption and even directly affects the energy cost of the factory. Factors that affect equipment operating efficiency include changes in external conditions, such as weather, operating time, etc. or changes in control parameter settings, or even directly caused by equipment failure. Production efficiency is poor. The factory's electricity-saving technology achieves energy-saving effects by improving the efficiency of electrical equipment, reducing the power system to avoid downtime losses, and using energy management systems. On the other hand, the cost of machine failure is too high, and unexpected shutdowns require machine cleaning and adjustment, which will affect production capacity and equipment availability. What is more serious is the situation where the cause of the equipment abnormality cannot be determined, which requires a lot of time and labor costs to find the abnormal problem.

An embodiment of the present disclosure provides a method for determining an efficiency impact factor, which is suitable for determining the efficiency impact factor of equipment by an electronic device. This method includes the following steps: capturing multiple operating parameters and operating efficiency of the current operation of the equipment, inputting the captured multiple operating parameters as current operating information into a machine learning model established based on these operating parameters to generate a predicted efficiency curve, and Compare with the current efficiency curve generated based on operating efficiency to adjust the current operating information input to the machine learning model, so that the predicted efficiency curve generated by the machine learning model fits the current efficiency curve, and adjust each operating parameter according to the adjusted current operating information Proportion, determine at least one influencing factor that affects the operating efficiency of the equipment from these operating parameters.

An embodiment of the present disclosure provides an efficiency impact factor determination device, which includes a data acquisition device, a storage device and a processor. Among them, the data acquisition device is used to connect the equipment. The storage device is used to store the machine learning model established using multiple operating parameters of the equipment. The processor is coupled to the data acquisition device and the storage device, and is configured to use the data acquisition device to continuously acquire multiple operating parameters and operating efficiencies of the current operation of the equipment, and use the acquired multiple operating parameters as current operating information input based on A machine learning model is established with multiple operating parameters to generate a predicted efficiency curve, and compared with the current efficiency curve generated based on operating efficiency to adjust the current operating information input to the machine learning model, so that the predicted efficiency curve generated by the machine learning model fits the current efficiency curve, and based on the adjustment ratio of each operating parameter in the current operating information after adjustment, at least one influencing factor that affects the operating efficiency of the equipment is determined from multiple operating parameters.

In order to make the above features and advantages of the present disclosure more obvious and understandable, embodiments are given below and described in detail with reference to the attached drawings.

The integrated efficiency impact factor determination method and device of the embodiment of the present invention uses machine learning modeling technology to establish a correlation model between parameters and efficiency, and uses this model to estimate the impact of changes in equipment operating parameters on efficiency. Embodiments of the present invention further utilize cloud service technology to analyze the operation information of equipment of the same type and compare it with the operation information of local equipment to find parameters that affect efficiency. Thereby, when the equipment is operating at low efficiency, embodiments of the present invention can find the most relevant factors and provide modification suggestions or issue warnings.

FIG. 1 is a schematic diagram of a method for determining efficiency impact factors according to an embodiment of the present invention. Please refer to Figure 1. The method of this embodiment is, for example, to collect operation information 101 from the target device, including the parameter settings of the device itself, power consumption, flow, noise, temperature, humidity of the surrounding environment, etc. related to the operation of the device. The operating parameters and the operating efficiency of the equipment are obtained, and the captured operating information 101 is stored in the local database 102.

The efficiency model module 103 uses the operation information 101 in the local database 102 to establish a machine learning model related to operation parameters and efficiency. The efficiency calibration module 104 inputs the current operating parameters of the equipment into the machine learning model to generate a predicted efficiency curve, and compares it with the current efficiency curve of the equipment to detect whether the efficiency of the equipment is abnormal. If the efficiency calibration module 104 determines that the equipment efficiency is abnormal, the parameter detection module 105 will adjust the operation information input to the machine learning model so that the predicted efficiency curve generated by the machine learning model can fit the current efficiency curve. The parameter detection module 105 also determines the influencing factors 107 that affect the equipment's operating efficiency based on the adjustment ratio of each adjusted operating parameter.

On the other hand, the equipment operation information stored in the local database 102 can be uploaded to the cloud database 110, and the cloud server (not shown) collects and analyzes the operation information of multiple equipment of the same type in different fields, so as to Obtain decision tree data that can represent the performance characteristics of equipment of the same type. Accordingly, the parameter analysis module 106 obtains the decision tree data from the cloud database 110 on the one hand, and also obtains the operation information of the local device from the local database 102 and analyzes it to obtain parameters that can represent the performance characteristics of the local device. Decision tree data, and then compare the two decision tree data to determine the influencing factors that affect the equipment's operating efficiency107.

FIG. 2 is a block diagram of an efficiency impact factor determination device according to an embodiment of the present invention. Please refer to Figure 2. The efficiency impact factor determination device 10 of this embodiment is, for example, a personal computer, a server, a workstation or other devices with computing functions, including a data acquisition device 12, a storage device 14 and a processor 16. Its functions The breakdown is as follows:

The data acquisition device 12 is, for example, a universal serial bus (USB), RS232, a universal asynchronous receiver/transmitter (UART), an internal integrated circuit (I2C), a serial peripheral interface ( Wired connection devices such as serial peripheral interface (SPI), display port, thunderbolt or local area network (LAN) interface, or support wireless fidelity (Wi-Fi) ), RFID, Bluetooth, infrared, near-field communication (NFC) or device-to-device (D2D) and other communication protocols wireless connection devices. The data acquisition device 12 can be connected to the local device 20 to acquire the operation information of the device 20, and can be connected to a remote device to access the cloud database located on the remote device. There is no limitation here.

The storage device 14 is, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hardware disk or other similar device, or a combination of these devices, for storing programs executable by the processor 16. In some embodiments, the storage device 14 can store the above-mentioned local database 102, efficiency model module 103, efficiency verification module 104, parameter detection module 105 and parameter analysis module 106. In other embodiments, the storage device 14 can also store a machine learning model established using equipment operation information. The machine learning model is, for example, a convolutional neural network (CNN) or a recurrent neural network (recurrent neural network). network, RNN) or long short term memory (long short term memory, LSTM) recurrent neural network, this disclosure is not limited to this.

The processor 16 is coupled to the data acquisition device 12 and the storage device 14 to control the operation of the efficiency impact factor determination device 10 . In some embodiments, the processor 16 is, for example, a central processing unit (CPU), or other programmable general-purpose or special-purpose microprocessor (microprocessor), digital signal processor (digital signal processor) , DSP), programmable controller, application specific integrated circuit (ASIC), field programmable gate array (FPGA), programmable logic controller, PLC) or other similar devices or a combination of these devices, and can load and execute the program stored in the storage device 14 to execute the efficiency impact factor determination method of the embodiment of the present disclosure.

FIG. 3 is a flow chart of a method for determining efficiency impact factors according to an embodiment of the present disclosure. Please refer to FIG. 2 and FIG. 3 at the same time. The method of this embodiment is applicable to the efficiency impact factor determination device 10 of FIG. 2. The details of the efficiency impact factor determination method of the present disclosure are explained below with each component of the efficiency impact factor determination device 10. steps.

In step S302, the processor 16 of the efficiency influence factor determination device 10 uses the data acquisition device 12 to acquire multiple operating parameters and operating efficiency of the current operation of the equipment 20. The operating parameters include parameter settings related to equipment operation and the operating efficiency of the equipment. Taking the ice machine as an example, the operating parameters may include ice water inlet temperature, ice water outlet temperature, ice water flow rate, partial load ratio (PLR), cooling water inlet temperature, cooling water outlet temperature, cooling water Temperature difference, outdoor humidity, outdoor temperature, and specific energy consumption (KW/RT).

In step S304, the processor 16 inputs the captured operating parameters as current operating information into a machine learning model established based on these operating parameters to generate a predicted efficiency curve, and compares the predicted efficiency curve with the current efficiency generated based on the equipment operating efficiency. Curve comparison is used to adjust the current operating information input to the machine learning model so that the predicted efficiency curve generated by the machine learning model can fit the current efficiency curve.

In some embodiments, the processor 16 selects important parameters with the greatest impact from all operating parameters that can affect equipment efficiency and uses them to build a machine learning model. In detail, FIG. 4 is a flowchart of an operating parameter selection method according to an embodiment of the present disclosure. Please refer to FIG. 2 and FIG. 4 at the same time. The method of this embodiment is applicable to the efficiency impact factor determination device 10 of FIG. 2 .

In step S402, the processor 16 will retrieve a plurality of pieces of operation information about the operation of the equipment 20, including a plurality of candidate operation parameters related to the operation of the equipment 20. For example, the processor 16 uses the collected parameter data of the device 20 as a source data set, which includes sensing data, calculation data, input data, etc. of the device 20 . As shown in Table 1, the source data in the source data set may include operational information captured by the processor 16 using the data acquisition device 12 when the equipment is running, and efficiency information when the equipment is running. Operation information Efficiency information when equipment is running Value of ice water entering water temperature (°C) Efficiency value of ice water host (KW/RT) Ice water flow value (LPM) Cooling water temperature difference value (°C) The value of PLR Cooling water inlet temperature value outdoor humidity value Cooling water outlet temperature value The numerical value of the ice water outlet temperature outdoor temperature value Table 1

In step S404, the processor 16 calculates the contribution of each operation information to the operation efficiency of the equipment 20, and sorts the operation information according to the magnitude of the contribution. Among them, the processor 16 randomly extracts N pieces of operation information from the source data set as training data, and calculates the contribution of each piece of operation information to the target value (ie, the efficiency information of equipment operation) based on the Shapley value (SHAP Value). .

In step S406, the processor 16 predicts a target operating efficiency using multiple pieces of previously sorted operating information, and calculates the error between the target operating efficiency and the actual operating efficiency of the equipment. The processor 16 calculates, for example, the mean absolute percentage error (MAPE) between the estimated target value and the actual target value based on the N pieces of operation information with the largest contribution.

In step S408, the processor 16 selects a plurality of candidate operating parameters corresponding to the operating information whose calculated error is less than an error threshold as operating parameters for establishing the machine learning model. Among them, for the N pieces of operation information whose MAPE is less than the MAPE threshold, the processor 16 uses, for example, the candidate operation parameters corresponding to the N pieces of operation information as the operation parameters used to establish the machine learning model.

For example, FIG. 5 is an example of operating parameter selection according to an embodiment of the present disclosure. Please refer to Figure 5. For the candidate operating parameters "ice water inlet temperature, ice water flow rate, cooling water temperature difference, PLR, cooling water inlet temperature, outdoor humidity, cooling water outlet temperature, ice water outlet temperature, outdoor temperature", processor 16 By calculating the contribution and comparing the average absolute percentage error, it can be concluded that "ice water inlet temperature, ice water flow, cooling water temperature difference, PLR" are the main factors affecting the operating efficiency of the equipment, and these operating parameters are used to establish a machine learning model.

When the equipment 20 is actually running, the processor 16 can use the data retrieval device 12 to retrieve a plurality of current operation information corresponding to a plurality of operation parameters when the equipment 20 is currently running. For example, the processor 16 can use the data retrieval device 12 to retrieve the current operation information "value A (°C) of the ice water inlet temperature" corresponding to the operating parameter "ice water inlet temperature" when the equipment 20 is currently operating, and corresponding to The current operation information "ice water flow rate value B (LPM)" of the operation parameter "ice water flow rate" and the current operation information "cooling water temperature difference value C (°C)" corresponding to the operation parameter "cooling water temperature difference" are included. Then, the processor 16 inputs the current operating information into the machine learning model to generate a predicted efficiency curve. For example, the processor 16 can input the current operating information "the value of the ice water inlet temperature A (°C)", "the value of the ice water flow rate B (LPM)" and "the value of the cooling water temperature difference C (°C)" into the machine Learn the model to produce a predictive efficiency curve.

For example, FIG. 6 is a schematic diagram illustrating changes in equipment operating efficiency according to an embodiment of the present disclosure. Please refer to Figure 6. When the equipment is operating normally, its unit energy consumption is shown in the historical calculation data. Based on this historical calculation data, an efficiency prediction model M1 for predicting the efficiency curve can be established. As the equipment continues to operate, its unit energy consumption gradually becomes worse. As shown in the current actual value in the figure, the actual value of unit energy consumption has deviated from the historical value of normal operation.

Accordingly, the processor 16, for example, compares the predicted efficiency curve with a current efficiency curve of the current operation of the equipment, and adjusts the value of at least one current operation information of the current operation of the equipment based on the degree of deviation between the predicted efficiency curve and the current efficiency curve. (Maintaining other operating parameters among the plurality of operating parameters unchanged), thereby inputting the adjusted current operating information into the machine learning model. For example, the processor 16 calculates a standard score (Z-Score) between the predicted efficiency curve and the current efficiency curve as the degree of deviation. By adjusting the values back and forth and detecting changes in the efficiency curve, the predicted efficiency curve generated by the machine learning model can finally be fitted to the current efficiency curve.

In some embodiments, the processor 16 adjusts each operating parameter in the operating information, for example, including upward and downward adjustments, and selects the adjustment direction that is closest to the model prediction and the actual efficiency to find out the consistency. The characteristic curve of the real efficiency information, and the adjusted predicted efficiency curve can fit the current efficiency curve of the equipment, and based on this, the main causes of equipment inefficiency can be obtained.

For example, during the first adjustment, the processor 16 can adjust the value of one of the multiple pieces of current operation information (for example, change the value A (°C) of the current operation information "ice water inlet temperature" to adjusted upward to A+1 (°C)), and keep other current operating information unchanged, and input these adjusted current operating information into the machine learning model to obtain the predicted efficiency curve 1.

Then, during the second adjustment, the processor 16 can adjust the values of the two pieces of current operation information in the current operation information (for example, adjust the value A (°C) of the current operation information "ice water inlet temperature" upward to A+1(°C), and adjust the value B(LPM) of the current operation information "ice water flow" upward to B+3(LPM)), and keep other current operation information unchanged, and adjust these The final current operating information is input into the machine learning model to obtain the predicted efficiency curve 2.

By comparing the predicted efficiency curve with the current efficiency curve and repeating the above adjustment steps, the predicted efficiency curve can finally be fitted to the current efficiency curve.

For example, FIG. 7 is an example of fitting a predicted efficiency curve to a current efficiency curve according to an embodiment of the present disclosure. As shown in Figure 7, the unit energy consumption (KW/RT) of the real efficiency curve (i.e., the current efficiency curve) is higher than the predicted efficiency curve. At this time, the value of the operating information input to the machine learning model can be adjusted so that the adjusted predicted efficiency curve can be shifted upward from the original position and fit to the real efficiency curve.

Returning to the flow of FIG. 3 , in step S306 , the processor 16 will determine at least one influencing factor that affects the operating efficiency of the equipment 20 from multiple operating parameters based on the adjustment ratio of each operating parameter in the adjusted current operating information.

For example, in the above example, if the predicted efficiency curve 2 is more fitting (such as the trend of the curve) to the current efficiency curve than the predicted efficiency curve 1, and the adjustment ratio of "ice water flow" is higher in the second adjustment , at this time, the processor 16 can determine from multiple operating parameters that the influencing factor that affects the current operating efficiency of the equipment is "ice water flow".

In some embodiments, the processor 16 may also give suggestions for adjusting the operating parameters of the device 20 based on the determined influencing factors based on the adjustment data during the operating parameter adjustment process.

For example, Table 2 below is an example of suggestions for adjusting equipment operating parameters. Among them, by adjusting the operating parameters, the processor 16 can determine that the "ice water outlet temperature, ice water temperature difference, and cooling water temperature difference" are the influencing factors that cause the unit energy consumption of the equipment 20 to increase. Therefore, based on changing the unit energy consumption from the actual value ( 1) Adjusting to the target of the model estimate (0.6), the processor 16 can adjust the ice water outlet temperature from 8° to 12°, adjust the ice water temperature difference from 3° to 4°, and adjust the cooling water temperature difference from 2.8 ° adjustment is recommended to be 2.7°. Operating parameters actual value Model estimates Adjustment suggestions Unit energy consumption (KW/RT) 1 0.6 1->0.6 Ice water outlet temperature 8 10 8->12 Ice water temperature difference 3 3 3->4 Cooling water temperature difference 2.8 2.8 2.8->2.7 Ice water flow 900 900 900 fan frequency 60 60 60 Table 2

In some embodiments, in addition to collecting the operation information of the equipment 20 locally and establishing an efficiency estimation model to find out the efficiency influencing factors that cause the efficiency of the equipment 20 to decrease, the efficiency influencing factor determination device 10 can also be combined with a cloud database. , conduct performance analysis and comparison of equipment of the same type to find out other factors that cause the efficiency of equipment 20 to decrease.

Specifically, through cloud big data analysis, the cloud database can integrate the input characteristics and energy efficiency data of the same type of equipment in different fields, and perform node intersection comparisons through the cloud decision tree and the decision tree of a single device in the target field. In this way, we can find the operating point difference between the operating energy efficiency of this equipment and the same type of equipment, and obtain the operating characteristics that cause the energy efficiency difference between the same type of equipment, so as to give suggestions for adjusting the input characteristics.

FIG. 8 is a flow chart of a method for determining efficiency impact factors according to an embodiment of the present disclosure. Please refer to FIG. 2 and FIG. 8 at the same time. The method of this embodiment is applicable to the efficiency impact factor determination device 10 of FIG. 2. The details of the efficiency impact factor determination method of the present disclosure are explained below with each component of the efficiency impact factor determination device 10. steps.

In step S802, the processor 16 of the efficiency influencing factor determination device 10 selects multiple operating parameters as multiple nodes of the local decision tree, and analyzes the current operating information to obtain the best path of the local decision tree.

In step S804 , the processor 16 uses the data retrieval device 12 to retrieve the global decision tree obtained by the remote device analyzing the operation information of multiple devices of the same type as the device 20 and the optimal path therein.

In step S806 , the processor 16 calculates the intersection of the best path of the single-machine decision tree and the best path of the global decision tree, thereby determining the influencing factors that affect the operating efficiency of the device 20 .

For example, FIG. 9 is an example of an efficiency impact factor determination method according to an embodiment of the present disclosure. Please refer to Figure 9. This embodiment takes a local ice machine as an example. Assuming that its specification is 500RT, a single machine decision tree analysis can be performed through the above method, and multiple data of different fields of the same 500RT can be obtained from the cloud database. Decision tree analysis results of a piece of equipment (hereinafter referred to as 500RT). Among them, the best and worst paths of the above two decision trees are shown in Table 3 below. decision tree best path Decision tree worst path Local ice machine (Cooling water temperature difference＞2.785°) & (ice water inlet temperature＞9.82°), unit energy consumption=0.728 (cooling water temperature difference 3.645°)&(ice water entering water temperature 8.932°), unit energy consumption=0.83 500RT (PLR>0.608), unit energy consumption=0.676 (cooling water temperature difference 1.998°)&(PLR 0.348), unit energy consumption=0.991 table 3

It can be seen from Table 3 that in the decision trees of both the local ice machine and the 500RT, the "cooling water temperature difference" is the main basis for segmentation. In the optimal path of the decision tree of the local ice machine, the cooling water temperature difference is greater than 2.785° and unit The energy consumption is 0.728, and the optimal path result of the decision tree of the same type of ice machine obtained from the cloud database is that the partial load ratio (PLR) is greater than 0.608 and the unit energy consumption is 0.676. This efficiency is compared with the efficiency of the local ice machine. Higher, therefore, through the intersection comparison of the two sets of decision trees, adjustment suggestions for increasing or improving the local ice-server can be obtained, so that the partial load rate of the local ice-server can be increased to improve the operating performance of the local ice-server.

In some embodiments, processor 16 may perform decision tree analysis on device 20 at regular intervals (eg, one month). For example, if the processor 16 has determined that the operating parameters "ice water inlet temperature, ice water flow rate, cooling water temperature difference" are the main factors affecting the efficiency of the equipment 20 during operation, the processor 16 can execute the decision tree on the equipment 20 at regular intervals. Analysis to determine which one (or which) of the operating parameters "ice water inlet temperature, ice water flow rate, and cooling water temperature difference" is the efficiency influencing factor that affects the equipment efficiency during this period of time.

In summary, the efficiency impact factor determination method and device of the present disclosure use the historical operation information of the equipment to establish an efficiency model of the equipment to simulate the impact of changes in operating parameters on the equipment efficiency, and can identify important parameters that affect equipment performance and Recommendations for changes. In addition, through decision tree analysis and integration of local and cloud analysis results, the operating point differences between local equipment and equipment of the same type can be found, and the operating parameters that cause performance differences between equipment of the same type can be found. Based on this, we can provide Make suggestions for changes. Through the above methods, equipment managers can be assisted to identify factors causing equipment inefficiency and make appropriate decisions.

Although the disclosure has been disclosed above through embodiments, they are not intended to limit the disclosure. Anyone with ordinary knowledge in the technical field may make slight changes and modifications without departing from the spirit and scope of the disclosure. Therefore, The scope of protection of this disclosure shall be determined by the scope of the appended patent application.

101: Operation information 102:Local database 103:Efficiency model module 104:Efficiency test module 105: Parameter detection module 106: Parameter analysis module 107: Factors affecting operating efficiency 110:Cloud database 10: Efficiency influencing factor determination device 12:Data acquisition device 14:Storage device 16: Processor 20:Equipment S302~S306, S402~S408, S802~S806: steps

FIG. 1 is a schematic diagram of a method for determining efficiency impact factors according to an embodiment of the present invention. FIG. 2 is a block diagram of an efficiency impact factor determination device according to an embodiment of the present invention. FIG. 3 is a flow chart of a method for determining efficiency impact factors according to an embodiment of the present disclosure. FIG. 4 is a flowchart of an operating parameter selection method according to an embodiment of the present disclosure. FIG. 5 is an example of operating parameter selection according to an embodiment of the present disclosure. FIG. 6 is a schematic diagram illustrating changes in equipment operating efficiency according to an embodiment of the present disclosure. FIG. 7 is an example of fitting a predicted efficiency curve to a current efficiency curve according to an embodiment of the present disclosure. FIG. 8 is a flow chart of a method for determining efficiency impact factors according to an embodiment of the present disclosure. FIG. 9 is an example of a method for determining efficiency impact factors according to an embodiment of the present disclosure.

S302~S306: steps

Claims (6)

An efficiency impact factor determination method, suitable for determining the efficiency impact factor of an equipment by an electronic device, the method includes the following steps: capturing multiple operating parameters and operating efficiency of the current operation of the equipment; The plurality of operating parameters are input as current operating information into a machine learning model established based on the plurality of operating parameters to generate a predicted efficiency curve, and compared with a current efficiency curve generated based on the operating efficiency to adjust the input to the machine The current operating information of the learning model is such that the predicted efficiency curve generated by the machine learning model fits the current efficiency curve, wherein the predicted efficiency curve is generated and compared with the predicted efficiency curve generated based on the operating efficiency. The step of comparing current efficiency curves to adjust the current operating information input to the machine learning model includes: calculating a standard score (Z-Score) between the predicted efficiency curve and the current efficiency curve as the offset degree; adjust the value of at least one of the plurality of operating parameters in the current operating information, and maintain the other operating parameters of the plurality of operating parameters unchanged; and adjust the adjusted current operating parameter Information is input into the machine learning model, and the steps of generating the predicted efficiency curve and adjusting the current operating information are repeated until the predicted efficiency curve generated by the machine learning model fits the current efficiency curve; and According to the adjustment ratio of each of the plurality of operation parameters in the adjusted current operation information, at least one influencing factor that affects the operation efficiency of the equipment is determined from the plurality of operation parameters. The efficiency impact factor determination method as described in claim 1 further includes: acquiring multiple pieces of operation information on the operation of the equipment, the operation information including multiple candidate operation parameters related to the operation of the equipment; calculating each of the The contribution of the operation information to the operation efficiency of the equipment, and the operation information is sorted according to the magnitude of the contribution; a plurality of the operation information in the front are used to predict a target operation efficiency, and the target operation is calculated The error between the efficiency and the actual operating efficiency of the equipment; and selecting a plurality of candidate operating parameters corresponding to the operating information whose calculated error is less than an error threshold as the said machine learning model for establishing operating parameters. The efficiency influencing factor determination method as described in claim 1 further includes: selecting the plurality of operating parameters as multiple nodes of a local decision tree to analyze the current operation information to obtain the optimal result of the local decision tree. The best path; acquiring a global decision tree obtained by the remote device analyzing the operation information of multiple devices of the same type as the device and the best path in the global decision tree; and Calculate the intersection of the optimal path of the single-machine decision tree and the optimal path of the global decision tree to determine the influencing factor that affects the operating efficiency of the equipment. An efficiency impact factor determination device includes: a data acquisition device connected to a device; a storage device that stores a machine learning model established using multiple operating parameters of the device; and a processor coupled to the data acquisition The device and the storage device are configured to: use the data acquisition device to continuously acquire the multiple operating parameters and operating efficiencies of the current operation of the equipment; use the acquired multiple operating parameters as the current The operating information is input into the machine learning model established based on the plurality of operating parameters to generate a predicted efficiency curve, and compared with a current efficiency curve generated based on the operating efficiency to adjust the current input into the machine learning model. Operating information such that the predicted efficiency curve generated by the machine learning model fits the current efficiency curve, wherein the predicted efficiency curve is generated and compared with the current efficiency curve generated based on the operating efficiency to adjust The step of inputting the current operating information of the machine learning model includes: calculating a standard score (Z-Score) between the predicted efficiency curve and the current efficiency curve as the degree of deviation; Adjust the value of at least one of the plurality of operating parameters in the current operating information, and maintain the other operating parameters of the plurality of operating parameters unchanged; and input the adjusted current operating information The machine learning model, and repeat the steps of generating the predicted efficiency curve and adjusting the current operating information until the predicted efficiency curve generated by the machine learning model fits the current efficiency curve; and according to the adjusted The adjustment ratio of each of the plurality of operation parameters in the current operation information is used to determine at least one influencing factor that affects the operation efficiency of the equipment from the plurality of operation parameters. As for the efficiency impact factor determination device described in claim 4, the processor further uses the data acquisition device to acquire a plurality of pieces of operation information about the operation of the equipment, and the operation information includes a plurality of pieces of operation information related to the operation of the equipment. candidate operation parameters, calculate the contribution of each operation information to the operation efficiency of the equipment, sort the operation information according to the magnitude of the contribution, and predict a target using multiple pieces of the operation information sorted in front operating efficiency, and calculate the error between the target operating efficiency and the actual operating efficiency of the equipment, and select a plurality of candidate operating parameters corresponding to the operating information whose calculated error is less than an error threshold as the The operating parameters of the machine learning model are established. According to the efficiency influencing factor determination device of claim 4, the processor further selects the plurality of operating parameters as multiple nodes of a local decision tree to analyze the current operation information to obtain the local decision tree. the best path, Utilize the data acquisition device to acquire a global decision tree obtained by a remote device analyzing the operation information of multiple devices of the same type as the device and the best path in the global decision tree, and calculate the obtained The intersection of the best path of the single-machine decision tree and the best path of the global decision tree determines the influencing factor that affects the operating efficiency of the equipment.