TWI831033B - Electricity usage scheduling system and electricity usage scheduling method - Google Patents

Abstract

A electricity usage scheduling system includes several machines, a smart box, and a process planning device. The several machines include several power load data and several temperature data. The smart box is connected to the several machines, and the smart box is configured to input several power load data and several temperature data into a neural network like model to generate power consumption prediction results. The process planning device is configured to generate power usage schedules for the several machines according to the power consumption prediction results.

Description

Electricity scheduling system and electricity scheduling method

The embodiments described in this disclosure relate to a power usage scheduling system and a power usage scheduling method, and in particular to a power usage scheduling system and power usage scheduling method for power consumption prediction.

In a factory area containing many machines, the power consumption of the machine equipment is extremely high. If the power load is accidentally exceeded, the contracted quota will be exceeded. In the past, we relied on manual monitoring, regular inspections or passively shutting down the power consumption of other equipment, but the efficiency needs to be improved. Traditionally, production schedules can only passively accept the impact of power overload. In this way, the production quality of machine equipment will also be affected.

Some embodiments of the present disclosure relate to a power scheduling system, including multiple machines, smart set-top boxes, and process planning devices. Multiple machines include multiple power load data and multiple temperature data. The smart set-top box is connected to multiple machines to input multiple power load data and multiple temperature data into the neural network model to generate power consumption prediction results. The process planning device is used to generate power consumption schedules for multiple machines based on the power consumption prediction results.

Some embodiments of the present disclosure relate to a power scheduling method, which includes the following steps. Multiple power load data and multiple temperature data of multiple machines are transmitted from multiple machines to the smart set-top box. The smart set-top box inputs multiple power load data and multiple temperature data into the neural network model to generate power consumption prediction results. The process planning device generates power consumption schedules for multiple machines based on the power consumption prediction results.

The term "coupling" used in this article may also refer to "electrical coupling", and the term "connection" may also refer to "electrical connection". "Coupling" and "connection" can also refer to the cooperation or interaction of two or more components with each other.

Refer to Figure 1. FIG. 1 is a schematic diagram of a power scheduling system 100 according to some embodiments of the present disclosure.

Taking the example in Figure 1 as an example, the power scheduling system 100 includes a plurality of machines 110A to 110C, a smart set-top box 130, a process planning device 150 and an ordering device 170. In terms of connection relationship, the smart set-top box 130 is connected to a plurality of machines 110A to 110C, the process planning device 150 is connected to the set-top box, and the ordering device 170 is connected to the process planning device 150 .

The power scheduling system 100 shown in Figure 1 is only for illustration, and the implementation of the present application is not limited thereto. For example, in some other embodiments, the process planning device 150 is connected to multiple smart set-top boxes, and the multiple smart set-top boxes are connected to multiple machines respectively. Various configurations of the power scheduling system 100 are within the scope of this disclosure.

Taking a dyeing and finishing factory as an example, the machines 110A to 110C in Figure 1 can be, respectively, a loom, a dyeing machine, a setting machine, a dehydration machine, a spreading machine, a cloth inspection machine, a cloth slitting machine, and a dyeing auxiliary automatic metering system. , air conditioning compressor, cloth rolling machine, waste water washing treatment device, steam boiler, single needle sewing machine, factory temperature control equipment, etc.

The detailed operation method of the power scheduling system 100 will be explained below with reference to Figure 2 .

FIG. 2 is a flowchart of a power scheduling method 200 according to some embodiments of the present disclosure. The power scheduling method 200 can be applied to the power scheduling system 100 shown in FIG. 1 . Please refer to Figure 1 to Figure 2 below.

In step S210, multiple power load data and multiple temperature data of multiple machines are transmitted to the smart set-top box. In some embodiments, step S210 is performed by the machines 110A to 110C as shown in FIG. 1 transmitting their respective power load data and temperature data to the smart set-top box 130 .

In some embodiments, the plurality of power load data includes an average load parameter and a peak load parameter, and the plurality of temperature data includes an average temperature parameter, a maximum temperature parameter, and a minimum temperature parameter.

In step S230, a plurality of electric load data and a plurality of temperature data are input into the neural network model to generate a power consumption prediction result. In some embodiments, step S230 is performed by the smart set-top box 130 as shown in Figure 1 .

See picture 3. FIG. 3 is a schematic diagram of a neural network model 300 according to some embodiments of the present disclosure. As shown in Figure 3, the neural network model 300 includes two hidden layers H1 and H2 and an output layer OP.

Specifically, the input layer includes a vector I composed of multiple electrical load data and multiple temperature data. Vector B11 is generated based on vector I, and vector B11 is calculated by function F11 of hidden layer H1 to generate vector L1. After the vector L1 is input to the hidden layer H2, the vector L2 is generated after being calculated by the function F12 of the hidden layer H2. Then, the vector B2 is generated according to the vector L1 and the vector L2, and the vector B2 is passed through the function F2 of the output layer OP to generate the power consumption prediction result S.

In some embodiments, the transfer functions of function F11 and function F12 are tangent double bending transfer functions, and the function F2 is a linear transfer function. In some embodiments, the number of nodes in the hidden layer H1 is 12, and the number of nodes in the hidden layer H2 is 12. In some embodiments, the network training method of the neural network model 300 is the LM algorithm (Lewenberg-Marquardt algorithm) combined with BasDavid Mackay's Bayesian structure, and the network training method is The maximum number of cycles is 6000.

In some embodiments, the output layer OP of the neural network model 300 generates a prediction model. The prediction model is estimated power consumption = first parameter × the average load parameter - second parameter × the maximum temperature parameter × all The peak load parameter - the third parameter × the average temperature parameter + the fourth parameter × the lowest temperature parameter × the average load parameter + the fifth parameter × (the peak load parameter) ^1/2 × all Describe the average load parameters.

In some embodiments, the first parameter, the second parameter, the third parameter, the fourth parameter and the fifth parameter are generated by the neural network model 300 .

In some embodiments, the electricity consumption prediction result S in Figure 3 is generated based on the prediction model to predict the electricity consumption in each time interval.

In some embodiments, 7.21≦first parameter≦8.81, 18.33≦second parameter≦22.40, 1.44≦third parameter≦1.76, 1.10≦fourth parameter≦1.34, 2.22≦fifth parameter≦2.72. In a preferred embodiment, the first parameter is 8.0136, the second parameter is 20.363, the third parameter is 1.6, the fourth parameter is 1.22, and the fifth parameter is 2.471.

Please refer back to Figure 2. In step S250, power consumption schedules for multiple machines are generated based on the power consumption prediction results. In some embodiments, step S250 is performed by the process planning device 150 as shown in FIG. 1 .

In some embodiments, step S250 further includes sending a plurality of customer data and a plurality of shipping data to the process planning device 150 from the order device 170 in Figure 1 . The process planning device 150 generates power consumption schedules for the plurality of machines 110A to 110C based on the power consumption prediction results, multiple customer data, and multiple shipment data.

See picture 4. FIG. 4 is a schematic diagram of a power schedule 400 according to some embodiments of the present disclosure. As shown in Figure 4, the power consumption schedule 400 includes the working time of the machine 110A, the machine 110B and the machine 110C in Figure 1 from 00:00 to 09:00 for processes 1 to 3. The power schedule 400 shown in Figure 4 is only for illustration, and the implementation of the present application is not limited thereto.

In some embodiments, the process planning device 150 can provide the optimized power schedule 400 for reference by process personnel to reduce the time of revising the power schedule 400 . In the power consumption schedule 400, the process planning device 150 can classify processes 1 to 3 according to the amount of power consumption. For example, the process planning device 150 can classify processes 1 to 3 into three colors: red, yellow, and green according to their high, medium, and low power consumption. In this way, the process personnel can adjust the order of processes 1 to 3 according to the classification of the power schedule 400 . Taking the process waiting time and shipping time into consideration, the process personnel can appropriately adjust the sequence or overlapping timing of processes 1 to 3 to avoid high power load situations and thereby improve the reliability of the power system or reduce the Related electricity bills. In some embodiments, the process planning device 150 can further record and analyze the adjusted power consumption schedule 400 to provide more accurate suggestions next time.

In some embodiments, the power scheduling system 100 and the power scheduling method 200 of this case are suitable for dyeing and finishing factories. However, the implementation of this case is not limited to this.

To sum up, this disclosure provides a power scheduling system and power scheduling method that collects and integrates the power load data and temperature data of multiple machines through a smart set-top box, and uses a neural network to The model analyzes and predicts the power consumption forecast results of multiple machines, and then the process planning device integrates the power consumption forecast results, customer information, and shipping data to generate power consumption schedules for multiple machines. After referring to the power consumption schedule, process personnel can more appropriately adjust the operation sequence or timing of multiple processes on multiple machines. In this way, power consumption overload when multiple machines are performing multiple processes at the same time can be effectively avoided, thereby controlling power consumption and electricity costs in the factory.

Various functional elements are disclosed herein. It will be apparent to one of ordinary skill in the art that functional elements may be implemented by circuitry, whether dedicated circuitry or general purpose circuitry operating under the control of one or more processors and coded instructions.

Although the disclosure has been disclosed in the above embodiments, it is not intended to limit the disclosure. Anyone with ordinary knowledge in the art can make various modifications and modifications without departing from the spirit and scope of the disclosure. Therefore, the disclosure The scope of protection shall be subject to the scope of the patent application attached.

100: Electricity scheduling system 110A, 110B, 110C: Machine 130:Smart set-top box 150:Process planning device 170:Order device 200: Electricity scheduling method S210, S230, S250: steps 300:Neural network model I: vector H1, H2: hidden layer OP: output layer B11,B2: vector L1, L2: vector F11, F12, F2: function S: Electricity consumption forecast results 400: Electricity schedule

In order to make the above and other objects, features, advantages and embodiments of the present disclosure more obvious and understandable, the accompanying drawings are described as follows: Figure 1 is a schematic diagram of a power scheduling system according to some embodiments of the present disclosure; Figure 2 is a flow chart of a power scheduling method according to some embodiments of the present disclosure; Figure 3 is a schematic diagram of a neural network model according to some embodiments of the present disclosure; and FIG. 4 is a schematic diagram of power schedule according to some embodiments of the present disclosure.

100: Electricity scheduling system

110A, 110B, 110C: Machine

130:Smart set-top box

150:Process planning device

170:Order device

Claims (8)

A power scheduling system includes: a plurality of machines, including a plurality of electric load data and a plurality of temperature data; a smart set-top box, connected to the plurality of machines, for connecting the plurality of electric loads The quantity data and the plurality of temperature data are input into the neural network model to generate power consumption prediction results; a process planning device; and an order device to transmit a plurality of customer data and a plurality of shipment data to the process. Planning device; wherein the process planning device is further used to generate the power consumption schedules for the plurality of machines based on the power consumption prediction results, the plurality of customer data and the plurality of shipment data. The power scheduling system according to claim 1, wherein the neural model includes two hidden layers and an output layer. The power scheduling system according to claim 1, wherein the plurality of power load data includes an average load parameter and a peak load parameter, and the plurality of temperature data includes an average temperature parameter, a maximum temperature parameter, and a minimum temperature parameter. temperature parameters. The power consumption scheduling system according to claim 3, wherein the neural-like model is used to generate a prediction model to generate the power consumption prediction result through the prediction model, and the prediction model is [estimated power consumption = The first parameter × the average load parameter - the second parameter × the maximum temperature parameter × the peak load parameter - the third parameter × the average temperature parameter + the fourth parameter × the minimum temperature parameter × the Average load parameter + fifth parameter × (the peak load parameter) ^1/2 × the average load parameter]. A power scheduling method includes: transmitting multiple power load data and multiple temperature data of multiple machines to a smart set-top box from multiple machines; using the smart set-top box to transmit the multiple power The load data and the plurality of temperature data are input into the neural network model to generate a power consumption prediction result; the ordering device transmits the plurality of customer data and the plurality of shipment data to the process planning device; and the process planning device The planning device generates the power consumption schedules for the plurality of machines based on the power consumption prediction results, the plurality of customer data, and the plurality of shipping data. The power consumption scheduling method according to claim 5, further comprising: generating a prediction model from the neural-like model to generate the power consumption prediction result through the prediction model. The power scheduling method as described in claim 6, wherein the prediction model is [estimated power consumption = first parameter × average load parameter - second parameter × maximum temperature parameter × peak load parameter - third parameter × average temperature parameter + fourth parameter × minimum temperature parameter × said average load parameter + fifth parameter × (peak load parameter) ^1/2 × average load parameter]. The power scheduling method as described in request item 7, wherein 7.21≦first parameter≦8.81, 18.33≦second parameter≦22.40, 1.44≦third parameter≦1.76, 1.10≦fourth parameter≦1.34, 2.22<fifth parameter Parameter ≦2.72.

Images