CN-121993433-A - Cooling tower fan energy-saving optimization method and system based on reinforcement learning

Abstract

The invention belongs to the technical field of cooling towers, and particularly relates to a cooling tower fan energy-saving optimization method and system based on reinforcement learning. The method comprises the steps of obtaining historical operation data of a cooling tower fan, processing the historical operation data to divide a time range covered by the historical operation data into at least one operation stage, obtaining target power of the operation stage for each operation stage, generating a structured compensation signal according to a current operation deviation value when the current operation power of the cooling tower fan is monitored to deviate from the target power, and adjusting the operation power of the cooling tower fan based on the structured compensation signal until the operation power is stable within a preset target power interval. According to the invention, the updated regulation and control reference path is generated based on the regulation and control result, so that the stability and reliability of regulation and control are enhanced, and energy conservation is realized on the premise of ensuring the cooling effect.

Inventors

• HOU YULONG
• ZHENG MIN

Assignees

• 丰城市天壕新能源有限公司

Dates

Publication Date

20260508

Application Date

20260316

Claims (10)

1. 1. The cooling tower fan energy-saving optimization method based on reinforcement learning is characterized by comprising the following steps of: When the current operation power of the cooling tower fan is monitored to deviate from the preset target power of the current operation stage, executing: Generating a structured compensation signal according to the current operation deviation value between the current operation power and the target power; Adjusting the running power of the cooling tower fan based on the structured compensation signal until the running power is stabilized in a preset target power interval; and before adjusting the operating power of the cooling tower fan, further executing the steps of acquiring historical operating data of the cooling tower fan, wherein the historical operating data comprises operating power data and heat exchange amount data corresponding to the operating power data, and processing the historical operating data to divide a time range covered by the historical operating data into at least one operating stage.
2. 2. The reinforcement learning-based cooling tower fan energy saving optimization method according to claim 1, wherein the step of processing the historical operation data to divide the time range covered by the historical operation data into at least one operation stage comprises clustering the historical operation data to determine inflection points, and dividing the at least one operation stage according to the inflection points.
3. 3. The reinforcement learning-based cooling tower fan energy saving optimization method of claim 2, wherein clustering historical operating data to determine inflection points comprises: The method comprises the steps of establishing a function relation taking heat exchange data as an independent variable and running power data as a dependent variable, calculating a second derivative of the function relation, and determining a point, in the second derivative, with an absolute value of a numerical value larger than a preset threshold value as an inflection point.
4. 4. The reinforcement learning-based cooling tower fan energy conservation optimization method of claim 1, wherein dividing at least one operation phase according to inflection points comprises: the method comprises the steps of establishing a function relation taking heat exchange data as an independent variable and running power data as an independent variable, calculating a first derivative of the function relation, extracting extreme points with the first derivative being zero, and respectively determining adjacent extreme points as a starting point and an ending point of an running stage.
5. 5. The reinforcement learning-based cooling tower fan energy conservation optimization method of claim 1, further comprising: And multiplying the running power average value by a preset amplitude adjusting coefficient to obtain the target power of the running stage.
6. 6. The reinforcement learning-based cooling tower fan energy conservation optimization method of claim 1, wherein the structured compensation signal comprises: a first compensation signal part calculated based on a reference parameter preset for the current operation stage; and a second compensation signal portion calculated from real-time feedback data of cooling tower fan operation.
7. 7. The reinforcement learning based cooling tower fan energy conservation optimization method of claim 1, wherein adjusting cooling tower fan operating power based on the structured compensation signal comprises: Dividing a power regulation interval of a current operation stage into a plurality of subintervals, and generating an updated regulation reference path based on a regulation result of a cooling tower fan in each subinterval; the structured compensation signal is adjusted based on the updated regulation reference path.
8. 8. The cooling tower fan energy-saving optimization system based on reinforcement learning is characterized by comprising the following modules: The operation phase dividing module is used for acquiring historical operation data of the cooling tower fan, dividing a time range covered by the historical operation data into at least one operation phase and determining target power of each operation phase; The operation power monitoring module is used for monitoring the current operation power of the cooling tower fan and determining the current operation deviation value between the current operation power and the target power determined by the operation phase dividing module; A compensation signal generation module configured to generate a structured compensation signal in response to the current operating bias value determined by the operating power monitoring module; The operation power regulation and control module is used for regulating the operation power of the cooling tower fan according to the structured compensation signal generated by the compensation signal generation module; And a regulation and control path updating module for monitoring a regulation and control result executed by the operation power regulation and control module to generate an updated regulation and control reference path, and feeding back the updated regulation and control reference path to the compensation signal generating module for adjusting the structured compensation signal.
9. 9. The reinforcement learning based cooling tower fan energy saving optimization system of claim 8, wherein the operation phase partitioning module is configured to cluster historical operation data to determine inflection points and partition at least one operation phase according to the inflection points; The operation phase division module is further configured to calculate an operation power average value in the historical operation data corresponding to each operation phase, and multiply the operation power average value with a preset amplitude adjustment coefficient to obtain a target power of the operation phase.
10. 10. The reinforcement learning based cooling tower fan energy conservation optimization system of claim 8, wherein the structured compensation signal comprises: a first compensation signal part calculated based on a reference parameter preset for the current operation stage; and a second compensation signal portion calculated from real-time feedback data of cooling tower fan operation.

Description

Cooling tower fan energy-saving optimization method and system based on reinforcement learning Technical Field The invention belongs to the technical field of cooling towers, and particularly relates to a cooling tower fan energy-saving optimization method and system based on reinforcement learning. Background The cooling tower fan is used as key power equipment in an industrial cooling system, and is characterized in that heat exchange between circulating cooling water and ambient air is completed through forced air convection, so that the temperature stability of a core production process is ensured. With the intelligent development of industry, the operation efficiency and the intelligent level of the cooling tower fan are required to be improved. In the existing energy-saving control technology of the cooling tower fan, the control strategy lacks self-adaptive capability to dynamic working conditions, the control method is mainly dependent on a preset operation curve or fixed temperature, pressure threshold and the like to carry out passive adjustment, complex system changes which are jointly formed by various factors such as environment temperature and humidity, atmospheric pressure, production load fluctuation and the like cannot be responded in real time, energy redundancy waste occurs in a low-load period, response delay occurs in the case of load sudden increase, and optimal cooling effect cannot be ensured. Based on the problems, the invention provides a cooling tower fan energy-saving optimization method and system based on reinforcement learning. Disclosure of Invention The invention aims to provide a cooling tower fan energy-saving optimization method and system based on reinforcement learning, which can set an independent power adjustment mechanism for each operation stage and ensure the adjustment precision of the fan operation state. The technical scheme adopted by the invention is as follows: a cooling tower fan energy-saving optimization method based on reinforcement learning comprises the following steps: When the current operation power of the cooling tower fan is monitored to deviate from the preset target power of the current operation stage, executing: Generating a structured compensation signal according to the current operation deviation value between the current operation power and the target power; Adjusting the running power of the cooling tower fan based on the structured compensation signal until the running power is stabilized in a preset target power interval; and before adjusting the operating power of the cooling tower fan, further executing the steps of acquiring historical operating data of the cooling tower fan, wherein the historical operating data comprises operating power data and heat exchange amount data corresponding to the operating power data, and processing the historical operating data to divide a time range covered by the historical operating data into at least one operating stage. Preferably, the processing of the historical operating data to divide the time range covered by the historical operating data into at least one operating phase includes clustering the historical operating data to determine inflection points and dividing the at least one operating phase according to the inflection points. Preferably, clustering the historical operating data to determine the inflection point includes: The method comprises the steps of establishing a function relation taking heat exchange data as an independent variable and running power data as a dependent variable, calculating a second derivative of the function relation, and determining a point, in the second derivative, with an absolute value of a numerical value larger than a preset threshold value as an inflection point. Preferably, the partitioning of the at least one operational phase in terms of inflection points comprises: the method comprises the steps of establishing a function relation taking heat exchange data as an independent variable and running power data as an independent variable, calculating a first derivative of the function relation, extracting extreme points with the first derivative being zero, and respectively determining adjacent extreme points as a starting point and an ending point of an running stage. Preferably, the method further comprises: And multiplying the running power average value by a preset amplitude adjusting coefficient to obtain the target power of the running stage. Preferably, the structured compensation signal comprises: a first compensation signal part calculated based on a reference parameter preset for the current operation stage; And calculating a second compensation signal part according to the real-time feedback data of the operation of the cooling tower fan. Preferably, adjusting the cooling tower fan operating power based on the structured compensation signal comprises: Dividing a power regulation interval of a current operation stage into a plurality of subinterva