Water purifier user behavior prediction and dynamic heating energy-saving control method based on AI

Abstract

The invention discloses an AI-based water purifier user behavior prediction and dynamic heating energy-saving control method, and relates to the field of water purifier control, and the method comprises the following steps: predicting the weighted water consumption of a corresponding single hour period, and further obtaining a water consumption prediction sequence of the next 24 hours; a basic temperature set point of the day is determined according to the relation between the total water consumption of the day and a statistical threshold value; and dynamically adjusting the heating mode based on the predicted water consumption and the actual water consumption. According to the method, the multi-dimensional water consumption behavior mode of the user is learned based on the AI algorithm, then the water consumption demand in the next 24 hours is predicted, a scientific basis is provided for dynamic heating, the temperature maintained by the water tank is dynamically selected according to the water consumption behavior so as to adapt to the current situation of energy waste caused by low water consumption in most scenes, energy consumption is optimized, and the energy consumption is reduced. And real intelligent and personalized energy-saving control is realized.

Description

Technical Field

[0001] This invention relates to the field of water purifier control, specifically to an AI-based method for predicting user behavior and dynamically controlling energy-saving heating in water purifiers. Background Technology

[0002] Traditional water purifier heating control methods mainly employ a fixed-temperature heating mode, where the water tank is always maintained at a set temperature (e.g., 100℃), ensuring the outlet water temperature through constant temperature control. While this method is simple and reliable, it has significant energy consumption issues: regardless of the user's water usage, the system needs to continuously heat to maintain a high temperature, resulting in substantial energy waste.

[0003] Improvements in existing technologies include timed heating control, which controls heating periods according to a preset schedule, and simple control based on water consumption detection, which activates heating when water consumption is detected (instantaneous heating). However, these methods still have the following problems: 1) Inability to predict users' actual water demand leads to inaccurate preheating; 2) Lacks the ability to learn from users' water usage behavior and cannot adapt to the usage habits of different users; 3) The impact of multiple factors such as water usage intensity, frequency, and duration on energy consumption was not considered; 4) It lacks a dynamic strategy selection mechanism and cannot select the optimal heating scheme based on actual energy consumption. Summary of the Invention

[0004] To address the aforementioned shortcomings in existing technologies, the AI-based user behavior prediction and dynamic heating energy-saving control method for water purifiers provided by this invention solves the problem that existing fixed-temperature heating modes or instant heating modes cannot learn from user water usage behavior, resulting in high energy consumption or difficulty in meeting large water usage demands.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: A method for predicting user behavior and dynamically controlling energy-saving heating in water purifiers based on AI is provided, which includes the following steps: A weight is introduced for each historical water consumption data point. Based on this weight and the corresponding historical water consumption data, the weighted water consumption for the corresponding single hour segment is predicted, thereby obtaining the water consumption prediction sequence for the next 24 hours. The baseline temperature setpoint for the day is determined based on the relationship between the total daily water consumption and the statistical threshold; the total daily water consumption is calculated based on the water consumption forecast sequence for the next 24 hours; and the statistical threshold is calculated based on the historical daily total water consumption over the past m days. The heating method is dynamically adjusted based on predicted and actual water consumption. If the actual water consumption in a certain hour is greater than the set multiple of the weighted water consumption for that hour, the base temperature set point for the day will be raised immediately. If the weighted water consumption for the remaining time is less than a set multiple of the total water consumption from today to the present, the base temperature set point will be marked as closed, i.e., active heating will be stopped for the day; the total water consumption from today to the present is obtained by summing the actual water consumption for the day and the weighted water consumption for the remaining time. When there is a period of time during which the predicted water consumption is 0, check whether there is a non-zero predicted water consumption in the subsequent period of that hour. If so, set the water tank temperature to the minimum temperature during the period of time during which the predicted water consumption is 0. Otherwise, mark the basic temperature set point as closed, that is, stop active heating for the day. When users use water, it is heated instantly.

[0006] Furthermore, the historical water consumption data includes the timestamp, volume, and hour of each water usage; the historical water consumption data is pre-processed data, and the pre-processing methods include using... The criteria are used for data cleaning; the expression for the cleaned historical water consumption data is as follows:

[0007] in This is historical water consumption data after cleaning; For the historical water usage data of the j-th hour, This represents the historical average water consumption for the h-th hour. is the standard deviation of historical water consumption in the h-th hour.

[0008] Furthermore, the expression for calculating the weight introduced for each historical water consumption data point is as follows:

[0009] in For the first i Weights of historical water consumption data points; It is a natural constant; The attenuation coefficient; The current time; For the first i The timestamp of each historical water consumption data point.

[0010] Furthermore, the formula for calculating the weighted water consumption for a single hourly period is:

[0011] For the first Hourly water consumption forecast; For the first Weighted average water consumption per hour; For the first The first hour segment i Historical water consumption data; For the first Total number of historical water consumption data points within an hourly period.

[0012] Furthermore, the specific methods for determining the daily baseline temperature setpoint based on the relationship between the total daily water consumption and the statistical threshold include: If the total water consumption for the day is greater than or equal to the statistical threshold, the basic temperature setpoint for the day will be set to 75℃; if the total water consumption for the day is less than the statistical threshold, the basic temperature setpoint for the day will be set to 70℃.

[0013] Furthermore, if the actual water consumption in a certain hour is greater than the weighted water consumption of that hour, the set multiple is 1.5.

[0014] Furthermore, if the weighted water consumption for the remaining time is less than the set multiple of the total water consumption from today to the present, the set multiple is 0.25.

[0015] Furthermore, the minimum temperature is 65℃.

[0016] Furthermore, after obtaining the water usage forecast sequence for the next 24 hours, the following operations are also included: Based on the historical water consumption standard deviation in hour h and the arithmetic average of historical water consumption in the hth hour Define and obtain the confidence level of water consumption prediction for hour h+1. Its expression is: ; Standard deviation of historical water usage frequency based on h-hour and the arithmetic mean of historical water usage times in the h-th hour Define and obtain the prediction confidence level of water usage frequency in hour h+1. Its expression is: ; Based on the standard deviation of historical water usage duration in hour h and the arithmetic mean of historical water usage duration in the hth hour Define and obtain the prediction confidence level of water usage frequency in hour h+1. Its expression is: ; Calculate the overall confidence level at hour h+1. Its expression is: ; right Compared with the confidence threshold, if If the water usage prediction sequence for the next 24 hours is greater than or equal to the first confidence threshold, then the sequence is retained; if... If the water usage forecast is less than the first confidence threshold but greater than or equal to the second confidence threshold, the probability of retaining the water usage prediction sequence for the next 24 hours is reduced; if If the water usage prediction sequence for the next 24 hours is less than the second confidence threshold, then discard the prediction sequence.

[0017] Furthermore, for days when water usage forecast sequences are not retained, the heating control method is preset by the user or set on-site.

[0018] The beneficial effects of this invention are as follows: Improved Prediction Accuracy and Decision-Making Dimensions: By introducing water consumption prediction, accurate quantitative prediction of future water use behavior is achieved, overcoming the limitation of existing technologies that can only perform qualitative scenario identification, and providing a more reliable data foundation for control decisions.

[0019] Enhanced dynamic adaptability: Based on the dynamic correction mechanism of daily predicted total water consumption and real-time water usage, this method can adaptively adjust the operation strategy, effectively cope with the daily fluctuations in user habits, and solve the problem of poor adaptability of fixed mode control.

[0020] Energy efficiency optimization is more refined: by establishing a precise mapping relationship between water intensity and heating strategy, and intelligently switching to energy-saving mode or turning off heating during predicted low-load periods, a paradigm shift from "continuous guarantee" to "on-demand supply" has been achieved, reducing unnecessary energy consumption.

[0021] Improved system robustness: Through confidence assessment mechanism and multi-level decision rules, the system can reasonably balance prediction risks and control strategies, achieve energy efficiency optimization while ensuring user experience, and improve the reliability of this method in practical applications. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the method. Detailed Implementation

[0023] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0024] like Figure 1 As shown, the AI-based water purifier user behavior prediction and dynamic heating energy-saving control method includes the following steps: S1. Introduce a weight to each historical water consumption data point, and predict the weighted water consumption for the corresponding single hour based on the weight and the corresponding historical water consumption data, thereby obtaining the water consumption prediction sequence for the next 24 hours. S2. Determine the base temperature setpoint for the day based on the relationship between the total daily water consumption and the statistical threshold. S3. Dynamically adjust the heating method based on predicted and actual water consumption: ① If the actual water consumption in a certain hour exceeds the set multiple of the weighted water consumption for that hour, the base temperature set point for the day will be immediately increased; the corresponding judgment expression is:

[0025] in To set a multiplier, you can set it to 1.5; ② If the weighted water consumption for the remaining time is less than a set multiple of the total water consumption from today to date, the base temperature setpoint will be marked as closed, i.e., active heating will stop for the day. The total water consumption from today to date is obtained by summing the actual water consumption for the day and the weighted water consumption for the remaining time, and the corresponding expression is:

[0026]

[0027] This represents the total water consumption from today to date. Weighted water consumption for the remaining time; To set the multiplier, you can set it to 0.25; ③ When there is a period of time when the predicted water consumption is 0, check whether there is a non-zero predicted water consumption in the subsequent period of that hour. If so, set the water tank temperature to the minimum temperature (in this embodiment, it can be set to 65℃ or 40℃) during the period when the predicted water consumption is 0. Otherwise, mark the basic temperature set point as closed, and the water tank temperature will cool down naturally and will no longer be actively heated until the early morning of the next day. Then, heating will be resumed based on the predicted start time of water consumption in the early morning of the next day, and active heating will be stopped for the day. When users use water, it is heated instantly.

[0028] In this embodiment, the collected raw water usage dataset (historical water usage data) is used. It can be represented as:

[0029] in For the first i The timestamp of the first water usage; For the first iThe volume of water used in a single application is expressed in liters. For the first i The hourly segment for each water usage is defined as starting at 0 and ending at 23 per day. This represents the total number of data points in the original water usage dataset.

[0030] To eliminate interference from anomalous data, the original water dataset... use The criteria are used to clean the historical water consumption data. The expression is:

[0031] in For the historical water usage data of the j-th hour, This represents the historical average water consumption for the h-th hour. , The total number of historical data points in the h-th hour; The standard deviation of historical water consumption in the h-th hour. .

[0032] In this embodiment, to reflect the time-varying nature of user habits, an exponentially decaying weight is introduced for each historical water consumption data point, and its calculation expression is as follows:

[0033] in For the first i Weights of historical water consumption data points; It is a natural constant; The attenuation coefficient is... ; The current time; For the first i The timestamp of each historical water consumption data point.

[0034] Based on this, the weighted average water consumption for a single hourly period is calculated. This value is the predicted water consumption for that hourly period, and the corresponding calculation expression is:

[0035] For the first Hourly water consumption forecast; For the first Weighted average water consumption per hour; For the first The first hour segment i Historical water consumption data; For the first Total number of historical water consumption data points within an hourly period.

[0036] Finally, the water usage forecast sequence for the next 24 hours can be output. :

[0037] in This indicates the predicted water consumption for a single hour.

[0038] In this embodiment, to quantify the reliability of the prediction, the following operation is further included after obtaining the water usage prediction sequence for the next 24 hours: Based on the historical water consumption standard deviation in hour h and the arithmetic average of historical water consumption in the hth hour Define and obtain the confidence level of water consumption prediction for hour h+1. Its expression is: ; Standard deviation of historical water usage frequency based on h-hour and the arithmetic mean of historical water usage times in the h-th hour Define and obtain the prediction confidence level of water usage frequency in hour h+1. Its expression is: ; Based on the standard deviation of historical water usage duration in hour h and the arithmetic mean of historical water usage duration in the hth hour Define and obtain the prediction confidence level of water usage frequency in hour h+1. Its expression is: ; Calculate the overall confidence level at hour h+1. Its expression is: ; right Compared with the confidence threshold, if If the water usage prediction sequence for the next 24 hours is greater than or equal to the first confidence threshold, then the sequence is retained; if... If the water usage forecast is less than the first confidence threshold but greater than or equal to the second confidence threshold, the probability of retaining the water usage prediction sequence for the next 24 hours is reduced; if If the water usage prediction sequence for the next 24 hours is less than the second confidence threshold, then discard the prediction sequence.

[0039] For days when water usage forecast sequences are not retained, the heating control method is preset by the user or set on-site. Configurable heating control methods include: maintaining 100°C or instantaneous heating starting from a set temperature (e.g., room temperature).

[0040] In one embodiment of the present invention, the total daily water consumption It is calculated based on the water usage forecast sequence for the next 24 hours, and its expression is: The statistical threshold is calculated based on the historical daily total water consumption over the most recent m days. The calculation expression is:

[0041]

[0042]

[0043] in The average daily water consumption (L) over the past m days. This represents the water consumption on day j within the recent m days. Let be the standard deviation of daily water consumption (L) over the recent m days. This is a statistical threshold; This is the standard deviation multiplier, which defaults to 1 and is a dimensionless quantity.

[0044] In practical implementation, this method determines the basic temperature setpoint for the day based on the relationship between the total daily water consumption and the statistical threshold, including the following specific methods: If the total water consumption for the day is greater than or equal to the statistical threshold, the basic temperature setpoint for the day will be set to 75℃; if the total water consumption for the day is less than the statistical threshold, the basic temperature setpoint for the day will be set to 70℃.

[0045] In this embodiment, the data acquisition and calculation in steps S1 and S2 can both be processed by AI. Based on the specific temperature values ​​given in this embodiment, the heating status of the water purifier is determined. This can be summarized as follows:

[0046] ON indicates heating; OFF indicates no heating. This indicates that the base temperature setpoint is marked as off; This refers to the current actual water temperature; This is the current base temperature setpoint.

[0047] Correspondingly, when At that time, the energy consumption for water tank reheating can be expressed as:

[0048] in Energy consumption for replenishing the water tank temperature (kWh); This is the specific heat capacity of water; This refers to the volume of the water tank. Indicates the positive operator; This refers to the heating efficiency of the water tank.

[0049] The energy consumption for instant heating in the water tank is:

[0050] in Energy consumption (kWh) for replenishing heat to the water tank via instant heating. To predict water consumption (L); The target outlet water temperature set by the user has a default value of 100℃. The effective temperature (°C) of the water inlet of the instant heating module in the water tank. but ,like but This is equal to the water temperature in the tank after natural cooling. The instant heating efficiency of the water tank instant heating module.

[0051] In one embodiment of the present invention, the following comparative experiment was conducted to verify the energy-saving effect of the method: Experimental group: Busy hours are set from 9:00 AM to 6:00 PM. Off-peak hours are other time periods. During busy hours, the water tank temperature is maintained at 75 degrees Celsius, with instant heating at 25 degrees Celsius. During off-peak hours, the temperature is maintained at 40 degrees Celsius. User behavior calculations are not used; only busy and off-peak time periods are set.

[0052] Control group (the present invention): No busy time setting is required. Based on the dynamic calculation of water usage data, the temperature is maintained at 75 degrees when it is necessary to maintain the water tank temperature, which is equivalent to heating by 25 degrees.

[0053] The outlet water temperature is 100 degrees Celsius, and the ambient temperature is 25 degrees Celsius. Both waters start heating from an initial temperature of 25 degrees Celsius.

[0054] The energy consumption data of the control group and the experimental group under high-intensity water use scenarios (water use is 0 at other times) are shown in Table 1.

[0055] Table 1

[0056] It can be seen that in a high-intensity water use environment, the energy consumption of the experimental group and the control group is the same.

[0057] In low-intensity water use scenarios (water consumption is 0 at other times), the energy consumption data of the control group and the experimental group are shown in Table 2.

[0058] Table 2

[0059] It can be seen that in a low-intensity water use environment, the control group (this method) saved 12.6% of energy consumption compared to the experimental group.

[0060] The following conclusions can also be drawn from the above controlled experiment: 1. With a water tank temperature of 75 degrees Celsius, the instant heating temperature is 25 degrees Celsius. It takes approximately 17 seconds to fill a 500ml cup with water. With a water tank temperature of 100 degrees Celsius, no heating is required, and it takes approximately 17 seconds to fill a 500ml cup with water.

[0061] 2. A water tank temperature of 75 degrees Celsius, with an instant heating mode of 25 degrees Celsius, is significantly more energy-efficient than a water tank temperature of 100 degrees Celsius, because it does not require heating the entire tank of water. At the same time, the user experience is the same (it takes about 17 seconds to fill a cup of water).

[0062] 3. The water tank temperature is 70 degrees Celsius, with an instant heating temperature of 25 degrees Celsius. It takes approximately 20 seconds to fill a 500ml cup. When water usage is low, users will hardly notice this 3-second delay. However, if water usage is frequent and many people need to wait often, this 3-second delay will be more noticeable. Therefore, in scenarios with low water usage, the water tank temperature should be lowered.

[0063] 4. The lower the water tank temperature, the lower the heat loss. A 75-degree water tank will drop about 3 degrees in 1.5 hours, while a 70-degree water tank will only drop 2.5-2.7 degrees in 1.5 hours. That is, the heat loss of maintaining a water tank at 70 degrees is about 10-20% lower than that at 75 degrees.

[0064] In summary, this invention uses AI algorithms to learn users' multi-dimensional water usage behavior patterns, thereby predicting water demand for the next 24 hours. This provides a scientific basis for dynamic heating, dynamically selecting the water tank temperature based on water usage behavior to adapt to the current situation where low water consumption leads to energy waste in most scenarios, optimizing energy consumption and achieving truly intelligent and personalized energy-saving control.

Claims

1. A method for predicting user behavior and dynamically controlling energy-saving heating in water purifiers based on AI, characterized in that, Includes the following steps: A weight is introduced for each historical water consumption data point. Based on this weight and the corresponding historical water consumption data, the weighted water consumption for the corresponding single hour segment is predicted, thereby obtaining the water consumption prediction sequence for the next 24 hours. The base temperature setpoint for the day is determined based on the relationship between the total daily water consumption and the statistical threshold; the total daily water consumption is calculated based on the water consumption forecast sequence for the next 24 hours; the statistical threshold is calculated based on the historical daily total water consumption over the past m days. The heating method is dynamically adjusted based on predicted and actual water consumption. If the actual water consumption in a certain hour is greater than the set multiple of the weighted water consumption for that hour, the base temperature set point for the day will be raised immediately. If the weighted water consumption for the remaining time is less than the set multiple of the total water consumption from today to the present, the base temperature set point will be marked as closed, i.e., active heating will be stopped for the day. The total water consumption from the current day to the present is obtained by summing the actual water consumption of the day and the weighted water consumption for the remaining time. When there is a period of time during which the predicted water consumption is 0, check whether there is a non-zero predicted water consumption in the subsequent period of that hour. If so, set the water tank temperature to the minimum temperature during the period of time during which the predicted water consumption is 0. Otherwise, mark the basic temperature set point as closed, that is, stop active heating for the day. When users use water, it is heated instantly.

2. The AI-based water purifier user behavior prediction and dynamic heating energy-saving control method according to claim 1, characterized in that: Historical water consumption data includes the timestamp, volume, and hour of each water usage; the historical water consumption data is pre-processed data, and the pre-processing methods include... The criteria are used for data cleaning; the expression for the cleaned historical water consumption data is as follows: in This is historical water consumption data after cleaning; For the historical water usage data of the j-th hour, This represents the historical average water consumption for the h-th hour. is the standard deviation of historical water consumption in the h-th hour.

3. The AI-based water purifier user behavior prediction and dynamic heating energy-saving control method according to claim 1, characterized in that, The formula for calculating the weight introduced for each historical water consumption data point is as follows: in For the first i The weight of each historical water consumption data point; It is a natural constant; The attenuation coefficient; The current time; For the first i The timestamp of each historical water consumption data point.

4. The AI-based water purifier user behavior prediction and dynamic heating energy-saving control method according to claim 3, characterized in that, The formula for calculating the weighted water consumption for a single hour is: For the first Hourly water consumption forecast; For the first Weighted average water consumption per hour; For the first The first in the hour segment i Historical water consumption data; For the first Total number of historical water consumption data points within an hourly period.

5. The AI-based water purifier user behavior prediction and dynamic heating energy-saving control method according to claim 1, characterized in that, The specific methods for determining the daily baseline temperature setpoint based on the relationship between the total daily water consumption and the statistical threshold include: If the total water consumption for the day is greater than or equal to the statistical threshold, the basic temperature setpoint for the day will be set to 75℃; if the total water consumption for the day is less than the statistical threshold, the basic temperature setpoint for the day will be set to 70℃.

6. The AI-based user behavior prediction and dynamic heating energy-saving control method for water purifiers according to claim 1, characterized in that, If the actual water consumption in a certain hour is greater than the weighted water consumption for that hour, the set multiple is 1.

5.

7. The AI-based water purifier user behavior prediction and dynamic heating energy-saving control method according to claim 1, characterized in that, If the weighted water consumption for the remaining time is less than the set multiple of the total water consumption from today to the present, the set multiple is 0.

25.

8. The AI-based water purifier user behavior prediction and dynamic heating energy-saving control method according to claim 1, characterized in that, The minimum temperature is 65℃.

9. The AI-based water purifier user behavior prediction and dynamic heating energy-saving control method according to claim 1, characterized in that, After obtaining the water usage forecast sequence for the next 24 hours, the following operations are also included: Based on the historical water consumption standard deviation in hour h and the arithmetic average of historical water consumption in the hth hour Define and obtain the confidence level of water consumption prediction for hour h+1. Its expression is: ； Standard deviation of historical water usage frequency based on h-hour and the arithmetic mean of historical water usage times in the h-th hour Define and obtain the prediction confidence level of water usage frequency in hour h+1. Its expression is: ； Based on the standard deviation of historical water usage duration in hour h and the arithmetic mean of historical water usage duration in the hth hour Define and obtain the prediction confidence level of water usage frequency in hour h+1. Its expression is: ； Calculate the overall confidence level at hour h+1. Its expression is: ； right Compared with the confidence threshold, if If the water usage prediction sequence for the next 24 hours is greater than or equal to the first confidence threshold, then the sequence is retained; if... If the water usage forecast is less than the first confidence threshold but greater than or equal to the second confidence threshold, the probability of retaining the water usage prediction sequence for the next 24 hours is reduced; if If the water usage prediction sequence for the next 24 hours is less than the second confidence threshold, then discard the prediction sequence.

10. The AI-based water purifier user behavior prediction and dynamic heating energy-saving control method according to claim 9, characterized in that, For days when water usage forecast sequences are not retained, the heating control method is set by the user in advance or on-site.

Images