A method for adjusting a smart home system

Abstract

The present invention discloses a method for adjusting a smart home system, the method comprising: arranging indoor sensors according to needs, collecting data from each sensor within a set time period, and collecting control information of an indoor air-conditioning system; obtaining outdoor meteorological information; establishing a regression tree machine learning model for temperature changes; after cleaning all data, inputting the data into a training model to obtain a prediction result about the temperature; inputting the training set into the model, and dynamically adjusting the temperature setting of the air-conditioner to maintain the indoor temperature within a set comfortable range.

Description

Technical Field

[0001] The technical field to which this technology belongs is the field of Internet of Things, machine learning and HVAC system control technology, and specifically relates to a method for adjusting a smart home system. Background Art

[0002] Traditional house air conditioning and HVAC systems are simply controlled according to a schedule, which causes the house to be sometimes too cold and sometimes too hot, and wastes electricity for air conditioning. This patent proposes to extract the temperature change pattern of the house through machine learning, so that the air conditioning can be better controlled according to the actual use of the house, so that the temperature distribution of the house is even and consistent, bringing the effect of physical comfort and energy saving. Summary of the invention

[0003] In order to solve the defects of the above-mentioned prior art, this patent proposes to extract the temperature change pattern of the house through machine learning, so that the air conditioner can be better controlled according to the actual use of the house, so that the temperature distribution of the house is even and consistent, bringing about physical comfort and energy-saving effects.

[0004] The above effects are specifically achieved through the following technical solutions:

[0005] A method for adjusting a smart home system, the method comprising:

[0006] Arrange indoor sensors according to needs, collect data from each sensor within a set time period, and collect control information of the indoor air-conditioning system; obtain outdoor weather information;

[0007] Build a regression tree machine learning model on temperature change;

[0008] After cleaning all the data, the data is input into the model to train the model and obtain the prediction result about the temperature;

[0009] The training set is input into the model, and the temperature setting of the air conditioner is dynamically adjusted to maintain the indoor temperature within the set comfortable range.

[0010] Furthermore, the control information of the air-conditioning system includes the switch status, mode setting, temperature setting, and wind speed setting information of the air-conditioning or air-conditioning indoor unit, and radiator equipment.

[0011] Furthermore, the sensor data includes indoor temperature, humidity, and carbon dioxide concentration;

[0012] The outdoor meteorological data includes temperature and weather conditions.

[0013] Furthermore, the model was validated to have a root mean square error of less than 1% - 5% of the average reading.

[0014] Furthermore, the establishment of a regression tree machine learning model for temperature change is specifically as follows:

[0015] The temperature reading of each sensor point is taken as the target in turn, and the temperature readings of the sensor and other sensors within the set distance at the previous time point, the temperature and wind speed setting of the air conditioner within the set distance from the sensor, and the outdoor weather data at the next time point are taken as features. A data sample set is obtained every set number of minutes within a week, and a machine learning model of regression tree for temperature changes is established;

[0016] The targets are classified according to the sample characteristics to obtain a regression tree model.

[0017] Furthermore, based on the regression tree model of temperature change, the initial sensor temperature reading of the day and the outdoor weather forecast data, the specific process of obtaining the temperature forecast value at the next time point is as follows:

[0018] 6.1 Perform the temperature and control scheme deduction at the preset time every day, input the temperature reading of the indoor sensor at the preset time and the predicted value of the outdoor temperature at the next time point by the weather forecast into the regression tree model, and obtain the temperature distribution of the sensor at the next time point; then use the deduced temperature distribution as a known feature and repeat step 6.1 until the air conditioning system is turned on;

[0019] 6.2 Predict the most energy-efficient air conditioning setting at the next test time and the expected sensor temperature reading under the air conditioning setting according to the following method:

[0020] If the day is in the cooling period, first predict whether the sensor reading will be higher than the set comfort temperature if the air conditioner is turned off. If not, complete the prediction at that time point and determine that the final air conditioner setting should be "off". If yes, adjust the air conditioner setting hypothesis to: "on", "cooling", "temperature setting: set degrees Celsius A";

[0021] Under this setting, it is determined again whether there is a sensor reading higher than the set comfort temperature. If not, the air conditioning temperature setting is adjusted to: increase one degree Celsius based on the last set temperature value A. It is determined whether there is a sensor reading higher than the set comfort temperature. If yes, the air conditioning temperature setting is adjusted to: decrease one degree Celsius based on the last set temperature.

[0022] Under this setting, it is determined again whether there will be a sensor reading higher than the set comfort temperature. If so, the air conditioner temperature setting is adjusted to: lower one degree Celsius based on the last set temperature value; if not, it is determined whether the last setting assumption will cause the sensor reading to be higher than the set comfort temperature; if not, the air conditioner temperature setting is adjusted to: increase one degree Celsius based on the last set temperature value; if so, the prediction for this time point is completed, and the final air conditioner setting is determined to be "on" or "cooling", and the temperature is set to the temperature setting in the final setting assumption;

[0023] If the day is in the heating period, first predict whether the sensor reading will be lower than the set comfort temperature if the air conditioner is turned off; if not, complete the prediction at that time point and determine that the final air conditioner setting should be "off"; if yes, adjust the air conditioner setting hypothesis to: "on", "heating", "temperature setting: set degrees Celsius";

[0024] Under this setting, it is determined again whether any sensor reading is lower than the set comfort temperature; if not, the air conditioning temperature setting is assumed to be adjusted to: reduce one degree Celsius based on the last set temperature value; if yes, the air conditioning temperature setting is assumed to be adjusted to: increase one degree Celsius based on the last set temperature value;

[0025] Under this setting, it is determined again whether there will be a sensor reading higher than the set comfort temperature; if so, the air conditioner temperature setting is adjusted to: increase one degree Celsius based on the last set temperature value; if not, it is determined whether the previous setting will cause the sensor reading to be lower than the set comfort temperature; if not, the air conditioner temperature setting is adjusted to: reduce one degree Celsius based on the last set temperature value; if so, the prediction for this time point is completed, and the final air conditioner setting is determined to be "on" and "heating", and the temperature is set to the temperature setting in the final setting assumption;

[0026] 6.3 Store the temperature reading derived in the previous step into the model, and then derive the air conditioning settings and temperature readings at the next test time until the end of the day.

[0027] Compared with the prior art, the present invention has the following advantages:

[0028] 1. The present invention can improve the temperature comfort of the house.

[0029] 2. The present invention can reduce the electricity consumption of air conditioning and heating systems, thereby reducing carbon emissions.

[0030] 3. The present invention enables building managers to understand indoor environmental data in fine granularity and implement remote control. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solution in the present invention, the drawings required for use in the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0032] Figure 1 is a logic flow chart of model adjustment in an embodiment of the present invention;

[0033] Figure 2 is a flowchart of the implementation of the user modification model in an embodiment of the present invention;

[0034] Figure 3 It is a schematic diagram of the overall implementation of the method of the present invention;

[0035] Figure 4 is a schematic diagram of the position of the indoor unit of the air conditioner in an embodiment of the present invention;

[0036] Figure 5 is a schematic diagram of the position of the rolling curtain in an embodiment of the present invention;

[0037] Figure 6 is a schematic diagram of the position of lamps in an embodiment of the present invention;

[0038] Figure 7 is a schematic diagram of sensor arrangement in an embodiment of the present invention;

[0039] Figure 8 is a schematic diagram of a strategy for intelligently controlling a house in an embodiment of the present invention;

[0040] Fig. 9 is a thermal diagram of temperature distribution before the house is controlled in an embodiment of the present invention;

[0041] Fig.10 It is a thermodynamic diagram of temperature distribution after the house is controlled in an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example

[0043] The temperature changes of a house are caused by the main heat and cold sources of the house. The main heat and cold sources in a house generally include: air conditioning and heating systems, sunlight heat gain through windows, the flow of people, and the switching of lighting systems and equipment.

[0044] The method for regulating the air conditioning and heating system and the lighting system is implemented in two stages.

[0045] Phase 1 - Building the model;

[0046] 1. First, deploy a sensor matrix in the room that needs to be controlled to collect environmental data from each area of ​​the room. To control the air conditioning and heating system, the data that needs to be collected include: temperature and carbon dioxide concentration. To control the lighting system, brightness data also needs to be collected.

[0047] 2. At the same time, it is necessary to collect outdoor meteorological data, including temperature and weather conditions.

[0048] 3. At the same time, it is necessary to collect the control data of the air conditioning and HVAC system, including but not limited to the switch status, mode setting, temperature setting, and wind speed setting of the air conditioner or air conditioner indoor unit, and radiator equipment.

[0049] 4. At the same time, it is necessary to collect the control data of the lighting system, including the on / off status of each group of lights. If the lights can adjust the brightness, the brightness settings also need to be recorded.

[0050] 5. Since the opening or covering of the roller shutter will affect both the HVAC system and the lighting system, it is also necessary to collect the position information of each roller shutter, that is, the roller shutter coverage percentage data.

[0051] 6. Collect the above data every 5 minutes, and collect samples continuously throughout the day for one week or more.

[0052] 7. In order to ensure that the system can fully predict the impact of the air conditioning and heating system, lighting system, and roller blind system on indoor temperature, humidity, and brightness, the system should be configured as much as possible in different combinations at this stage.

[0053] 8. Clean the collected data and use linear regression to complete the data missed by the sensor.

[0054] 9. Build a scalable boosted tree regression machine learning model for the data.

[0055] a. When building the temperature change model, the temperature reading of each sensor point is used as the "learning target" in turn, and the temperature and carbon dioxide concentration sensor readings within 5 meters of the point, and the corresponding setting information of the air conditioning or HVAC equipment corresponding to the air conditioning outlet within 10 meters, and the location data of the roller shutter equipment in the room in the previous 10 minutes, 20 minutes and 30 minutes are used as "features";

[0056] b. When building a brightness change model, the brightness reading of each sensor point is used as the "learning target" in turn, and the brightness sensor reading within 5 meters of the point, the lighting settings within 10 meters, and the location data of the roller shutter device in the room are used as "features".

[0057] 10. Validation of the Model The RMS error of the model is less than 1% - 5% of the mean reading.

[0058] Phase 2 - Deploy control logic and continuously improve the model;

[0059] 11. Based on the established model and after all control preparations are made, the control plan for the next day is exported on the day of preparation for control.

[0060] a. According to the established model, Figure 1 The process shown derives the most energy-efficient air conditioning setting for the next time point and the expected sensor temperature readings at that air conditioning setting.

[0061] b. Store the temperature reading derived in the previous step into the model, and then derive the air conditioning settings and temperature readings for the next time node until the end of the day.

[0062] c. Export the derived temperature change and air conditioning setting plan for the day and push it to the administrator.

[0063] 12. After receiving the control plan, the administrator can check and make changes to the air conditioning settings based on the control plan.

[0064] If a modification is proposed, the system recalculates the temperature distribution and air conditioning settings from that point in time until the end of the day.

[0065] 13. The implementation plan is carried out according to the time.

[0066] 14. During the execution, users in the room can adjust the air conditioning temperature setting through the curtain panel or other networked methods. If the user adjusts the air conditioning setting, Figure 8 The policy update system shown.

[0067] 15. At each time point of the day, the sensors and air conditioners upload corresponding data. Every night at 11:30, the system updates the model, generates a control plan for the next day, and pushes it to the administrator again.

[0068] In this embodiment, the heating period refers to the outdoor temperature below 16 degrees Celsius, and the cooling period refers to the outdoor temperature above 24 degrees Celsius; the time division of the heating period and the cooling period is set according to the conditions of the location and user preferences. Taking 30 degrees north latitude as an example, December to April is the heating period, and May to October is the cooling period.

[0069] 1. The present invention can improve the temperature comfort of the house.

[0070] 1) Traditional air conditioners rely on the temperature of one point on the air conditioner panel to determine the temperature in the room. Often, the air conditioner panel is installed in the same position, causing the temperature to be too low or too high during actual use. By deploying a sensor matrix, the present invention can ensure the correspondence between the area covered by the air conditioner and the feedback temperature.

[0071] 2) The present invention can adjust the air conditioning and heating system in advance to avoid overcooling or overheating in some areas by identifying the changing patterns of the heat source efficiency and temperature of the house.

[0072] 2. The present invention can reduce the electricity consumption of air conditioning and heating systems, thereby reducing carbon emissions.

[0073] 1) The present invention can estimate the possible temperature changes in the time after the setting according to the air conditioner settings through machine learning algorithms, so that the startup time can be minimized during the air conditioning cooling period, or the air conditioning temperature setting can be increased; during the air conditioning / heating period, the startup time can be minimized, or the temperature setting can be lowered. For air conditioning systems with variable frequency compressors, temperature adjustment will correspondingly reduce the use of the compressor, thereby reducing the power consumption of the entire system, thereby reducing carbon emissions.

[0074] 3. The present invention enables building managers to understand indoor environmental data in fine granularity and implement remote control.

[0075] 1) The present invention provides managers with the temperature distribution of the entire room through sensor matrix data, providing a data basis for the comfort of the indoor environment.

[0076] 2) The present invention provides a fine-grained remote control device function, so that the manager can remotely control the air conditioner without going to the room where the air conditioner is to be adjusted.

[0077] The above are only preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for adjusting a smart home system, It is characterized in that The method comprises: Arrange indoor sensors according to demand, collect data from each sensor within a set time period, and collect control information of the indoor air conditioning system. The specific process is as follows: 6.1 Perform the temperature and control scheme deduction at the preset time every day, input the temperature reading of the indoor sensor at the preset time and the predicted value of the outdoor temperature at the next time point by the weather forecast into the regression tree model, and obtain the temperature distribution of the sensor at the next time point; then use the deduced temperature distribution as a known feature and repeat step 6.1 until the air conditioning system is turned on; 6.2 Predict the most energy-efficient air conditioning setting at the next test time and the expected sensor temperature reading under the air conditioning setting according to the following method: If the day is in the cooling period, first predict whether the sensor reading will be higher than the set comfort temperature if the air conditioner is turned off. If not, complete the prediction at that time point and determine that the final air conditioner setting should be "off". If yes, adjust the air conditioner setting hypothesis to: "on", "cooling", "temperature setting: set degrees Celsius A"; Under this setting, it is determined again whether there is a sensor reading higher than the set comfort temperature. If not, the air conditioning temperature setting is adjusted to: increase one degree Celsius based on the last set temperature value A. It is determined whether there is a sensor reading higher than the set comfort temperature. If yes, the air conditioning temperature setting is adjusted to: decrease one degree Celsius based on the last set temperature. Under this setting, it is determined again whether there will be sensor readings higher than the set comfort temperature. If so, the air conditioner temperature setting is adjusted to: lower one degree Celsius based on the last set temperature value; if not, it is determined whether the last setting assumption will cause the sensor reading to be higher than the set comfort temperature; if not, the air conditioner temperature setting is adjusted to: increase one degree Celsius based on the last set temperature value; if so, the prediction for this time point is completed, and the final air conditioner setting is determined to be "on" and "cooling", and the temperature is set to the temperature setting in the final setting assumption; If the day is in the heating period, first predict whether the sensor reading will be lower than the set comfort temperature if the air conditioner is turned off; if not, complete the prediction at that time point and determine that the final air conditioner setting should be "off"; if yes, adjust the air conditioner setting hypothesis to: "on", "heating", "temperature setting: set degrees Celsius"; Under this setting, it is determined again whether any sensor reading is lower than the set comfort temperature; if not, the air conditioning temperature setting is assumed to be adjusted to: reduce one degree Celsius based on the last set temperature value; if yes, the air conditioning temperature setting is assumed to be adjusted to: increase one degree Celsius based on the last set temperature value; Under this setting, it is determined again whether there will be a sensor reading higher than the set comfort temperature; if so, the air conditioner temperature setting is adjusted to: increase one degree Celsius based on the temperature value set last time; if not, it is determined whether the previous setting will cause the sensor reading to be lower than the set comfort temperature; if not, the air conditioner temperature setting is adjusted to: reduce one degree Celsius based on the temperature value set last time; if so, the prediction for this time point is completed, and the final air conditioner setting is determined to be "on" and "heating", and the temperature is set to the temperature setting in the final setting assumption; 6.3 Store the temperature reading derived in the previous step into the model, and then derive the air conditioning settings and temperature readings at the next test time until the end of the day; obtain outdoor weather information; Establishing a regression tree machine learning model for temperature change; the establishment of a regression tree machine learning model for temperature change is specifically as follows: The temperature reading of each sensor point is taken as the target in turn, and the temperature readings of the sensor and other sensors within the set distance at the previous time point, the temperature and wind speed setting of the air conditioner within the set distance from the sensor, and the outdoor weather data at the next time point are taken as features. A data sample set is obtained every set number of minutes within a week, and a machine learning model of regression tree for temperature changes is established; Classify the target according to the sample characteristics and obtain the regression tree model; After cleaning all the data, the data is input into the model to train the model and obtain the prediction result about the temperature; The training set is input into the model to dynamically adjust the temperature setting of the air conditioner to maintain the indoor temperature within the set comfortable range.

2. A method for adjusting a smart home system according to claim 1, It is characterized in that The control information of the air conditioning system includes the switch status, mode setting, temperature setting, and wind speed setting information of the air conditioner or the indoor unit of the air conditioner and the radiator equipment.

3. A method for adjusting a smart home system according to claim 1, It is characterized in that The sensor data includes indoor temperature, humidity, and carbon dioxide concentration; The outdoor meteorological information includes temperature and weather conditions.

4. A method for adjusting a smart home system according to claim 1, It is characterized in that Validation Model The RMS error of the model was less than 1% - 5% of the mean reading.

Images