WO2017008321A1

WO2017008321A1 - Smart home energy management method based on smart wearable device behavior detection

Info

Publication number: WO2017008321A1
Application number: PCT/CN2015/084679
Authority: WO
Inventors: 刘烃; 沈超; 季建廷; 陈思运; 车煜林; 徐占伯; 管晓宏
Original assignee: 西安交通大学
Priority date: 2015-07-10
Filing date: 2015-07-21
Publication date: 2017-01-19
Also published as: CN105045234A; CN105045234B

Abstract

Disclosed is a smart home energy management method based on smart wearable device behavior detection, comprising: S101) collecting sensor data and external environment data of a wearable device; S102) preprocessing the data; S103) determining a sensor type; S104) detecting a motion behavior of a user; S105) detecting a physical state of the user; S106) estimating a living environment demand of the user; S107) updating and determining a user demand; S108) optimizing a home appliance control policy in an event-driven manner; and S109) updating a control policy of a smart home control module. The present invention estimates and updates, by monitoring various sensors on a wearable device of a user, demands of the user for current and future living environments, and accordingly generates a scheduling policy for various smart home appliance devices and actively controls the home appliance devices, thereby meeting comfort requirements of the user on the home environment, and reducing the electricity costs.

Description

Intelligent home energy management method based on behavioral awareness of smart wearable devices

[Technical Field]

The invention relates to the field of intelligent home optimization control, in particular to a smart home energy management method based on behavioral awareness of a smart wearable device.

【Background technique】

The optimized control of a personalized smart home depends on the user's behavior and demand perception. Accurate user behavior-aware results can lead to reliable demand-aware input for smart home optimization control. Reliable demand-aware input helps optimize the control system to customize the optimal control strategy that is more user-friendly. In the end, the smart home optimization control system can satisfy the user comfort to the greatest extent and help users save electricity. The location and motion information of the smart wearable device can be used to analyze the current activity behavior of the user; the analysis and processing result of the user activity behavior can be used to clarify the user requirements; the acquisition of the user requirement can be used in the smart home with the environmental parameter data of the user family. Optimize control to maximize user satisfaction.

Traditional smart home control is unified and monotonous. For users with different behaviors and needs, almost the same scheduling control strategy is adopted. However, users are often personalized, have their own unique lifestyle and household energy use, and a unified scheduling strategy can not meet the user's comfort and economic expectations to the greatest extent.

[Summary of the Invention]

It is an object of the present invention to provide a smart home energy management method based on behavioral awareness of smart wearable devices to solve the above technical problems. The invention collects sensor data of the user intelligent wearable device in real time; performs user behavior perception; collects external environment parameters in real time, estimates user life environment requirements; determines whether user needs are updated according to the user social behavior perception result; if user demand is triggered Update, execute smart home optimization scheduling algorithm, generate control strategy; update smart home control module control strategy.

In order to achieve the above object, the present invention adopts the following technical solutions:

A smart home energy management method based on behavioral awareness of smart wearable devices includes the following steps:

S101), sensor data of the wearable device and external environment data collection: collecting data of the motion sensor, the biosensor, and the environmental sensor in the smart wearable device carried by the user; and collecting the weather, traffic, and electricity price information of the current location of the user; The sensor acquisition data and the external environment data are stored in an environment database for predicting a comfortable target desired by the user for the environment; the data collected by the motion sensor and the biosensor is transferred to step S102);

S102), sensor data preprocessing of the wearable device: preprocessing formats the data, including time synchronization and data source classification, wherein the time synchronization refers to performing sensor data acquired on different wearable devices according to the local time of the system. Time synchronization processing; data source classification is to classify and identify data according to wearable devices, collected sensors and data structure attributes of the data source;

S103) sensor type determination: according to the type of sensor data, for the data collected by the motion sensor, proceeds to step S104); for the data collected by the biosensor, proceeds to step S105);

S104) User behavioral awareness: the user's behavior includes behavioral and behavioral states; the behavioral context refers to the environment in which the user's athletic behavior occurs; combined with the location information base, using the user's location information acquired by GPS, and implementing the behavioral context according to the location information base Estimation; behavioral state refers to the type of motion behavior currently being performed by the user; by analyzing the data acquired by the acceleration sensor, the gyroscope, the geomagnetic sensor or the electronic compass sensor, the behavior state classifier is used to realize the perception of the behavior state;

S105) The user's physical state perception: estimating the current state of the user's body according to the combination of different biometrics according to the current heart rate, blood pressure, body surface temperature, and blood glucose data collected by the biosensor;

S106), user living environment demand estimation: user living environment demand refers to the user's expectation of temperature, humidity, illuminance, hot water, diet in the current and future environment, and the corresponding time; according to step S104) and step S105) Perceived results and environmental data acquired in step S101), combined with pre-built user motion behavior - user body state - external A correspondence table between the environment and the user's living environment requirements, estimating the user's living environment requirements, and dynamically updating the demand correspondence table in the form of feedback according to the actual behavior of the user in the later stage;

S107), the user demand update judgment: comparing step S106) obtaining the latest estimated user living environment demand, comparing with the previous estimation result, determining whether the demand changes; if no change occurs, proceeding to step S101) to continue collecting data; If a change occurs, proceed to step S108);

S108), event-driven electrical equipment control strategy optimization: using event-driven online optimization method, combined with the environmental data collected in step S101), to meet the user's living environment requirements and electrical equipment constraints, to save the entire household electricity costs Minimize the target and generate a joint operational control strategy for various devices;

S109), updating the control policy of the smart home control module according to the result of step S108), and proceeding to step S101) to continue monitoring the wearable device sensor data of the user.

Further improvements of the present invention include: a motion sensor including a GPS, an acceleration sensor, a gyroscope, a geomagnetic sensor, and an electronic compass sensor; the biosensor includes a blood glucose sensor, a blood pressure sensor, an electrocardiogram sensor, a myoelectric sensor, a body temperature sensor, a brain wave sensor; The sensor includes a temperature and humidity sensor, an atmospheric pressure sensor, a gas sensor, a pH sensor, an ultraviolet sensor, an ambient light sensor, a particulate matter sensor, a barometric pressure sensor, and a microphone.

A further improvement of the present invention is that the formatting of the wearable device sensor data in step S102) includes time synchronization processing and data source classification; the specific method of time synchronization processing is: uniformly uploading the collected data to the server, discarding the collected data. Local timestamp, according to the server time, the time field of the data is marked with a uniform timestamp; the data source classification is as follows: the source field is added to the attribute list of the acquired data, and the source of the data is filled according to the source point of the data. Attribute field.

A further improvement of the present invention is that the location information base in step S104) specifically includes: location information of a local sensitive building, the sensitive building includes an office place, a sports place, a shopping place, a restaurant, a home building; a behavior state classifier and a body state classifier. The training method is: using Adaboost to train user behavior based on historical multidimensional sensor data vectors. State classifier and body state classifier.

A further improvement of the present invention is that the method for constructing and updating the correspondence table of the user's exercise behavior-user physical state-external environment and the user's living environment requirement in step S106) is: according to the user historical behavior data statistics, respectively, the user's exercise behavior-life The demand correspondence table, the user's physical state-life demand correspondence table, the external environment-life demand correspondence table, according to the prior expert knowledge and statistical rules, the three tables are given equal weights, and the three sub-tables are weighted and combined, that is, the user behavior is obtained. The life needs, the life needs of the user's physical state and the life needs of the external environment are linearly superimposed to obtain the final life demand, and then the initialized user's motor behavior - body state - external environment and life demand correspondence table; when there is user's When historically operating the behavior data, the weights of the three tables are adjusted according to the user's past operational behavior, the user's physical state, and the correlation coefficient of the external environment; during the use process, the user's manual operation behavior and evaluation information of the home environment are recorded. Analyze the user's home ring The level of satisfaction of the environment, continuously adjust the weight of the three tables, and constantly update the user life needs correspondence table;

The method for estimating the user's living environment requirement in step S106 is specifically: according to the user's behavioral situation and behavioral state, mapping based on the requirement rule base, obtaining an environment requirement corresponding to the latest user behavior; and the demand time is performed according to the GPS information and the road condition information. Estimate; the user requirements rule base is empirically built based on defined user behavior.

A further improvement of the present invention is that the update policy for the existing user requirements in step S107) includes: comparing the latest user requirements with the current user requirements; if the demand periods do not overlap, increasing the corresponding user requirements; if the demand period occurs Overlap will replace current user needs with the latest user requirements.

A further improvement of the present invention is: in step S108), the change of the user requirement is taken as the event e, including the event detection time T _d , the event type A, the user requirement R; wherein the event detection time T _d refers to the time when the event is detected; The event type A refers to the user behavior description corresponding to the event; the user requirement R is a set of time series corresponding to each user requirement, that is, R={r _T (k), r _H (k), r _L (k), r _W (k) )}, where r _T (k), r _H (k), r _L (k), r _W (k) represent the temperature, humidity, lighting and hot water demand of the user at time k, respectively, defined when the user does not have time k When required, the corresponding demand variable r(k)=0.

A further improvement of the present invention is that the optimization problem in step S108) is aimed at minimizing the total electricity cost of the home appliance operation running in the optimized scheduling period, that is,

Where t ₀ represents the initial time of the optimization period, t _d represents the end time of the optimization period, N is the number of scheduled electrical devices, p(k) represents the price of electricity at time k, and q _i (k) represents the time of device i at time k The power is the decision variable.

A further improvement of the present invention is that the constraints of the optimization problem in step S108) include three types: a power consumption model of the electrical equipment, an operational constraint of the electrical equipment, and a comfort requirement constraint; wherein the power consumption model of the electrical equipment generally adopts a state equation. Establish the connection between the electrical power of the equipment and the user demand it provides; the operational constraints of the electrical equipment refer to the physical limitations during the operation of the electrical equipment, usually including the maximum power constraints of the electrical equipment, the operational safety constraints of the electrical equipment; the comfort requirement constraints refer to the user's room environment Variables need to meet current comfort needs.

A further improvement of the present invention is that the optimization problem in step S108) is driven by an event, the specific process is: when the event e is detected at the time T _d , the user requirement R contained therein is read; and the length L of the time window is defined, at the T _The optimization problem is solved based on the user demand R during the period from _d to T _d +L; the obtained optimization strategy is used to control the electrical equipment until a new event is detected, and the drive optimization problem is recalculated to obtain a new strategy.

Compared with the prior art, the present invention has the following beneficial effects: the present invention monitors the user's motion behavior by using motion sensor data by monitoring various types of sensors on the user wearable device, and uses the biosensor to recognize the user's physical state and combines the external environmental information. Estimate and update the user's demand for current and future living environment; adopt event-driven electrical equipment control strategy optimization method, and formulate scheduling strategies for various types of electrical equipment in smart home based on updated user living environment requirements, environmental status and dynamic electricity price information And actively control household appliances to meet the user's comfort requirements for the home environment, while reducing the cost of electricity.

[Description of the Drawings]

FIG. 1 is a block diagram of a smart home energy management method based on behavioral awareness of smart wearable devices.

【Detailed ways】

Please refer to FIG. 1 , which is a block diagram of a smart home energy management method based on behavioral awareness of a smart wearable device according to the present invention; and a basic framework of a smart home energy management method based on behavioral awareness of the smart wearable device is shown.

The invention provides a smart home energy management method based on behavioral awareness of a smart wearable device, comprising the following steps:

S101), sensor data of the wearable device and external environment data collection: the smart wearable device defined by the invention includes: a smart bracelet, a smart phone, a smart watch, a smart eye, and the like, which can be integrated on the user's body or can be carried by the user. The device data of the wearable device is divided into three categories: (a) motion sensor, including GPS, acceleration sensor, gyroscope, geomagnetic sensor, electronic compass sensor; (b) biosensor, including blood glucose sensor, blood pressure sensor, heart Electric sensor, myoelectric sensor, body temperature sensor, brain wave sensor; (c) environmental sensor, including temperature and humidity sensor, atmospheric pressure sensor, gas sensor, pH sensor, ultraviolet sensor, ambient light sensor, particulate sensor or dust sensor, air pressure sensor, microphone. The external environment data defined by the present invention is the weather, traffic, and electricity price information of the user's current location, and such information is actively acquired through the weather, traffic, and power grid websites of the Internet. The environmental sensor acquisition data and the external environment data are stored in the environment database for predicting a comfortable target desired by the user to the environment; the data collected by the motion sensor and the biosensor is transferred to step S102).

S102), sensor data preprocessing of the wearable device: preprocessing formats the data, including time synchronization and data source classification, wherein the time synchronization refers to performing sensor data acquired on different wearable devices according to the local time of the system. Time synchronization processing; data source classification is to classify and identify data based on wearable devices, collected sensors, and data structure attributes of the data source.

S103) Sensor type determination: according to the type of the sensor data, the data collected by the motion sensor is transferred to step S104); for the data collected by the biosensor, the process proceeds to step S105).

S104) User motion behavior perception: The user motion behavior defined by the present invention includes a behavioral situation and a behavioral state. The behavioral scenario refers to the environment in which the user's sports behavior occurs, including housing, office, shopping mall, highway, sports venue, restaurant, etc.; combined with the location information base, the user location information obtained by GPS is used to estimate the behavioral scenario according to the location information database. The behavior state refers to the type of exercise behavior currently being performed by the user, including walking, exercising, regular stillness, sleeping, etc.; by analyzing the data acquired by the acceleration sensor, the gyroscope, the geomagnetic sensor, and the electronic compass sensor, using the behavior state classifier to realize Perception of behavioral state.

S105) User body state perception: according to the user's current heart rate, blood pressure, body surface temperature, blood glucose data collected by the biosensor, according to the combination of different biometrics, the current state of the user's body is estimated, and the state includes: normal, excited, fatigue, cold, hot .

S106), user living environment demand estimation: The user living environment requirement defined by the present invention refers to the user's expectation of temperature, humidity, illuminance, hot water, diet in the current and future environment, and the corresponding time. According to the sensing result of step S104) and step S105) and the environmental data acquired in step S101), combined with the pre-built user motion behavior-user physical state-correspondence table of the external environment and the user's living environment demand, the user's living environment demand is estimated. According to the actual behavior of the user in the later stage, the demand correspondence table is dynamically updated in the form of feedback.

S107), user demand update judgment. The comparison step S106) obtains the latest estimated user living environment requirement, and compares with the previous estimation result to determine whether the demand changes. If no change has occurred, the process proceeds to step S101) to continue collecting data; if a change has occurred, the process proceeds to step S108).

S108), Event-Driven Electrical Equipment Control Strategy Optimization: The event defined by the present invention refers to a change in the user's living environment requirements. Using the event-driven online optimization method, combined with the environmental data collected in step S101), under the constraints of the user's living environment requirements and electrical equipment, to save the entire household electricity cost to minimize the goal, to generate a combination of various devices Run the control strategy.

S109), the control strategy of the control module is updated according to the result of step S108), and the process proceeds to step S101) to continue monitoring the wearable device sensor data of the user.

The formatting of the wearable device sensor data in step S102) of the present invention mainly includes time synchronization processing and data source classification. The specific method of time synchronization processing is: uploading the collected data to the server uniformly, due to the locality of each system. The time may be different, so the local timestamp of the collected data is discarded, and the time field of the data is uniformly time stamped according to the server time. The data source classification is as follows: the source field is added to the attribute list of the acquired data, and the source attribute field of the data is filled according to the source point of the data.

The location information library in step S104) of the present invention specifically includes: location information of a local sensitive building, and the sensitive building includes an office place, a sports place, a shopping place, a restaurant, and a home building. The training method of the behavior state classifier and the body state classifier is: based on the historical multidimensional sensor data vector (including motion sensor, biosensor, environmental sensor data vector), Adaboost is used to train the user behavior state classifier and the body state classifier.

The method for constructing and updating the correspondence table of the user's exercise behavior-user body state-external environment and the user's living environment requirement in step S106) is: according to the user historical behavior data statistics, respectively obtaining the user's exercise behavior-life demand correspondence table, the user Body state - life demand correspondence table, external environment - life demand correspondence table, according to prior knowledge and statistical rules, three tables are given equal weights, three child tables are weighted and combined, that is, the user needs to get the life needs, users The life needs obtained by the physical state and the life needs obtained by the external environment are linearly superimposed to obtain the final life demand, and then the initialized user's exercise behavior-physical state-external environment and life demand correspondence table is present; when the user's historical operational behavior data exists According to the user's past operational behavior, the user's physical state and the correlation coefficient of the external environment, the weights of the three tables are adjusted; during the use process, the user's manual operation behavior and evaluation information of the home environment are recorded, and the user is analyzed for the home environment. Satisfaction, continuous Adjust the weight of the three tables and continuously update the user life requirements correspondence table.

The method for estimating the user's living environment requirement in step S106) of the present invention is specifically: according to the user's behavioral situation and behavioral state, mapping based on the requirement rule base, obtaining an environment requirement corresponding to the latest user behavior; and the demand time is based on GPS information and road conditions. Information is estimated. The user requirement rule base is empirically established according to the defined user behavior. For example, the behavioral scenario is motion, and the corresponding air conditioning temperature demand will be lowered during the demand period; the behavioral scenario is at home, and the behavior state is sleep, and the corresponding air conditioning demand temperature Will rise.

The update strategy for the existing user requirements in step S107) of the present invention includes: comparing the latest user requirements with the current User requirements; if the demand periods do not overlap, the corresponding user requirements are increased; if the demand periods overlap, the current user requirements are replaced with the latest user requirements.

In the step S108) of the present invention, the change of the user demand is taken as the event e, including the event detection time T _d , the event type A, and the user requirement R. The event detection time T _d refers to the time when the event is detected; the event type A refers to the user behavior description corresponding to the event; the user requirement R is a set of time series corresponding to each user requirement, that is, R={r _T (k), r _H (k), r _L (k), r _W (k)}, where r _T (k), r _H (k), r _L (k), r _W (k) represent the temperature of the user at time k, Humidity, lighting, and hot water demand define the corresponding demand variable r(k) = 0 when the user has no demand at time k.

The optimization problem in the step S108) of the present invention is that the total electricity cost of the home appliance operation running in the optimized scheduling period is minimum, that is,

The constraints of the optimization problem in the step S108) of the present invention include three categories: the power consumption model of the electrical equipment, the operational constraints of the electrical equipment, and the comfort requirement constraints. Among them, the power consumption model of electrical equipment usually adopts the state equation to establish the connection between the electrical power of the equipment and the user demand it provides; the operational constraints of the electrical equipment refer to the physical limitations during the operation of the electrical equipment, usually including the maximum power constraints of the electrical equipment, electrical equipment Operational safety constraints, etc.; comfort requirement constraints mean that the user's room environment variables need to meet current comfort requirements.

The optimization problem in step S108) of the present invention is driven by an event. The specific process is: when the event e is detected at the time T _d , the user requirement R contained therein is read; and a time window length L is defined, at T _d to T _d + The optimization problem is solved based on the user demand R during the time period of L; the obtained optimization strategy is used to control the electrical equipment until a new event is detected, and the drive optimization problem is recalculated to obtain a new strategy.

1. The specific implementation process of behavior sensing and intelligent control driven by a motion sensor

Scene 1: Work

1) Real-time collection of various sensor data of the user's smart wearable device, including three-axis accelerator data, gyroscope data, skin temperature sensor data, skin sensor data, GPS data;

2) Upload the data to the server, format the data, discard the local timestamp of the collected data, and assign a uniform server timestamp to the time attribute field of the data according to the server time; add a column of source fields in the data attribute field. And populate the source field based on the source point of the data.

3) Determine the source of the data based on the source field of the data. The motion sensor data is processed as follows: according to the GPS data and the established location information database, the scene in which the user is working can be determined; the three-axis acceleration data and the gyroscope data input into the behavior state classifier can be obtained by the user in real time. The behavior state, when the user is working, the behavior state is the regular quiescent state. The biosensor is processed as follows: the biosensor data is input into the body state classifier, and the user's physical state can be obtained. Assuming that the user is all normal at this time, a normal state result can be obtained.

4) Comparing the user behavior state with the user's physical state and the behavior state-physical state-external environment-user requirement rule base to obtain the real-time demand of the user, and input the real-time demand into the demand update judgment module.

5) Compare the latest estimated demand with the previous estimated result. Since the user's normal work, work plan and physical state have not changed, the estimated demand is the same as the previous estimate, and the demand is not updated. Step 1) Continue to collect data.

Scene 2: Sports

3) Determine the source of the data based on the source field of the data. The motion sensor data is processed as follows: according to GPS The data and the established location information database can determine that the user is in the sports field at the moment, and the motion situation is obtained; the three-axis acceleration data and the gyroscope data input into the behavior state classifier can obtain the real-time behavior state of the user, and the user is exercising at this time. Then its behavior state is motion state. The biosensor is processed as follows: the biosensor data is input into the body state classifier, and the user's physical state can be obtained. When the user is exercising, the body state is excited.

4) comparing the user behavior state and the user's physical state with the behavior state-physical state-external environment-user requirement rule base to obtain the real-time demand of the user. Since the user is exercising, according to the demand rule library, the user can get the user back home. The demand for water heaters and air conditioners can be estimated by the magnitude and timing of demand.

5) Comparing the latest estimated demand obtained with the previous estimated result, since the user goes to exercise and changes compared with the previous estimated plan, the estimated demand is different from the previous estimated result, and the demand is updated and transferred. The event-driven electrical equipment control strategy optimization module generates a control strategy.

6) Combining the environmental data collected in step 1), under the constraints of the user's living environment requirements and electrical equipment, to save the entire household electricity consumption to minimize the goal, to generate a joint operation control strategy of various devices.

7) Update the control strategy of the control module and go to step 1) to continue monitoring the user's wearable device sensor data.

2. The specific implementation process of biosensor-driven behavioral awareness and intelligent control:

Scene 1: At home - sick

3) Determine the source of the data based on the source field of the data. The motion sensor data is processed as follows: according to GPS The data and the established location information database can determine that the user is at home at this time; inputting the three-axis acceleration data and the gyroscope data into the behavior state classifier can obtain the real-time behavior state of the user, and the user is at home, assuming that the user is at this time While watching TV, its behavior is in a normal state of rest. The biosensor is processed by inputting biosensor data into a body state classifier such as skin temperature and skin sensor data. After the classification algorithm is calculated, the user's physical state can be obtained. At this time, the patient can get the result of being sick-fatigued.

4) Comparing the user behavior state with the user's physical state and the behavior state-physical state-external environment-user requirement rule base can obtain the real-time demand of the user. Since the user is detected to be sick, the demand for the air-conditioning temperature rise and the need are generated. To soothe the demand for music, enter the real-time requirements into the demand update judgment module.

5) Comparing the latest estimated demand obtained with the previous estimated result, since the user is sick at home and has changed compared with the previous user's physical state estimation, the estimated demand is different from the previous estimated result, and the demand is updated. Transfer to the event-driven electrical equipment control strategy optimization module to generate a control strategy.

Thus, a smart home energy management process based on the behavioral awareness of smart wearable devices was completed.

Claims

The smart home energy management method based on the behavior sensing of the smart wearable device is characterized in that the method comprises the following steps:

S101), sensor data of the wearable device and external environment data collection: collecting data of the motion sensor, the biosensor, and the environmental sensor in the smart wearable device carried by the user; and collecting the weather, traffic, and electricity price information of the current location of the user; The sensor acquisition data and the external environment data are stored in an environment database for predicting a comfortable target desired by the user for the environment; the data collected by the motion sensor and the biosensor is transferred to step S102);

S102), sensor data preprocessing of the wearable device: preprocessing formats the data, including time synchronization and data source classification, wherein the time synchronization refers to performing sensor data acquired on different wearable devices according to the local time of the system. Time synchronization processing; data source classification is to classify and identify data according to wearable devices, collected sensors and data structure attributes of the data source;

S103) sensor type determination: according to the type of sensor data, for the data collected by the motion sensor, proceeds to step S104); for the data collected by the biosensor, proceeds to step S105);

S104) User behavioral awareness: the user's behavior includes behavioral and behavioral states; the behavioral context refers to the environment in which the user's athletic behavior occurs; combined with the location information base, using the user's location information acquired by GPS, and implementing the behavioral context according to the location information base Estimation; behavioral state refers to the type of motion behavior currently being performed by the user; by analyzing the data acquired by the acceleration sensor, the gyroscope, the geomagnetic sensor or the electronic compass sensor, the behavior state classifier is used to realize the perception of the behavior state;

S105) The user's physical state perception: estimating the current state of the user's body according to the combination of different biometrics according to the current heart rate, blood pressure, body surface temperature, and blood glucose data collected by the biosensor;

S106), user living environment demand estimation: user living environment demand refers to the user's expectation of temperature, humidity, illuminance, hot water, diet in the current and future environment, and the corresponding time; according to step S104) and step S105) The perceived result and the environmental data obtained in step S101) are combined with the pre-built user motion behavior-user physical state-correspondence table of the external environment and the user's living environment requirement, and the user's living environment demand is estimated, and according to the actual user's actual user's actual line. In order to dynamically update the demand correspondence table in the form of feedback;

S107), the user demand update judgment: comparing step S106) obtaining the latest estimated user living environment demand, comparing with the previous estimation result, determining whether the demand changes; if no change occurs, proceeding to step S101) to continue collecting data; If a change occurs, proceed to step S108);

S108), event-driven electrical equipment control strategy optimization: using event-driven online optimization method, combined with the environmental data collected in step S101), to meet the user's living environment requirements and electrical equipment constraints, to save the entire household electricity costs Minimize the target and generate a joint operational control strategy for various devices;

S109), updating the control policy of the smart home control module according to the result of step S108), and proceeding to step S101) to continue monitoring the wearable device sensor data of the user.
The smart home energy management method based on smart wearable device behavior sensing according to claim 1, wherein the motion sensor comprises a GPS, an acceleration sensor, a gyroscope, a geomagnetic sensor, an electronic compass sensor, and the biosensor comprises a blood glucose sensor and a blood pressure sensor. Sensors, ECG sensors, myoelectric sensors, body temperature sensors, brain wave sensors; environmental sensors include temperature and humidity sensors, atmospheric pressure sensors, gas sensors, pH sensors, ultraviolet sensors, ambient light sensors, particulate matter sensors, air pressure sensors, microphones.
The smart home energy management method based on smart wearable device behavior sensing according to claim 1, wherein the formatting of the wearable device sensor data in step S102) comprises time synchronization processing and data source classification; time synchronization processing The specific method is: uploading the collected data to the server uniformly, discarding the local timestamp of the collected data, and marking the time field of the data according to the server time; the specific method of the data source classification is: The source field is added to the attribute list, and the source attribute field of the data is populated based on the source point of the data.
The intelligent home energy management method based on the behavior of the smart wearable device according to claim 1, wherein the location information database in step S104) specifically includes: location information of a local sensitive building, and the sensitive building includes an office location and a sports place. , shopping, restaurants, home construction; behavioral state classifiers and body state classifier training The method is: according to the historical multi-dimensional sensor data vector, Adaboost is used to train the user behavior state classifier and the body state classifier.
The smart home energy management method based on the smart wearable device behavior perception according to claim 1, wherein the user sports behavior-user physical state-the construction of the correspondence table between the external environment and the user living environment requirement in step S106) The updating method is: according to the user historical behavior data statistics, respectively, the user motion behavior-life demand correspondence table, the user body state-life demand correspondence table, the external environment-life demand correspondence table, and the three children are given according to the prior expert knowledge and statistical rules. The corresponding weights of the table, and the weights of the three sub-tables are combined to obtain the initial user motion behavior-body state-external environment and life demand correspondence table; the online learning method is used to collect the feedback result of the user satisfaction in real time, and recalculate Sub-table weights, continuously update the user life requirements correspondence table;

The method for estimating the user's living environment requirement in step S106 is specifically: according to the user's behavioral situation and behavioral state, mapping based on the requirement rule base, obtaining an environment requirement corresponding to the latest user behavior; and the demand time is performed according to the GPS information and the road condition information. Estimate; the user requirements rule base is empirically established based on defined user behavior;

The steps of assigning corresponding weights to the three sub-tables according to prior knowledge and statistical rules, and weighting the three sub-tables to obtain an initial user motion behavior-physical state-external environment and life demand correspondence table include: The three tables are set to equal weights, that is, the life requirements corresponding to the user behavior, the life requirements corresponding to the user's physical state, and the life requirements corresponding to the external environment are linearly superimposed; when there is historical operation behavior data of the user, according to the past operational behavior of the user The weight of the correlation coefficient of the user's physical state and the external environment, adjust the weight of the three tables; during the use process, record the user's manual operation behavior and evaluation information of the home environment, analyze the user's satisfaction with the home environment, and continuously adjust The weight of the three tables.
The intelligent home energy management method based on the smart wearable device behavior sensing device of claim 1 , wherein the updating strategy for the existing user requirements in step S107) comprises: comparing the latest user requirements with current user requirements. If the demand periods do not overlap, increase the corresponding user requirements; if the demand periods overlap, they will be used The latest user requirements replace current user needs.
The smart home energy management method based on smart wearable device behavior sensing according to claim 1, wherein in step S108), the change of user demand is taken as event e, including event detection time T d , event type A, user Requirement R; wherein, the event detection time T d refers to the time when the event is detected; the event type A refers to the user behavior description corresponding to the event; the user requirement R is a set of time series corresponding to each user requirement, that is, R={r T (k ), r H (k), r L (k), r W (k)}, where r T (k), r H (k), r L (k), r W (k) represent the user at time k The temperature, humidity, lighting, and hot water requirements define the corresponding demand variable r(k) = 0 when the user has no demand at time k.
The smart home energy management method based on the smart wearable device behavior perception according to claim 7, wherein the optimization problem in step S108) is that the total electricity cost of the home appliance operation running in the optimized scheduling period is minimum, that is,
Where t 0 represents the initial time of the optimization period, t d represents the end time of the optimization period, N is the number of scheduled electrical devices, p(k) represents the price of electricity at time k, and q i (k) represents the time of device i at time k The power is the decision variable.
The intelligent home energy management method based on the smart wearable device behavior perception according to claim 8, wherein the constraint of the optimization problem in step S108) comprises three types: a power consumption model of the electrical device, an operation constraint of the electrical device, and The comfort requirement constraint; among them, the power consumption model of the electrical equipment usually adopts the state equation to establish the connection between the electrical power of the equipment and the user demand provided by the equipment; the operational constraint of the electrical equipment refers to the physical limitation during the operation of the electrical equipment, which usually includes the largest electrical equipment. Power constraints, electrical equipment operating safety constraints; comfort requirements constraints means that the user room environment variables need to meet current comfort requirements.
The smart home energy management method based on the smart wearable device behavior perception according to claim 9, wherein the optimization problem in step S108) is driven by an event, and the specific process is: when the event e is detected at the time T d , the read Take the user requirement R contained therein; define a time window length L, and solve the optimization problem based on the user demand R during the period from T d to T d + L; the obtained optimization strategy is used to control the electrical equipment until new The event is detected and the drive optimization problem is recalculated to get a new strategy.