CN109656137A - The daily behavior prediction technique of smart home environment servant - Google Patents
The daily behavior prediction technique of smart home environment servant Download PDFInfo
- Publication number
- CN109656137A CN109656137A CN201811551611.XA CN201811551611A CN109656137A CN 109656137 A CN109656137 A CN 109656137A CN 201811551611 A CN201811551611 A CN 201811551611A CN 109656137 A CN109656137 A CN 109656137A
- Authority
- CN
- China
- Prior art keywords
- data
- behavior
- time
- sensor
- window
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses the daily behavior prediction techniques of energy domestic environment servant a kind of, and step 1: behavioral data pretreatment: daily behavior identification needs to establish behavior model using existing historical behavior data set, and is labeled to data, and data are generated by sensor；Step 2: behavior model is established using the algorithm of logistic regression and establishes behavior model；Step 3: real-time behavior prediction carries out real-time behavior prediction using the model established in step 2, wherein the different sliding window length of behavior and the speed of behavior prediction.The present invention is able to achieve quick identification；It is mainly used in the prediction under smart home environment to daily behavior.
Description
Technical field
The present invention relates to the daily behavior prediction techniques of smart home environment servant a kind of.
Background technique
The daily behavior prediction technique of smart home environment servant, the prior art are to pass through image recognition skill using camera
Art identifies behavior, and this method can damage privacy of user in terms of daily behavior.Existing sensor-based identification
Main problem present in method is to can not achieve quick identification.
Summary of the invention
The purpose of the present invention is to provide the daily behavior of smart home environment servant for being able to achieve and quickly identifying a kind of is pre-
Survey method.
The technical solution of the invention is as follows:
A kind of daily behavior prediction technique of smart home environment servant, it is characterized in that: including the following steps:
Step 1: behavioral data pretreatment: daily behavior identification needs to establish behavior mould using existing historical behavior data set
Type, and data are labeled, data are generated by sensor；
Step 1.1: data format processing, by original sensor data Format adjusting be < timeStamp, SensorID,
sensorValue, Activity>；TimeStamp is the time that sensing data generates, and sensorID is that sensor is compiled
Number, value range (0-N) N is sensor sum；SensorValue is sensor reading；Activity behavior label,
It can be sky；
Step 1.2: deleting the incomplete record of data intensive data, be empty record including sensorValue；
Step 1.3: the standardization of sensing data.By temperature (D temp), light sensor data (D light) mapping
To [0,1] space；
Step 1.4: calculating the minimum M in temp of temperature data, and maximum value Max temp, standardized data is public
Formula DT temp=(D temp-Min temp)/(Max temp-Min temp)；
Step 1.5: calculating the minimum M in light of light data, and maximum value Max light, standardized data
Formula DT light=(D light-Min light)/(Max light-Min light)；
Step 1.6: the adjustment standardization of behavior label, including apparent error record is deleted, and modification misspelling record etc., output
Generate data set DS org；
Step 1.7: for the speed of accuracy and the identification of raising Activity recognition, different behaviors being handled and built respectively
Formwork erection type；Initially set up model of having a meal；Data set DS org < timeStamp, SensorID, sensorValue are modified,
Activity>in the domain Activity, value range is revised as<Eating (having a meal), and Non-Eeating(does not have a meal)
>；
Step 1.8: further the domain Activity being modified, Eating is updated to 1, Non-Eating and is updated to
0.
Step 1.9: setting time window (Sliding Window) length is 1 minute (60 seconds), is slided using recursion
Window, algorithm export new data set DS 60sec Eat as shown in Fig. 1；
Step 1.10: output sliding window data collection DS 60sec Eat format is as follows: < t j, S 1 t (j-k) ~ tj
, 2 t of S (j-k) ~ tj, 3 t of S (j-k) ~ tj ... S n t (j-k) ~ tj, (j, k were used as to the time Act tj >
Counting, t indicates the time, and S indicates sensing data sequence, and n indicates that sliding window counts, and S n t (j-k) ~ tj is indicated
N-th sensing data sequence, initial time are t (j-k), and front slide is worked as in the expression of end time t j, Act tj
The behavior label of window；
Step 1.11: calculating the sum (Snum) for enlivening sensor in each window, the active time of each sensor
<T 0, T 1, T 2 ... T Snum>, and each sensor enlivens frequency<F 0, F 1, F 2
…… F Snum >；
Step 1.12: the average value (T mean) for the active time length that sensor is enlivened in each window is calculated, it is maximum
It is worth (T max), minimum value (T min) and variance (T sd)；
Step 1.13: domain newly-generated in step 11-12 being added in data set DS 60sec Eat, is generated new
Data set DS_2 60sec Eat, data format be < t j, S 1 t (j-k) ~ tj, 2 t of S (j-k) ~ tj, S
3 t (j-k) ~ tj ... S n t (j-k) ~ tj, T 0, T 1, T 2 ... T Snum, F 0, F 1, F 2
... F Snum, T mean, T max, T min, T sd, Act tj >；
Step 1.14: to cooking, behavioral data is pre-processed.Data set DS org < timeStamp is modified,
SensorID, sensorValue, Activity>in the domain Activity, value range is revised as<Cooking, Non-
Cooking>；
Step 1.15: further the domain Activity being modified, Cooking is updated to 1, Non-Cooking and is updated
It is 0；
Step 1.16: setting time window Sliding Window, length is 30 seconds, defeated using recursion sliding window
New data set DS 30sec cook out；
Step 1.17: executing step 1.10-1.13, export new data set DS_2 30sec cook；
Step 1.18: identical operation being executed to other behaviors, generates data set DS_2 10sec relax, DS_2 respectively
120sec sleep , DS_2 30sec work , DS_2 120sec dust , DS_2 10sec tv , DS_2
30sec bath；
Step 2: behavior model is established using the algorithm of logistic regression and establishes behavior model, the calculation formula of algorithm is as follows:
Step 2.1: utilizing data set (DS_2 60sec Eat, DS_2 the 30sec cook, DS_2 generated in step 1
10sec relax, DS_2 120sec sleep , DS_2 30sec work, DS_2 120sec dust , DS_2
10sec tv, DS_2 30sec bath) behavior model is established respectively, model DM_2 60sec Eat, DM_2 are generated respectively
30sec cook, DM_2 10sec relax, DM_2 120sec sleep, DM_2 30sec work, DM_2 120sec
dust ,DM_2 10sec tv, DM_2 30sec bath；
Step 3: real-time behavior prediction carries out real-time behavior prediction using the model established in step 2, wherein different behaviors
Sliding window length and behavior prediction speed；
Step 3.1: under real time environment, when user executes a certain item activity or behavior, just there is new real time sensor data
It generates, new data format<timeStamp, SensorID, sensorValue>, for different behaviors, generate different length
Time window；
Step 3.2: executing and calculated in step 1.11-1.12, the result of generation is added to actual time window, generate new
Window TW act len format < t j, S 1 t (j-k) ~ tj, 2 t of S (j-k) ~ tj, 3 t of S (j-k) ~ tj ...
S n t (j-k) ~ tj, T 0, T 1, T 2 ... T Snum, F 0, F 1, F 2 ... F Snum, T
Mean, T max, T min, T sd >；Wherein TW act len includes TW cook 30, TW eat 60, TW work
30, TW relax 10, TW sleep 120, TW dust 120, TW tv 10, TW bath 30,
Step 3.3: using the time window generated in step 3.2 as input, being separately input to corresponding in step 2.1
It is in model, i.e., predictable to export the behavior label being currently executing.
The present invention is able to achieve quick identification；It is mainly used in the prediction under smart home environment to daily behavior.Intelligence
The main sensors of domestic environment configuration include temperature, human body sensing, illumination.The present invention can be with Accurate Prediction (accuracy 95%
More than) behavior include cook (cook), have a meal (eat), work (work), rest (relax), sleep (sleep),
Cleaning (dust) sees TV (tv), and the time of (bath) that goes to the toilet prediction is 30 seconds respectively, 60 seconds, 30 seconds, and 10
Second, 120 seconds, 120 seconds, 10 seconds, 30 seconds.
Detailed description of the invention
Present invention will be further explained below with reference to the attached drawings and examples.
Fig. 1 is sliding window generating algorithm schematic diagram of the present invention.
Fig. 2 is smart home schematic diagram.
Specific embodiment
A kind of daily behavior prediction technique of smart home environment servant, including the following steps:
Step 1: behavioral data pretreatment: daily behavior identification needs to establish behavior mould using existing historical behavior data set
Type, and data are labeled, data are generated by sensor；
Step 1.1: data format processing, by original sensor data Format adjusting be < timeStamp, SensorID,
sensorValue, Activity>；TimeStamp is the time that sensing data generates, and sensorID is that sensor is compiled
Number, value range (0-N) N is sensor sum；SensorValue is sensor reading；Activity behavior label,
It can be sky；
Step 1.2: deleting the incomplete record of data intensive data, be empty record including sensorValue；
Step 1.3: the standardization of sensing data.By temperature (D temp), light sensor data (D light) mapping
To [0,1] space；
Step 1.4: calculating the minimum M in temp of temperature data, and maximum value Max temp, standardized data is public
Formula DT temp=(D temp-Min temp)/(Max temp-Min temp)；
Step 1.5: calculating the minimum M in light of light data, and maximum value Max light, standardized data
Formula DT light=(D light-Min light)/(Max light-Min light)；
Step 1.6: the adjustment standardization of behavior label, including apparent error record is deleted, and modification misspelling record etc., output
Generate data set DS org；
Step 1.7: for the speed of accuracy and the identification of raising Activity recognition, different behaviors being handled and built respectively
Formwork erection type；Initially set up model of having a meal；Data set DS org < timeStamp, SensorID, sensorValue are modified,
Activity>in the domain Activity, value range is revised as<Eating (having a meal), and Non-Eeating(does not have a meal)
>；
Step 1.8: further the domain Activity being modified, Eating is updated to 1, Non-Eating and is updated to
0.
Step 1.9: setting time window (Sliding Window) length is 1 minute (60 seconds), is slided using recursion
Window, algorithm export new data set DS 60sec Eat as shown in Fig. 1；
Step 1.10: output sliding window data collection DS 60sec Eat format is as follows: < t j, S 1 t (j-k) ~ tj
, 2 t of S (j-k) ~ tj, 3 t of S (j-k) ~ tj ... S n t (j-k) ~ tj, (j, k were used as to the time Act tj >
Counting, t indicates the time, and S indicates sensing data sequence, and n indicates that sliding window counts, and S n t (j-k) ~ tj is indicated
N-th sensing data sequence, initial time are t (j-k), and front slide is worked as in the expression of end time t j, Act tj
The behavior label of window；
Step 1.11: calculating the sum (Snum) for enlivening sensor in each window, the active time of each sensor
<T 0, T 1, T 2 ... T Snum>, and each sensor enlivens frequency<F 0, F 1, F 2
…… F Snum >；
Step 1.12: the average value (T mean) for the active time length that sensor is enlivened in each window is calculated, it is maximum
It is worth (T max), minimum value (T min) and variance (T sd)；
Step 1.13: domain newly-generated in step 11-12 being added in data set DS 60sec Eat, is generated new
Data set DS_2 60sec Eat, data format be < t j, S 1 t (j-k) ~ tj, 2 t of S (j-k) ~ tj, S
3 t (j-k) ~ tj ... S n t (j-k) ~ tj, T 0, T 1, T 2 ... T Snum, F 0, F 1, F 2
... F Snum, T mean, T max, T min, T sd, Act tj >；
Step 1.14: to cooking, behavioral data is pre-processed.Data set DS org < timeStamp is modified,
SensorID, sensorValue, Activity>in the domain Activity, value range is revised as<Cooking, Non-
Cooking>；
Step 1.15: further the domain Activity being modified, Cooking is updated to 1, Non-Cooking and is updated
It is 0；
Step 1.16: setting time window Sliding Window, length is 30 seconds, defeated using recursion sliding window
New data set DS 30sec cook out；
Step 1.17: executing step 1.10-1.13, export new data set DS_2 30sec cook；
Step 1.18: identical operation being executed to other behaviors, generates data set DS_2 10sec relax, DS_2 respectively
120sec sleep , DS_2 30sec work , DS_2 120sec dust , DS_2 10sec tv , DS_2
30sec bath；
Step 2: behavior model is established using the algorithm of logistic regression and establishes behavior model, the calculation formula of algorithm is as follows:
Step 2.1: utilizing data set (DS_2 60sec Eat, DS_2 the 30sec cook, DS_2 generated in step 1
10sec relax, DS_2 120sec sleep , DS_2 30sec work, DS_2 120sec dust , DS_2
10sec tv, DS_2 30sec bath) behavior model is established respectively, model DM_2 60sec Eat, DM_2 are generated respectively
30sec cook, DM_2 10sec relax, DM_2 120sec sleep, DM_2 30sec work, DM_2 120sec
dust ,DM_2 10sec tv, DM_2 30sec bath；
Step 3: real-time behavior prediction carries out real-time behavior prediction using the model established in step 2, wherein different behaviors
Sliding window length and behavior prediction speed；
Step 3.1: under real time environment, when user executes a certain item activity or behavior, just there is new real time sensor data
It generates, new data format<timeStamp, SensorID, sensorValue>, for different behaviors, generate different length
Time window；
Step 3.2: executing and calculated in step 1.11-1.12, the result of generation is added to actual time window, generate new
Window TW act len format < t j, S 1 t (j-k) ~ tj, 2 t of S (j-k) ~ tj, 3 t of S (j-k) ~ tj ...
S n t (j-k) ~ tj, T 0, T 1, T 2 ... T Snum, F 0, F 1, F 2 ... F Snum, T
Mean, T max, T min, T sd >；Wherein TW act len includes TW cook 30, TW eat 60, TW work
30, TW relax 10, TW sleep 120, TW dust 120, TW tv 10, TW bath 30,
Step 3.3: using the time window generated in step 3.2 as input, being separately input to corresponding in step 2.1
It is in model, i.e., predictable to export the behavior label being currently executing.
Algorithm parameter:
Parameter name parameter connotation
TimeStamp timestamp；
SensorID sensor number, the sensor sum configured in value 0-N, N smart home environment；
SensorValue sensor reading；
Activity behavior label；
Eat/eating has a meal；
Cook cooks；
Work work；
Relax rest；
Sleep sleep；
Tv sees TV；
Dust cleaning；
Bath goes to the toilet；
DtempTemperature data；
DlightLight data；
MintempTemperature minimum value；
MaxtempTemperature maximum；
DTtempTemperature data after standardization；
MinlightLight data minimum value；
MaxlightLight data maximums；
DTlightLight data after standardization；
DSorgPreliminary pretreated data set；
Sliding Window sliding window；
DS60sec EatFor the behavior of having a meal, the data set that sliding window length is 60 seconds；
The t time；
The counting of j, k to the time；
tjThe time of jth event in entire time series；
S1 t(j-k)~tjThe sum of data of the sensor that number is 1 from time t (j-k) to time t j；
Act tj tjThe behavior label at moment；
The sum for enlivening sensor in each window of Snum；
T x0-the Snum of active time x value of sensor；
F xSensor enlivens 0-Snum of frequency x value；
T meanThe average value of the active time length of sensor is enlivened in window；
T maxThe maximum value of the active time length of sensor is enlivened in window；
T minThe minimum value of the active time length of sensor is enlivened in window；
T sdThe variance yields of the active time length of sensor is enlivened in window；
DS60sec EatDS indicates that data set, 60sec indicate that time window length is 60 seconds, and Eat indicates behavior of having a meal；
DS_260sec EatDS indicates that data set, 60sec indicate that time window length is 60 seconds, and Eat indicates behavior of having a meal；
DS_230sec cookIt cooks behavioral data collection, DS indicates that data set, 3sec indicate that time window length is 30 seconds；
DS_2 10sec relaxRest behavioral data collection, DS_2 indicate that data set, 10sec indicate that time window length is 10
Second；
DS_2 120sec sleepSleep behavioral data collection, DS_2 indicate that data set, 120sec indicate that time window length is
120 seconds；
DS_2 30sec workWork behavior data set, DS_2 indicate that data set, 30sec indicate that time window length is 30
Second；
DS_2 120sec dust Cleaning behavioral data collection, DS_2 indicate that data set, 120sec indicate that time window length is
120 seconds；
DS_2 10sec tvSee that TV behavior data set, DS_2 indicate that data set, 10sec indicate that time window length is 10
Second；
DS_2 30sec bathGo to lavatory data set, DS_2 indicate that data set, 30sec indicate that time window length is 30
Second；
DM_2 60sec EatIt has a meal behavior model；
DM_2 30sec cookIt cooks behavior model；
DM_2 10sec relaxRest behavior model；
DM_2 30sec workWork behavior model.
Claims (1)
1. a kind of daily behavior prediction technique of smart home environment servant, it is characterized in that: including the following steps:
Step 1: behavioral data pretreatment: daily behavior identification needs to establish behavior mould using existing historical behavior data set
Type, and data are labeled, data are generated by sensor；
Step 1.1: data format processing, by original sensor data Format adjusting be < timeStamp, SensorID,
sensorValue, Activity>；TimeStamp is the time that sensing data generates, and sensorID is that sensor is compiled
Number, value range (0-N) N is sensor sum；SensorValue is sensor reading；Activity behavior label,
It can be sky；
Step 1.2: deleting the incomplete record of data intensive data, be empty record including sensorValue；
Step 1.3: the standardization of sensing data；By temperature (D temp), light sensor data (D light) mapping
To [0,1] space；
Step 1.4: calculating the minimum M in temp of temperature data, and maximum value Max temp, standardized data is public
Formula DT temp=(D temp-Min temp)/(Max temp-Min temp)；
Step 1.5: calculating the minimum M in light of light data, and maximum value Max light, standardized data
Formula DT light=(D light-Min light)/(Max light-Min light)；
Step 1.6: the adjustment standardization of behavior label, including apparent error record is deleted, and modification misspelling record etc., output
Generate data set DS org；
Step 1.7: for the speed of accuracy and the identification of raising Activity recognition, different behaviors being handled and built respectively
Formwork erection type；Initially set up model of having a meal；Data set DS org < timeStamp, SensorID, sensorValue are modified,
Activity>in the domain Activity, value range is revised as<Eating (having a meal), and Non-Eeating(does not have a meal)
>；
Step 1.8: further the domain Activity being modified, Eating is updated to 1, Non-Eating and is updated to
0.
Step 1.9: setting time window (Sliding Window) length is 1 minute (60 seconds), is slided using recursion
Window, algorithm export new data set DS 60sec Eat as shown in Fig. 1；
Step 1.10: output sliding window data collection DS 60sec Eat format is as follows: < t j, S 1 t (j-k) ~ tj
, 2 t of S (j-k) ~ tj, 3 t of S (j-k) ~ tj ... S n t (j-k) ~ tj, (j, k were used as to the time Act tj >
Counting, t indicates the time, and S indicates sensing data sequence, and n indicates that sliding window counts, and S n t (j-k) ~ tj is indicated
N-th sensing data sequence, initial time are t (j-k), and front slide is worked as in the expression of end time t j, Act tj
The behavior label of window；
Step 1.11: calculating the sum (Snum) for enlivening sensor in each window, the active time of each sensor
<T 0, T 1, T 2 ... T Snum>, and each sensor enlivens frequency<F 0, F 1, F 2
…… F Snum >；
Step 1.12: the average value (T mean) for the active time length that sensor is enlivened in each window is calculated, it is maximum
It is worth (T max), minimum value (T min) and variance (T sd)；
Step 1.13: domain newly-generated in step 11-12 being added in data set DS 60sec Eat, is generated new
Data set DS_2 60sec Eat, data format be < t j, S 1 t (j-k) ~ tj, 2 t of S (j-k) ~ tj, S
3 t (j-k) ~ tj ... S n t (j-k) ~ tj, T 0, T 1, T 2 ... T Snum, F 0, F 1, F 2
... F Snum, T mean, T max, T min, T sd, Act tj >；
Step 1.14: to cooking, behavioral data is pre-processed；Data set DS org < timeStamp is modified,
SensorID, sensorValue, Activity>in the domain Activity, value range is revised as<Cooking, Non-
Cooking>；
Step 1.15: further the domain Activity being modified, Cooking is updated to 1, Non-Cooking and is updated
It is 0；
Step 1.16: setting time window Sliding Window, length is 30 seconds, defeated using recursion sliding window
New data set DS 30sec cook out；
Step 1.17: executing step 1.10-1.13, export new data set DS_2 30sec cook；
Step 1.18: identical operation being executed to other behaviors, generates data set DS_2 10sec relax, DS_2 respectively
120sec sleep , DS_2 30sec work , DS_2 120sec dust , DS_2 10sec tv , DS_2
30sec bath；
Step 2: behavior model is established using the algorithm of logistic regression and establishes behavior model, the calculation formula of algorithm is as follows:
Step 2.1: utilizing data set (DS_2 60sec Eat, DS_2 the 30sec cook, DS_2 generated in step 1
10sec relax, DS_2 120sec sleep , DS_2 30sec work, DS_2 120sec dust , DS_2
10sec tv, DS_2 30sec bath) behavior model is established respectively, model DM_2 60sec Eat, DM_2 are generated respectively
30sec cook, DM_2 10sec relax, DM_2 120sec sleep, DM_2 30sec work, DM_2 120sec
dust ,DM_2 10sec tv, DM_2 30sec bath；
Step 3: real-time behavior prediction carries out real-time behavior prediction using the model established in step 2, wherein different behaviors
Sliding window length and behavior prediction speed；
Step 3.1: under real time environment, when user executes a certain item activity or behavior, just there is new real time sensor data
It generates, new data format<timeStamp, SensorID, sensorValue>, for different behaviors, generate different length
Time window；
Step 3.2: executing and calculated in step 1.11-1.12, the result of generation is added to actual time window, generate new
Window TW act len format < t j, S 1 t (j-k) ~ tj, 2 t of S (j-k) ~ tj, 3 t of S (j-k) ~ tj ...
S n t (j-k) ~ tj, T 0, T 1, T 2 ... T Snum, F 0, F 1, F 2 ... F Snum, T
Mean, T max, T min, T sd >；Wherein TW act len includes TW cook 30, TW eat 60, TW work
30, TW relax 10, TW sleep 120, TW dust 120, TW tv 10, TW bath 30,
Step 3.3: using the time window generated in step 3.2 as input, being separately input to corresponding in step 2.1
It is in model, i.e., predictable to export the behavior label being currently executing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811551611.XA CN109656137A (en) | 2018-12-19 | 2018-12-19 | The daily behavior prediction technique of smart home environment servant |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811551611.XA CN109656137A (en) | 2018-12-19 | 2018-12-19 | The daily behavior prediction technique of smart home environment servant |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109656137A true CN109656137A (en) | 2019-04-19 |
Family
ID=66113375
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811551611.XA Pending CN109656137A (en) | 2018-12-19 | 2018-12-19 | The daily behavior prediction technique of smart home environment servant |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109656137A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106445101A (en) * | 2015-08-07 | 2017-02-22 | 飞比特公司 | Method and system for identifying user |
| CN107241697A (en) * | 2017-06-30 | 2017-10-10 | 北京奇虎科技有限公司 | User behavior for mobile terminal determines method, device and mobile terminal |
| CN107248003A (en) * | 2017-08-03 | 2017-10-13 | 浙江大学 | Based on the adaptive soft-sensor Forecasting Methodology with sliding window Bayesian network |
-
2018
- 2018-12-19 CN CN201811551611.XA patent/CN109656137A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106445101A (en) * | 2015-08-07 | 2017-02-22 | 飞比特公司 | Method and system for identifying user |
| CN107241697A (en) * | 2017-06-30 | 2017-10-10 | 北京奇虎科技有限公司 | User behavior for mobile terminal determines method, device and mobile terminal |
| CN107248003A (en) * | 2017-08-03 | 2017-10-13 | 浙江大学 | Based on the adaptive soft-sensor Forecasting Methodology with sliding window Bayesian network |
Non-Patent Citations (5)
| Title |
|---|
| JAVIER ORTIZ LAGUNA等: "A Dynamic Sliding Window Approach for Activity Recognition", 《 KONSTAN J.A., CONEJO R., MARZO J.L., OLIVER N. (EDS) USER MODELING, ADAPTION AND PERSONALIZATION. UMAP 2011. LECTURE NOTES IN COMPUTER SCIENCE》 * |
| JUAN LORENZO HAGAD等: "Predicting Levels of Rapport in Dyadic Interactions through Automatic Detection of Posture and Posture Congruence", 《2011 IEEE THIRD INTERNATIONAL CONFERENCE ON PRIVACY, SECURITY, RISK AND TRUST AND 2011 IEEE THIRD INTERNATIONAL CONFERENCE ON SOCIAL COMPUTING》 * |
| 吕培卓等: "智能家居用户行为预测的方法研究", 《中国新技术新产品》 * |
| 李文锋等: "基于运动特征分析的人体异常行为模糊识别", 《华中科技大学学报(自然科学版)》 * |
| 毛驾燕等: "一种面向智能家居老人看护系统的实现方案", 《计算机技术与发展》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110251034B (en) | Control method for disinfection operation and dish-washing machine | |
| CN106597946B (en) | A kind of indoor occupant method for monitoring state and system | |
| CN104482656B (en) | The method of Intelligent adjustment water heater temperature, device and water heater | |
| US9501613B1 (en) | Health and wellness management technology | |
| JP2020506445A5 (en) | ||
| CN107205698A (en) | For the movable system and method for the daily life for monitoring people | |
| DE202014011334U1 (en) | Data Analysis System | |
| CN109887595A (en) | Heartbeat anomalous identification algorithm based on depth learning technology | |
| CN110197235B (en) | Human body activity recognition method based on unique attention mechanism | |
| US9251463B2 (en) | Knowledge transfer in smart environments | |
| CN105326434A (en) | Intelligent tissue box and working method thereof | |
| JP5504529B2 (en) | Watching robot, watching method, and watching program | |
| Ramírez-Morales et al. | Automated early detection of drops in commercial egg production using neural networks | |
| CN110036372A (en) | Data processing equipment, data processing method, setting managing device and data processing system | |
| CN116703227B (en) | Guest room management method and system based on intelligent service | |
| JP2016016295A (en) | Blood pressure estimation apparatus | |
| CN108732975A (en) | Hand cleanser follower goes out hydraulic control method, apparatus and hand cleanser follower | |
| CN110111815A (en) | Animal anomaly sound monitoring method and device, storage medium, electronic equipment | |
| CN109656137A (en) | The daily behavior prediction technique of smart home environment servant | |
| Najiyah et al. | Hadith degree classification for Shahih Hadith identification web based | |
| CN113456063B (en) | Artificial intelligence-based dead chicken disease detection system and detection method | |
| CN105232063B (en) | User mental health detection method and intelligent terminal | |
| CN107253202A (en) | A kind of method that robot of supporting parents wakes up | |
| CN109886721A (en) | A kind of pork price forecasting system algorithm | |
| CN115507519A (en) | Method and device for monitoring air conditioner, air conditioner and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190419 |