CN117271994A

CN117271994A - Door detection method and system for care monitoring of old people

Info

Publication number: CN117271994A
Application number: CN202311226009.XA
Authority: CN
Inventors: 王华华; 施展昊; 杨亚杰; 赵忠乾; 郑晓东
Original assignee: Beijing Bidi Intelligent Technology Co ltd; Global Business Intelligence Consulting Co
Current assignee: Beijing Bidi Intelligent Technology Co ltd; Global Business Intelligence Consulting Co
Priority date: 2023-09-22
Filing date: 2023-09-22
Publication date: 2023-12-22
Anticipated expiration: 2043-09-22
Also published as: CN117271994B

Abstract

The invention relates to the field of artificial intelligence, and discloses a door detection method and a door detection system for care monitoring of old people, which are used for realizing intelligent early warning for old people and improving early warning accuracy. The method comprises the following steps: acquiring intelligent door magnetic state data of a target door body and acquiring time sequence data; performing door body switching frequency characteristic analysis on the intelligent door magnetic state data to obtain door body switching frequency characteristic data; performing activity duration feature analysis to obtain activity duration feature data; performing eigenvalue mapping, generating a switch eigenvector and a duration eigenvector, and performing matrix conversion to obtain a target eigenvector; inputting the target feature matrix into an old man activity anomaly analysis model to perform old man activity anomaly analysis, so as to obtain an old man activity anomaly analysis result; and matching the target monitoring strategy according to the analysis result of the activity abnormality of the old, and generating intelligent reminding according to the target monitoring strategy.

Description

Door detection method and system for care monitoring of old people

Technical Field

The invention relates to the field of artificial intelligence, in particular to a door detection method and a door detection system for care monitoring of old people.

Background

With the increasing trend of aging of the population, care and health care of the elderly have become a focus of social attention. Elderly people face a series of health and safety problems in daily life, such as sudden illness, accidental falls, inconvenient actions, etc. Therefore, the intelligent monitoring system is developed, the state and the behavior of the old can be monitored in real time, early warning and help providing can be realized, and the intelligent monitoring system has important significance for maintaining the health and safety of the old. In this context, sensor technology-based geriatric care monitoring systems have received a great deal of attention. The sensor technology can acquire life activity data of the old in real time, such as door body state, walking path, heart rate and the like, and provide valuable information for guardianship personnel, so that the requirements and problems of the old can be responded in time. The monitoring of the door body state is particularly important, because the daily activities of the old are often related to the entrance and exit of rooms, and the change of the door body state reflects the activity habit and behavior rule of the old, and even predicts health problems or safety risks.

However, conventional care monitoring systems for elderly people often rely on manual intervention and observation, and it is difficult to accurately capture the activity states and abnormalities of the elderly in real time. Therefore, an automatic and intelligent door detection method is developed, real-time monitoring, analysis and early warning can be carried out through intelligent door magnetic data, and the limitation of a traditional monitoring system is solved.

Disclosure of Invention

The invention provides a gate detection method and a gate detection system for care monitoring of old people, which are used for realizing intelligent early warning for old people and improving the early warning accuracy.

The first aspect of the present invention provides a door detection method for care of the elderly, comprising:

acquiring intelligent door magnetic state data of a target door body based on a preset sampling time interval, and acquiring time sequence data corresponding to the intelligent door magnetic state data;

performing door body switching frequency characteristic analysis on the intelligent door magnetic state data to obtain door body switching frequency characteristic data;

according to the time sequence data and the intelligent door magnetic state data, performing activity duration characteristic analysis on the old user to obtain activity duration characteristic data;

performing feature value mapping on the door body switching frequency feature data to generate a switching feature vector, performing feature value mapping on the active duration feature data to generate a duration feature vector, and performing matrix conversion on the switching feature vector and the duration feature vector to obtain a target feature matrix;

inputting the target feature matrix into a preset old people activity abnormality analysis model to perform old people activity abnormality analysis, so as to obtain an old people activity abnormality analysis result;

And matching a corresponding target monitoring strategy according to the analysis result of the activity abnormality of the old people, generating corresponding intelligent reminding information according to the target monitoring strategy, and carrying out intelligent reminding on the old people through the target door body.

With reference to the first aspect, in a first implementation manner of the first aspect of the present invention, performing a gate body switching frequency feature analysis on the intelligent gate magnetic state data to obtain gate body switching frequency feature data includes:

performing outlier removal and missing value interpolation on the intelligent door magnetic state data to obtain standard state sequence data;

dividing the standard state sequence data to obtain a plurality of sub-state sequence data, and respectively calculating the switching frequency of the plurality of sub-state sequence data to obtain the switching frequency data of each sub-state sequence data;

frequency characteristic extraction is carried out on the switching frequency data of each sub-state sequence data to obtain first frequency characteristic data of each sub-state sequence data, wherein the first frequency characteristic data comprises average switching frequency, highest switching frequency and lowest switching frequency;

performing feature fusion on the first frequency feature data of each sub-state sequence data to obtain second frequency feature data;

Performing Fourier transform on the second frequency characteristic data to obtain target frequency domain data, and performing spectrum analysis on the target frequency domain data to obtain target spectrum data;

performing frequency component analysis on the target frequency spectrum data to obtain frequency component data, and performing periodic analysis on the frequency component data to obtain periodic frequency characteristics;

and generating corresponding gate body switching frequency characteristic data according to the periodic frequency characteristic and the second frequency characteristic data.

With reference to the first aspect, in a second implementation manner of the first aspect of the present invention, the performing, according to the time sequence data and the intelligent door magnetic state data, an activity duration feature analysis on an old user to obtain activity duration feature data includes:

carrying out data alignment on the time sequence data and the intelligent door magnetic state data, and calculating the time length of each time of the activities of the old user entering and exiting a room to obtain a plurality of activity time length data of the switch state change;

extracting activity duration features from the plurality of activity duration data to obtain a plurality of initial activity duration features, wherein the plurality of initial activity duration features comprise: average activity duration, longest activity duration, and shortest activity duration;

Calling a preset probability density distribution function to perform distribution operation on the plurality of activity duration data, and generating an activity duration probability density distribution map;

performing time period comparison on the activity duration probability density distribution map to obtain activity duration distribution characteristics;

and generating corresponding activity duration feature data according to the activity duration distribution feature and the plurality of initial activity duration features.

With reference to the first aspect, in a third implementation manner of the first aspect of the present invention, the performing eigenvalue mapping on the gate switching frequency characteristic data to generate a switching eigenvector, performing eigenvalue mapping on the active duration eigenvalue data to generate a duration eigenvector, and performing matrix conversion on the switching eigenvector and the duration eigenvector to obtain a target eigenvector, including:

carrying out standardization processing on the door body switching frequency characteristic data to obtain standardized frequency characteristic data, and carrying out characteristic value mapping on the standardized frequency characteristic data through a preset first mapping function to obtain a plurality of first characteristic values;

generating a plurality of first vector elements according to the plurality of first eigenvalues, and generating a switch eigenvector according to the plurality of first vector elements;

Performing standardization processing on the activity duration feature data to obtain standardized duration feature data, and performing feature value mapping on the standardized duration feature data through a preset second mapping function to obtain a plurality of second feature values;

generating a plurality of second vector elements according to the plurality of second eigenvalues, and generating a duration eigenvector according to the plurality of second vector elements;

performing weighting operation on the switch feature vector according to a preset first weight data set to obtain a weighted switch feature vector, and performing weighting operation on the duration feature vector according to a preset second weight data set to obtain a weighted duration feature vector;

and performing matrix conversion on the weighted switch feature vector and the weighted duration feature vector to obtain a target feature matrix.

With reference to the first aspect, in a fourth implementation manner of the first aspect of the present invention, inputting the target feature matrix into a preset senior citizen activity anomaly analysis model to perform senior citizen activity anomaly analysis, to obtain an senior citizen activity anomaly analysis result, includes:

model training is carried out on an isolated forest model through a preset isolated forest library, and a model parameter set of the isolated forest model is set to obtain an old man activity anomaly analysis model;

Inputting the target feature matrix into the senior citizen activity anomaly analysis model, and calculating a first anomaly score of each column vector in the target feature matrix through the senior citizen activity anomaly analysis model;

performing mean value operation on the first anomaly scores of each column vector to obtain second anomaly scores of the target feature matrix;

comparing the second anomaly score with a preset target threshold, and if the second anomaly score is larger than the target threshold, determining that the old activity anomaly analysis result is that the old activity anomaly exists;

and if the second anomaly score is smaller than or equal to the target threshold value, determining that the old people activity anomaly analysis result is that the old people activity is not abnormal.

With reference to the first aspect, in a fifth implementation manner of the first aspect of the present invention, the matching the corresponding target monitoring policy according to the analysis result of the activity anomaly of the elderly, generating corresponding intelligent reminding information according to the target monitoring policy, and performing intelligent reminding on the elderly user through the target gate body includes:

obtaining a monitoring policy bank, wherein the monitoring policy bank comprises a plurality of candidate monitoring policies;

Creating identification information of each candidate monitoring strategy in the monitoring strategy library, and constructing a target corresponding relation between the identification information and the second abnormal score;

according to the analysis result of the activity abnormality of the old and the second abnormality score, matching a corresponding target monitoring strategy from the plurality of candidate monitoring strategies;

and generating corresponding intelligent reminding information according to the target monitoring strategy, and carrying out intelligent reminding on the old people through the target gate body.

With reference to the first aspect, in a sixth implementation manner of the first aspect of the present invention, the door detection method for care monitoring of the elderly further includes:

after intelligent reminding is carried out, acquiring door body switching pressure data of the target door body, and calculating feedback parameters of the second abnormal score according to the door body switching pressure data to generate a third abnormal score;

performing policy gradient analysis on the elderly users based on the third anomaly score, and determining policy gradients corresponding to the elderly users;

and carrying out strategy updating through the strategy gradient and the third anomaly score to generate a feedback monitoring strategy.

A second aspect of the present invention provides a door detection system for senior citizen care monitoring, comprising:

The acquisition module is used for acquiring intelligent door magnetic state data of the target door body based on a preset sampling time interval and acquiring time sequence data corresponding to the intelligent door magnetic state data;

the frequency analysis module is used for carrying out door body switching frequency characteristic analysis on the intelligent door magnetic state data to obtain door body switching frequency characteristic data;

the time length analysis module is used for carrying out activity time length characteristic analysis on the old people user according to the time sequence data and the intelligent door magnetic state data to obtain activity time length characteristic data;

the conversion module is used for carrying out eigenvalue mapping on the door body switching frequency characteristic data to generate a switching eigenvector, carrying out eigenvalue mapping on the active duration characteristic data to generate a duration eigenvector, and carrying out matrix conversion on the switching eigenvector and the duration eigenvector to obtain a target eigenvector;

the abnormality analysis module is used for inputting the target feature matrix into a preset old people activity abnormality analysis model to perform old people activity abnormality analysis, so as to obtain an old people activity abnormality analysis result;

the generation module is used for matching the corresponding target monitoring strategy according to the analysis result of the activity abnormality of the old people, generating corresponding intelligent reminding information according to the target monitoring strategy, and carrying out intelligent reminding on the old people through the target door body.

A third aspect of the present invention provides a door body detection apparatus for care of elderly people, comprising: a memory and at least one processor, the memory having instructions stored therein; the at least one processor invokes the instructions in the memory to cause the gate detection device for geriatric care monitoring to perform the gate detection method for geriatric care monitoring described above.

A fourth aspect of the present invention provides a computer-readable storage medium having instructions stored therein, which when run on a computer, cause the computer to perform the gate detection method for geriatric care monitoring described above.

In the technical scheme provided by the invention, intelligent door magnetic state data of a target door body are acquired, and time sequence data are acquired; performing door body switching frequency characteristic analysis on the intelligent door magnetic state data to obtain door body switching frequency characteristic data; performing activity duration feature analysis to obtain activity duration feature data; performing eigenvalue mapping, generating a switch eigenvector and a duration eigenvector, and performing matrix conversion to obtain a target eigenvector; inputting the target feature matrix into an old man activity anomaly analysis model to perform old man activity anomaly analysis, so as to obtain an old man activity anomaly analysis result; according to the method, the system and the device, the target monitoring strategy is matched according to the abnormal activity analysis result of the old, and intelligent reminding is performed according to the target monitoring strategy, and the state change of the old entering and exiting a room is monitored in real time through the intelligent door magnetic data, so that the activity track and habit of the old can be accurately captured, and the guardian can know the life rule of the old. By analyzing the door body switching frequency characteristic and the activity duration characteristic, the method can identify abnormal conditions, such as long-time inactivity, frequent in-out and the like, and early warn potential health problems or safety risks in time. Can monitor the activity habit of old man, in time discover to fall down, unexpected circumstances such as lost, reduce the emergence of unexpected accidents through intelligent warning, match individualized guardianship tactics through unusual analytical model according to the actual conditions of old man, remind to specific activity period for example, improve guardianship's effectiveness, and then realized carrying out intelligent early warning to old man user to the rate of accuracy of early warning has been improved.

Drawings

FIG. 1 is a diagram illustrating an embodiment of a door detection method for geriatric care monitoring according to an embodiment of the present invention;

FIG. 2 is a flow chart of the feature analysis of the activity duration in an embodiment of the present invention;

FIG. 3 is a flow chart of matrix conversion in an embodiment of the invention;

FIG. 4 is a flow chart of an abnormality analysis of an elderly activity in an embodiment of the present invention;

FIG. 5 is a schematic diagram of one embodiment of a gate detection system for geriatric care monitoring in accordance with an embodiment of the present invention;

fig. 6 is a schematic diagram of an embodiment of a door detection apparatus for care monitoring of elderly people according to an embodiment of the present invention.

Detailed Description

The embodiment of the invention provides a door detection method and a door detection system for care monitoring of old people, which are used for realizing intelligent early warning for old people and improving the early warning accuracy. The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.

For ease of understanding, a specific flow of an embodiment of the present invention is described below with reference to fig. 1, and an embodiment of a door detection method for care monitoring of elderly people according to the embodiment of the present invention includes:

s101, acquiring intelligent door magnetic state data of a target door body based on a preset sampling time interval, and acquiring time sequence data corresponding to the intelligent door magnetic state data;

it will be appreciated that the execution subject of the present invention may be a gate detection system for care and monitoring of elderly people, or may be a terminal or a server, and is not limited herein. The embodiment of the invention is described by taking a server as an execution main body as an example.

Specifically, the server presets a suitable sampling time interval. This time interval selection requires balancing data accuracy and server burden, a common option being to sample every 5 minutes or 10 minutes. Shorter time intervals may provide more detailed data, but also result in increased data volume and processing complexity. Subsequently, the smart door magnetic sensor will periodically record the status of the door, i.e., open or closed, according to a preset time interval. These status information are recorded in the form of digital signals, which may be generally indicated as "1" for door open and "0" for door closed. Thus, each sampling time point generates a corresponding gate status data. Over time, these gate status data will constitute a time series data set. This dataset records the state of the gate at a series of time points, including the time stamp and the corresponding gate state. For example:

2023-08-18:08:00:00-door closed;

2023-08-18:08:10:00-door open;

2023-08-18:08:20:00-door closed;

2023-08-18:08:30:00-door closed;

2023-08-18:08:40:00-door open;

2023-08-18:08:50:00-door closed;

through this time series data set, the server knows the activity of the old in the home in real time. For example, in 8:10 a.m., the server records that the door is open, indicating that the elderly is going out. And at 8:30, the door is closed again, indicating that the elderly is moving indoors. Such time series data can be used not only for real-time monitoring, but also for subsequent analysis. By analyzing the data, the server calculates the characteristics of the old people such as the activity duration, the switching frequency of the door and the like. These features help determine the daily activity patterns of the elderly, thereby providing more accurate care and reminders. For example, if the server finds that the door switching frequency of the elderly is abnormally reduced, meaning that the activity of the elderly is reduced, the server generates a corresponding alert, encouraging the elderly to properly activity.

S102, performing door body switching frequency characteristic analysis on intelligent door magnetic state data to obtain door body switching frequency characteristic data;

specifically, in the door body switching frequency characteristic analysis process, the server needs to process original intelligent door magnetic state data to obtain door body switching frequency characteristic data. The initial intelligent door magnetic state data contains outliers and missing values, which can affect subsequent analysis. Therefore, data needs to be cleaned. Outliers can be detected and culled, and missing values can be filled in by interpolation methods to maintain data continuity and accuracy. The cleaned data will form a standard state sequence recording the change of the door's open and close state with time. This standard state sequence will be the basis for subsequent analysis. The standard state sequence may be partitioned according to a certain rule into a plurality of sub-state sequence data. By doing so, the switching behavior of the door body in different time periods can be captured better. The switching frequency, i.e. the ratio of the number of times the gate is switched to time, is calculated separately for each sub-state sequence data. This will result in switching frequency data for a plurality of sub-state sequence data. And carrying out frequency characteristic extraction on the switching frequency data of each sub-state sequence data. This may include calculating statistics of average switching frequency, highest switching frequency, and lowest switching frequency to characterize the frequency of the gate switches. And fusing the first frequency characteristic data of each sub-state sequence data to obtain second frequency characteristic data. This process includes weighted averaging or other integration of the features of the individual sub-states to integrate the switching frequency information for the different time periods. The second frequency characteristic data is fourier transformed and converted into the frequency domain. Spectral analysis is performed to obtain target spectral data, which will reveal intensity information of the different frequency components. And analyzing the target frequency spectrum data and extracting frequency component information in the target frequency spectrum data. This may help determine the dominant frequency component present in the gate switching frequency. And carrying out periodical analysis according to the frequency component data to identify a periodical pattern. This helps to capture the periodic behavior of the door switch. And combining the periodic frequency characteristic with the second frequency characteristic data to generate final door body switching frequency characteristic data. This characteristic data will take into account the switching frequency, periodicity and frequency component information of the gate body in combination. And finally, the server extracts rich door body switching frequency characteristic information from the intelligent door magnetic state data. Such characteristic information can reveal daily activity patterns of the elderly, further for monitoring, care and reminder. For example, consider an elderly person living in their own home, where a smart door magnetic sensor is installed. The door is opened and closed at different times of the day as follows:

08:00 AM: closing the door;

09:00 AM: opening the door;

09:30 AM: closing the door;

12:00 PM: opening the door;

12:30 PM: closing the door;

02:00 PM: opening the door;

04:00 PM: closing the door;

in this embodiment, the server will first clean the data, remove outliers and interpolate missing values to obtain standard state sequence data. The standard state sequence data is partitioned into different sub-state sequences, such as morning, noon and afternoon. For each sub-state sequence, the server calculates a switching frequency, e.g., 1 for the door open at noon, 1 for the door close, and 0.5 for the switching frequency. Likewise, the server calculates the switching frequency for other time periods. Next, statistical feature extraction is performed on the switching frequency data of each sub-state sequence, for example, calculating an average switching frequency, a highest switching frequency, and a lowest switching frequency. Assume that in this embodiment, the average switching frequency in the morning is 0.5, the highest switching frequency is 1, and the lowest switching frequency is 0.5. The server fuses the features to obtain second frequency feature data. And carrying out Fourier transform and spectrum analysis on the second frequency characteristic data to obtain target spectrum data. Suppose that the spectral analysis shows that the dominant frequency component is 0.5Hz. Through the periodic analysis, the server recognizes that the main period of the door switch is 2 seconds (reciprocal of 0.5 Hz). The server generates final gate switching frequency characteristic data, which will include switching frequency, periodicity, and frequency component information, by integrating the periodic frequency characteristic and the second frequency characteristic data.

S103, performing activity duration characteristic analysis on the old user according to the time sequence data and the intelligent door magnetic state data to obtain activity duration characteristic data;

the server aligns the time series data with the intelligent door magnetic state data to determine the door switch state change. By recording the continuous switch state changes, the server calculates the time length of each time the old user enters and exits the room. This will generate a plurality of activity duration data, each representing the duration of an activity in and out of the room. And for the plurality of activity duration data, the server performs activity duration feature extraction to acquire information about the activity habits of the old user. These characteristics may include average activity duration, longest activity duration, shortest activity duration, etc. The average activity duration may reflect typical activity durations of elderly users, while the longest and shortest activity durations may reveal ranges of activity durations. The server calls a preset probability density distribution function, performs distribution operation on a plurality of activity duration data, and generates an activity duration probability density distribution map. The graph can display the occurrence probability of different activity duration values, thereby helping to understand the distribution situation of the activity duration of the old user. By analyzing the activity duration probability density distribution map, the server compares the activity duration distribution over different time periods. This helps identify the activity patterns of the elderly user over different time periods, e.g. the difference in the day and night activity duration profiles. And generating final activity duration characteristic data by the server according to the activity duration distribution characteristics and the initial activity duration characteristics such as the average activity duration, the longest activity duration, the shortest activity duration and the like extracted before. The characteristic data comprehensively considers the activity duration habit, the activity mode and the distribution condition of the activity duration of the old people. For example, consider an elderly person living in a home, and a smart door magnetic sensor records the status of his switch to and from the room. And the server performs characteristic analysis of the activity duration through the time sequence data and the intelligent door magnetic state data. At 8 am, the old left the bedroom and the door state changed from closed to open. At 8:15 minutes, the old returns to the bedroom, and the door state changes from open to closed. At 9:30 minutes, the old left the bedroom again, and the door state changed from closed to open. At 11, the elderly returns to the bedroom and the door state changes from open to closed. With data alignment, the server calculates that the first exit from the bedroom lasted 15 minutes, the second exit lasted 1 hour 30 minutes, and finally the return to the bedroom lasted 1 hour 30 minutes. In the activity duration feature extraction stage, the server calculates the average activity duration to be 1 hour, the longest activity duration to be 1 hour and 30 minutes and the shortest activity duration to be 15 minutes. The server calls a probability density distribution function, calculates the activity duration data and generates an activity duration probability density distribution map. By analyzing the graph, the server finds that most of the activity duration of the elderly user is concentrated around 1 hour, but there are also some shorter and longer activity durations. Further, the server compares the activity duration distributions in different time periods, and finds that the activity duration distributions of the old user in the morning and the evening are different because the old user is more active in the morning and is quieter in the evening. And integrating the information such as the activity duration distribution characteristics, the average, the longest activity duration and the shortest activity duration, and the like, and generating final activity duration characteristic data by the server. Such data may provide important information to the care monitoring system regarding the length of daily activities of the elderly user for generating care and reminders for individual needs.

S104, performing feature value mapping on the door body switching frequency feature data to generate a switching feature vector, performing feature value mapping on the active duration feature data to generate a duration feature vector, and performing matrix conversion on the switching feature vector and the duration feature vector to obtain a target feature matrix;

specifically, the door body switching frequency characteristic data is subjected to standardized processing so as to ensure that the data are in the same scale range. And mapping the characteristic values of the standardized frequency characteristic data by using a preset first mapping function to obtain a plurality of first characteristic values. And similarly, carrying out standardization processing on the characteristic data of the activity duration, and carrying out characteristic value mapping by using a preset second mapping function to obtain a plurality of second characteristic values. Based on the first plurality of eigenvalues, a first plurality of vector elements are generated, which elements will constitute a switching eigenvector. Also, a plurality of second vector elements are generated from the plurality of second feature values to construct a duration feature vector. And carrying out weighting operation on the elements of the switch characteristic vector by using a preset first weight data set to obtain the weighted switch characteristic vector. And similarly, weighting the elements of the time length feature vector by using a preset second weight data set to generate a weighted time length feature vector. And combining the weighted switch characteristic vector and the weighted duration characteristic vector into a matrix, wherein the matrix is the target characteristic matrix. The matrix comprehensively reflects the door body switch frequency characteristic and the activity duration characteristic of the old user and can be used in subsequent analysis and care. For example, assuming an elderly person is in a home, the smart door magnetic sensor records the on-off status and the activity duration of his entering and exiting the room. Through the previous steps, the server has extracted switching frequency characteristic data and activity duration characteristic data from the data. Switching frequency characteristic data (after normalization and characteristic value mapping): 0.2,0.8,0.6, activity duration feature data (normalized and feature value mapped): [0.4,0.6,0.2]. The server is also provided with a first weight data set and a second weight data set which are preset and used for weighting operation. A first weight data set: [0.3,0.5,0.2], second weight dataset: [0.6,0.3,0.1]. And carrying out weighted operation on the switching characteristic vector elements according to the first weight data set: weighted switching eigenvector elements= [0.2x0.3,0.8x0.5,0.6x0.2] = [0.06,0.4,0.12]. Likewise, the time-long feature vector elements are weighted according to the second weight data set: weighted duration feature vector elements= [0.4x0.6,0.6x0.3,0.2x0.1] = [0.24,0.18,0.02]. Combining the weighted switch eigenvector elements and duration eigenvector elements into a target eigenvector matrix: the target feature matrix= [ [0.06,0.24], [0.4,0.18], [0.12,0.02] ], integrates the switching frequency features and the activity duration features of the elder user, and can be used for subsequent analysis, care and reminding. For example, the server determines whether the activity pattern of the elderly user is normal based on this feature matrix, thereby providing personalized care and advice.

S105, inputting the target feature matrix into a preset old people activity anomaly analysis model to perform old people activity anomaly analysis, and obtaining an old people activity anomaly analysis result;

specifically, the server trains an isolated forest model through a preset isolated forest library. The isolated forest is an anomaly detection method based on random forests and is used for detecting anomaly samples in data. In the model training process, a model parameter set needs to be set so as to ensure that the model can accurately capture abnormal modes in data. And inputting the target feature matrix generated by the previous step into a trained old man activity anomaly analysis model. The model has the capability of identifying abnormal data and can be used for judging whether the activity of the old is abnormal or not. And calculating the first anomaly score of each column vector in the target feature matrix through the old people activity anomaly analysis model. This score reflects the degree of anomaly of each feature throughout the dataset. A larger anomaly score indicates that the corresponding feature is anomalous in the activity. And carrying out mean value operation on the first abnormal score of each column vector to obtain a second abnormal score of the target feature matrix. The score summarizes the abnormal conditions of the characteristics and is the basis for judging whether the overall activity is abnormal or not. And comparing the second anomaly score with a preset target threshold. The target threshold is a predetermined limit for demarcating points of abnormal and normal activity. If the second anomaly score is greater than the target threshold, the server will determine that an anomaly exists in the senior citizen's activity. If the second anomaly score is less than or equal to the target threshold, the server will determine that there is no anomaly in the senior citizen's activity. For example, assume that the server has trained an orphan forest model and has a preset target threshold. There is now a target feature matrix, which is processed through the previous steps as follows: target feature matrix = [ [0.06,0.24], [0.4,0.18], [0.12,0.02] ]. Assume that the preset target threshold is 0.3. Performing isolated forest model analysis on each column vector, and calculating a first anomaly score: the first anomaly score= [0.7,0.6], and performing mean operation on the first anomaly score to obtain a second anomaly score: second anomaly score= (0.7+0.6)/2=0.65. Comparing the second anomaly score to a target threshold: 0.65>0.3, and the server judges that the activity of the old is abnormal. In this embodiment, the server calculates the anomaly score of the activity by inputting the target feature matrix into the senior citizen activity anomaly analysis model, and then compares the anomaly score with a preset target threshold. And the server judges that the activity of the old people has abnormal conditions because the second abnormal score is larger than the target threshold value.

S106, matching the corresponding target monitoring strategies according to the analysis results of the activity anomalies of the old, generating corresponding intelligent reminding information according to the target monitoring strategies, and carrying out intelligent reminding on the old through the target gate.

Specifically, the server obtains a preset monitoring policy library, which includes a plurality of candidate monitoring policies. These strategies may provide different care and support for different elderly activity anomalies. Identification information is created for each candidate monitoring policy to relate it to the second anomaly score. This identification information may be a textual description, number, etc. for matching the care policy to the abnormal situation. And according to the result of the activity anomaly analysis of the old and the second anomaly score, the server matches the applicable target monitoring strategy from the plurality of candidate monitoring strategies. The basis for the match may be the range of anomaly scores, the severity of the anomaly, etc. The server will select the most appropriate policy to provide targeted care and alerts. Based on the matched target monitoring strategy, the server generates corresponding intelligent reminding information. Such information may include text, speech, images, etc. for conveying care and advice to the elderly user. The reminder should be able to clearly communicate the abnormal situation, the measures taken and the solution. Finally, the server carries out intelligent reminding through the intelligent door body. This may be achieved by means of sound, light, screen display, etc. When the old user enters or leaves the room through the door body, the server triggers corresponding reminding, so that the user can learn important information about abnormal conditions on the premise of not disturbing daily activities. For example, assume that the monitoring policy library contains the following two candidate monitoring policies: strategy a: mild activity abnormality, identification information: mild abnormalities, corresponding range of abnormality scores: 0.3-0.5, reminding information: your activities are somewhat abnormal, please take care of resting and keep proper exercise. Policy B: serious activity anomaly, identification information: severe anomalies, corresponding anomaly score ranges: 0.5-0.8, reminding information: you have serious abnormal activities and ask to contact with medical staff as soon as possible to get further support. Assume that the result of the abnormality analysis of the activity of the elderly is that the second abnormality score is 0.6. The server will match policy B based on this score and the anomaly since 0.6 is within the anomaly score of policy B. The server generates intelligent reminding information corresponding to the strategy B. When an old man user enters a room through the intelligent door body, the server triggers a reminder, for example, through voice playing: the respectful user, according to the analysis of the server, has serious abnormal activity condition, please contact with medical staff in time to obtain further support. In this embodiment, the server matches an appropriate monitoring policy according to the analysis result of the activity anomaly of the elderly, generates corresponding intelligent reminding information, and transmits the reminding to the elderly through the intelligent portal to provide necessary care and support.

In the embodiment of the invention, intelligent door magnetic state data of a target door body are acquired, and time sequence data are acquired; performing door body switching frequency characteristic analysis on the intelligent door magnetic state data to obtain door body switching frequency characteristic data; performing activity duration feature analysis to obtain activity duration feature data; performing eigenvalue mapping, generating a switch eigenvector and a duration eigenvector, and performing matrix conversion to obtain a target eigenvector; inputting the target feature matrix into an old man activity anomaly analysis model to perform old man activity anomaly analysis, so as to obtain an old man activity anomaly analysis result; according to the method, the system and the device, the target monitoring strategy is matched according to the abnormal activity analysis result of the old, and intelligent reminding is performed according to the target monitoring strategy, and the state change of the old entering and exiting a room is monitored in real time through the intelligent door magnetic data, so that the activity track and habit of the old can be accurately captured, and the guardian can know the life rule of the old. By analyzing the door body switching frequency characteristic and the activity duration characteristic, the method can identify abnormal conditions, such as long-time inactivity, frequent in-out and the like, and early warn potential health problems or safety risks in time. Can monitor the activity habit of old man, in time discover to fall down, unexpected circumstances such as lost, reduce the emergence of unexpected accidents through intelligent warning, match individualized guardianship tactics through unusual analytical model according to the actual conditions of old man, remind to specific activity period for example, improve guardianship's effectiveness, and then realized carrying out intelligent early warning to old man user to the rate of accuracy of early warning has been improved.

In a specific embodiment, the process of executing step S102 may specifically include the following steps:

(1) Abnormal value removal and missing value interpolation are carried out on the intelligent door magnetic state data, and standard state sequence data are obtained;

(2) Dividing the standard state sequence data to obtain a plurality of sub-state sequence data, and respectively calculating the switching frequency of the plurality of sub-state sequence data to obtain the switching frequency data of each sub-state sequence data;

(3) Frequency characteristic extraction is carried out on the switching frequency data of each sub-state sequence data to obtain first frequency characteristic data of each sub-state sequence data, wherein the first frequency characteristic data comprises average switching frequency, highest switching frequency and lowest switching frequency;

(4) Performing feature fusion on the first frequency feature data of each sub-state sequence data to obtain second frequency feature data;

(5) Performing Fourier transform on the second frequency characteristic data to obtain target frequency domain data, and performing spectrum analysis on the target frequency domain data to obtain target spectrum data;

(6) Frequency component analysis is carried out on the target frequency spectrum data to obtain frequency component data, and periodic analysis is carried out on the frequency component data to obtain periodic frequency characteristics;

(7) And generating corresponding gate switching frequency characteristic data according to the periodic frequency characteristic and the second frequency characteristic data.

Specifically, the server removes abnormal values of the intelligent door magnetic state data, and eliminates abnormal data points so as to ensure the accuracy of subsequent analysis. For missing values, a proper interpolation method is adopted for filling so as to ensure the integrity of the data. After outlier removal and outlier interpolation, a set of standard state sequence data is obtained, which will be used for subsequent analysis and feature extraction. The standard state sequence data is divided into a plurality of sub-state sequence data. Each sub-state sequence data corresponds to a switching condition of the intelligent door magnetic state for a continuous time. For each sub-state sequence data, its switching frequency is calculated. The switching frequency represents the number of times the door is opened and closed in a unit time, usually in Hz. And carrying out frequency characteristic extraction on the switching frequency data of each sub-state sequence data. This includes calculating the average switching frequency, the highest switching frequency, the lowest switching frequency, etc. And fusing the frequency characteristics of each sub-state sequence data to obtain second frequency characteristic data. This feature will integrate the frequency information of the multiple sub-state sequence data, providing a more comprehensive frequency signature. And carrying out Fourier transform on the second frequency characteristic data to obtain target frequency domain data. And carrying out frequency spectrum analysis on the frequency domain data, and analyzing the component distribution condition of the data in the frequency domain. Frequency component data is extracted from the target spectrum data, and the main frequency components of the data are analyzed. And then, periodically analyzing the frequency component data to obtain periodic frequency characteristics. And generating final gate body switching frequency characteristic data according to the periodic frequency characteristic and the second frequency characteristic data. These characteristic data will comprehensively reflect the frequency characteristics of the door opening and closing condition for subsequent analysis and care. For example, assume that there is a set of intelligent door magnetic state data, which is subjected to outlier removal and missing value interpolation as follows: standard state sequence data: [0,1,1,0,1,0,0,1,1,1,0,1,1,0,0]. The server partitions the set of standard state sequence data into 3 sub-state sequence data: [0,1,1],[0,1,0,0],[1,1,0,1,1,0,0]. The switching frequency of each sub-state sequence data is respectively as follows: 1/3Hz,0.25Hz,0.4285Hz. Frequency characteristic extraction: sub-state sequence 1: average switching frequency=1/3 Hz, highest switching frequency=1/3 Hz, lowest switching frequency=1/3 Hz; sub-state sequence 2: average switching frequency=0.25 Hz, highest switching frequency=0.25 Hz, lowest switching frequency=0.25 Hz; sub-state sequence 3: average switching frequency= 0.4285Hz, highest switching frequency= 0.4285Hz, lowest switching frequency= 0.4285Hz. Obtaining second frequency characteristic data after characteristic fusion: [1/3,0.25,0.4285]. And obtaining target spectrum data after Fourier transformation and spectrum analysis. The main frequency component is extracted from the target spectrum data, for example, the main frequency is 0.1Hz. And judging whether the frequency component shows periodicity or not through periodicity analysis, and if so, obtaining the periodicity frequency characteristic. And integrating the second frequency characteristic data and the periodic frequency characteristic to generate final gate body switching frequency characteristic data.

In a specific embodiment, as shown in fig. 2, the process of performing step S103 may specifically include the following steps:

s201, data alignment is carried out on time sequence data and intelligent door magnetic state data, and the time length of each time of room entering and exiting activities of an old user is calculated, so that a plurality of activity time length data of switch state changes are obtained;

s202, extracting activity duration features of a plurality of pieces of activity duration data to obtain a plurality of initial activity duration features, wherein the plurality of initial activity duration features comprise: average activity duration, longest activity duration, and shortest activity duration;

s203, a preset probability density distribution function is called to carry out distribution operation on a plurality of activity duration data, and an activity duration probability density distribution map is generated;

s204, comparing the time periods of the probability density distribution map of the activity duration to obtain the distribution characteristics of the activity duration;

s205, generating corresponding activity duration feature data according to the activity duration distribution feature and the plurality of initial activity duration features.

Specifically, the server aligns the time sequence data and the intelligent door magnetic state data, so that the time stamps of the two data sources are consistent. And calculating the activity duration of the old user entering and exiting the room each time according to the switch change of the door magnetic state. The activity duration refers to the time interval from entering the room to leaving the room again. And forming a list by the calculated activity duration data, wherein the list comprises a plurality of activities of the elder user entering and exiting the room, and the duration of each activity is recorded. And extracting the characteristics of the plurality of activity duration data to obtain a plurality of initial activity duration characteristics. These characteristics may include average activity duration, longest activity duration, shortest activity duration, etc. And calling a preset probability density distribution function, and calculating a plurality of activity duration data to generate an activity duration probability density distribution map. This profile will reflect the profile of the length of time the elderly user is moving into and out of the room. And analyzing the probability density distribution map of the activity duration, and comparing the activity duration distribution conditions in different time periods. From which the activity duration distribution characteristics, such as peak hours, low peak hours, etc., can be extracted. And generating final activity duration characteristic data according to the activity duration distribution characteristics and the plurality of initial activity duration characteristics. The data comprehensively reflects the time length conditions of the activities of the elder users entering and exiting the room, and provides basis for subsequent analysis and care. For example, assume that there is a set of activity duration data, which is calculated and extracted as follows: activity duration data: [10, 15, 20,8, 25, 12]. The server extracts the characteristics of the activity duration data to obtain initial activity duration characteristics: average activity duration: (10+15+20+8+25+12)/6=15 Hz, longest active duration: 25Hz, shortest activity duration: 8Hz. And calling a probability density distribution function, and carrying out distribution operation on the activity duration data to obtain an activity duration probability density distribution map. From the graph, it can be found that the activity duration distribution is higher between 8 and 10 pm. By analyzing the probability density distribution diagram, the activity duration distribution characteristics can be obtained, namely, the old user can more perform activities of entering and exiting a room between 8 and 10 points at night. The server generates final activity duration characteristic data by combining the initial activity duration characteristic and the activity duration distribution characteristic, and the data can provide important information for the subsequent monitoring strategy formulation and reminding generation.

In a specific embodiment, as shown in fig. 3, the process of executing step S104 may specifically include the following steps:

s301, carrying out standardization processing on the door body switching frequency characteristic data to obtain standardized frequency characteristic data, and carrying out characteristic value mapping on the standardized frequency characteristic data through a preset first mapping function to obtain a plurality of first characteristic values;

s302, generating a plurality of first vector elements according to the first characteristic values, and generating a switching characteristic vector according to the first vector elements;

s303, carrying out standardization processing on the activity duration feature data to obtain standardized duration feature data, and carrying out feature value mapping on the standardized duration feature data through a preset second mapping function to obtain a plurality of second feature values;

s304, generating a plurality of second vector elements according to the plurality of second characteristic values, and generating a duration characteristic vector according to the plurality of second vector elements;

s305, carrying out weighting operation on the switch characteristic vector according to a preset first weight data set to obtain a weighted switch characteristic vector, and carrying out weighting operation on the time length characteristic vector according to a preset second weight data set to obtain a weighted time length characteristic vector;

S306, performing matrix conversion on the weighted switch feature vector and the weighted duration feature vector to obtain a target feature matrix.

Specifically, the server performs standardization processing on the door body switching frequency characteristic data, normalizes the door body switching frequency characteristic data into data with a mean value of 0 and a standard deviation of 1, and obtains standardized frequency characteristic data. And mapping the characteristic values of the standardized frequency characteristic data by using a preset first mapping function to obtain a plurality of first characteristic values. And forming a vector by the plurality of first eigenvalues to generate a first eigenvalue vector. A switching eigenvector is generated using the first eigenvalue vector. And carrying out standardization processing on the activity duration characteristic data to enable the activity duration characteristic data to have a distribution with a mean value of 0 and a standard deviation of 1, so as to obtain the standardized duration characteristic data. And mapping the characteristic values of the standardized duration characteristic data by using a preset second mapping function to obtain a plurality of second characteristic values. And forming a plurality of second eigenvalues into a vector to generate a second eigenvalue vector. And generating a duration feature vector by using the second feature value vector. And carrying out weighting operation on the switch characteristic vector by using a preset first weight data set to obtain a weighted switch characteristic vector. And weighting the time length feature vector by using a preset second weight data set to obtain a weighted time length feature vector. And performing matrix conversion on the weighted switch feature vector and the weighted duration feature vector to generate a target feature matrix. This matrix will reflect the weighting of the switch characteristics and the duration characteristics in combination. For example, assume that after feature value mapping, the resulting first feature value vector is [0.5,0.7,0.3 ] ]The second eigenvalue vector is [0.8,0.6,0.9 ]]. Assume that the preset first weight data set is [0.4,0.3,0.5 ]]The second weight data set is [0.7,0.2,0.6 ]]. The weighted switch characteristic vector is obtained by weight weighting operation:，，/>。

the weighted duration feature vector is:，，/>. Performing matrix conversion on the weighted switch characteristic vector and the weighted duration characteristic vector to obtain a target characteristic matrix:

[0.56，0.82]；

[0.62，0.46]；

[0.47，0.79]；

through the steps of standardization, eigenvalue mapping, weight weighting, matrix conversion and the like, a target eigenvalue matrix reflecting the switch characteristics and the duration characteristics can be generated, and more comprehensive characteristic information is provided for the care monitoring server of the aged.

In a specific embodiment, as shown in fig. 4, the process of performing step S105 may specifically include the following steps:

s401, performing model training on an isolated forest model through a preset isolated forest library, and setting a model parameter set of the isolated forest model to obtain an old man activity anomaly analysis model;

s402, inputting the target feature matrix into an old man activity anomaly analysis model, and calculating a first anomaly score of each column vector in the target feature matrix through the old man activity anomaly analysis model;

s403, carrying out mean value operation on the first abnormal score of each column vector to obtain a second abnormal score of the target feature matrix;

S404, comparing the second anomaly score with a preset target threshold, and if the second anomaly score is larger than the target threshold, determining that the old activity anomaly analysis result is that the old activity anomaly exists;

and S405, if the second anomaly score is smaller than or equal to the target threshold value, determining that the old people activity anomaly analysis result is that the old people activity is not abnormal.

In particular, the server prepares a library of pre-set isolated forests, typically containing implemented isolated forest algorithms. The prepared training data (which are some normal activity data) is trained by using a preset isolated forest library to construct an isolated forest model. During the training process, a set of model parameters, such as the number of trees, the number of sub-sampled samples, etc., may be set. And taking the generated target feature matrix as input, and transmitting the generated target feature matrix into the constructed isolated forest model. And performing isolated forest model calculation on each column vector (corresponding to different features) in the target feature matrix to obtain a first anomaly score of each feature. And carrying out mean value operation on the first abnormal scores of each feature to obtain a summarized second abnormal score. This score will reflect the degree of abnormality of the whole. And comparing the calculated second anomaly score with a preset target threshold. And if the second abnormal score is larger than the target threshold, indicating that the activity of the old is abnormal, and if the second abnormal score is smaller than or equal to the target threshold, indicating that the activity of the old is not abnormal. For example, assume that an isolated forest model has been constructed using a preset isolated forest library and has been trained. According to the previous steps, a target feature matrix is obtained, wherein each column of vectors corresponds to a first anomaly score of a feature. For example, assume that the first anomaly score is calculated as follows: first anomaly score for feature 1: 0.9, first anomaly score for feature 2: 0.8, first anomaly score for feature 3: 0.7. and carrying out average operation on the first anomaly scores to obtain second anomaly scores: (0.9+0.8+0.7)/3=0.8. Assume that the preset target threshold is 0.75. Since the second abnormality score (0.8) is greater than the target threshold (0.75), it can be determined that there is an abnormality in the activity of the elderly according to the determination rule.

In a specific embodiment, the process of executing step S106 may specifically include the following steps:

(1) Obtaining a monitoring policy bank, wherein the monitoring policy bank comprises a plurality of candidate monitoring policies;

(2) Creating identification information of each candidate monitoring strategy in the monitoring strategy library, and constructing a target corresponding relation between the identification information and the second abnormal score;

(3) According to the result of the activity anomaly analysis of the old and the second anomaly score, matching a corresponding target monitoring strategy from a plurality of candidate monitoring strategies;

(4) And generating corresponding intelligent reminding information according to the target monitoring strategy, and carrying out intelligent reminding on the old people through the target gate body.

Specifically, a plurality of candidate monitoring strategies are obtained from a previously prepared monitoring strategy library. These policies may be alerts and treatment schemes for different abnormal situations, such as suddenly leaving the room, inactivity for a long period of time, etc. Unique identification information is created for each candidate monitoring policy, ensuring differentiation between policies. And constructing a target corresponding relation table, and associating the identification information of the monitoring strategy with the second abnormal score. This will facilitate the subsequent matching process. And selecting a proper target monitoring strategy from the monitoring strategy library according to a preset matching rule by utilizing the result of the activity abnormality analysis of the old and the second abnormality score. The matching rules are based on factors such as anomaly score range, anomaly type, etc. And extracting corresponding intelligent reminding information from the strategy according to the matched target monitoring strategy. Such information may be in the form of text, sound, images, etc. for conveying critical information to the elderly user. And conveying the generated intelligent reminding information to the old people through the intelligent door body server. This can be achieved by means of a display screen, voice prompts, etc. on the door body so that the elderly user can timely know the abnormal situation and take necessary actions. For example, assume that there are two candidate monitoring strategies in the monitoring strategy library: strategy a: when the second anomaly score is greater than 0.7 and the activity anomaly analysis result of the old is anomaly, sending a voice prompt: "respectful user, you are abnormal in activities, ask you to confirm if help is needed. "policy B: when the second anomaly score is greater than 0.6 and the activity anomaly analysis result of the old is anomaly, displaying a text prompt: "you have not been active for a long time, please ensure your safety. "assuming that the result of the abnormality analysis of the activity of the elderly is abnormality, the second abnormality score is 0.75. According to the matching rule, policy a can be matched because the anomaly score satisfies a condition greater than 0.7. Generating intelligent reminding information according to the strategy A: "respectful user, you are abnormal in activities, ask you to confirm if help is needed. The generated intelligent reminding information is transmitted to the old people through the intelligent gate system, the old people receive the voice reminding, know the abnormal situation in time, and take proper measures according to the situation.

In a specific embodiment, the process of performing the gate detection method for care monitoring of elderly people may further specifically include the steps of:

(1) After intelligent reminding is carried out, door body switching pressure data of the target door body are collected, feedback parameters of the second abnormal score are calculated according to the door body switching pressure data, and a third abnormal score is generated;

(2) Performing strategy gradient analysis on the elderly users based on the third anomaly score, and determining strategy gradients corresponding to the elderly users;

(3) And carrying out strategy updating through the strategy gradient and the third anomaly score to generate a feedback monitoring strategy.

Specifically, after intelligent reminding is carried out, the opening and closing state of the target door body is monitored, and door body opening and closing pressure data are collected in real time. These data reflect the actual operation of the elderly user. And calculating feedback parameters by using the collected door body switch pressure data and combining the second abnormal fraction and other influencing factors. The parameters can reflect feedback behaviors of the old people, such as whether to respond to intelligent reminding, frequency of operating the door body and the like. And generating a third anomaly score based on the calculated feedback parameter in combination with the second anomaly score. The third anomaly score comprehensively considers the feedback behavior and the anomaly condition of the old user and provides more comprehensive information for policy updating. And carrying out strategy gradient analysis by using the third anomaly score, and analyzing the influence of the behavior of the old user on the anomaly score. This will help determine the policy gradient to which the elderly user corresponds, i.e. how to adjust the monitoring policy to better adapt to the habits and feedback of the elderly user. And updating the monitoring strategy by combining the strategy gradient and the third anomaly score. According to the analysis result, the reminding frequency, the reminding mode, the threshold setting and the like can be adjusted so as to provide more accurate monitoring service. For example, assume that after intelligent reminding, pressure data of an old man user operating the door body is collected. The calculated feedback parameter is 0.85, representing the positive degree of the user to respond to the prompt. Meanwhile, the second anomaly score is 0.8, and the calculation formula of the third anomaly score is as follows: third anomaly score = 0.8 x 0.85 = 0.68. Based on the third anomaly score, policy gradient analysis finds that the elderly user responds more positively to door body operation after intelligent reminding, but the anomaly score has limited descending amplitude due to fewer operation times. Thus, the strategic gradient analysis suggests an appropriate increase in alert frequency to better guide user behavior. In the strategy updating stage, according to the proposal of strategy gradient analysis, the monitoring strategy is adjusted: the alert frequency is adjusted from once per hour to once per half hour and provides a more pronounced alert pattern such as a sound and flashing display.

The method for detecting a door for care of an elderly person in the embodiment of the present invention is described above, and the system for detecting a door for care of an elderly person in the embodiment of the present invention is described below, referring to fig. 5, one embodiment of the system for detecting a door for care of an elderly person in the embodiment of the present invention includes:

the acquiring module 501 is configured to acquire intelligent door magnetic state data of a target door body based on a preset sampling time interval, and acquire time sequence data corresponding to the intelligent door magnetic state data;

the frequency analysis module 502 is configured to perform door body switching frequency characteristic analysis on the intelligent door magnetic state data to obtain door body switching frequency characteristic data;

a duration analysis module 503, configured to perform activity duration feature analysis on the old user according to the time sequence data and the intelligent door magnetic state data, to obtain activity duration feature data;

the conversion module 504 is configured to perform eigenvalue mapping on the gate switching frequency characteristic data to generate a switching eigenvector, perform eigenvalue mapping on the active duration characteristic data to generate a duration eigenvector, and perform matrix conversion on the switching eigenvector and the duration eigenvector to obtain a target eigenvector;

The anomaly analysis module 505 is configured to input the target feature matrix into a preset senior citizen activity anomaly analysis model to perform senior citizen activity anomaly analysis, so as to obtain an senior citizen activity anomaly analysis result;

the generating module 506 is configured to match the corresponding target monitoring policy according to the analysis result of the activity anomaly of the elderly, generate corresponding intelligent reminding information according to the target monitoring policy, and intelligently remind the elderly user through the target gate.

Acquiring intelligent door magnetic state data of a target door body and acquiring time sequence data through the cooperative cooperation of the components; performing door body switching frequency characteristic analysis on the intelligent door magnetic state data to obtain door body switching frequency characteristic data; performing activity duration feature analysis to obtain activity duration feature data; performing eigenvalue mapping, generating a switch eigenvector and a duration eigenvector, and performing matrix conversion to obtain a target eigenvector; inputting the target feature matrix into an old man activity anomaly analysis model to perform old man activity anomaly analysis, so as to obtain an old man activity anomaly analysis result; according to the method, the system and the device, the target monitoring strategy is matched according to the abnormal activity analysis result of the old, and intelligent reminding is performed according to the target monitoring strategy, and the state change of the old entering and exiting a room is monitored in real time through the intelligent door magnetic data, so that the activity track and habit of the old can be accurately captured, and the guardian can know the life rule of the old. By analyzing the door body switching frequency characteristic and the activity duration characteristic, the method can identify abnormal conditions, such as long-time inactivity, frequent in-out and the like, and early warn potential health problems or safety risks in time. Can monitor the activity habit of old man, in time discover to fall down, unexpected circumstances such as lost, reduce the emergence of unexpected accidents through intelligent warning, match individualized guardianship tactics through unusual analytical model according to the actual conditions of old man, remind to specific activity period for example, improve guardianship's effectiveness, and then realized carrying out intelligent early warning to old man user to the rate of accuracy of early warning has been improved.

The above detailed description of the door detection system for care of the elderly in the embodiment of the present invention is given in fig. 5 from the point of view of the modularized functional entity, and the following detailed description of the door detection device for care of the elderly in the embodiment of the present invention is given in terms of hardware processing.

Fig. 6 is a schematic structural diagram of a door detection device for care of the elderly people, where the door detection device 600 for care of the elderly people may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 610 (e.g., one or more processors) and a memory 620, one or more storage media 630 (e.g., one or more mass storage devices) storing applications 633 or data 632. Wherein the memory 620 and the storage medium 630 may be transitory or persistent storage. The program stored on the storage medium 630 may include one or more modules (not shown), each of which may include a series of instruction operations in the gate detection device 600 for geriatric care monitoring. Still further, the processor 610 may be configured to communicate with the storage medium 630 to execute a series of instruction operations in the storage medium 630 on the door check device 600 for geriatric care monitoring.

The gate detection device 600 for geriatric care monitoring may also include one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input/output interfaces 660, and/or one or more operating systems 631, such as Windows Server, mac OS X, unix, linux, freeBSD, and the like. It will be appreciated by those skilled in the art that the gate detection device structure for geriatric care monitoring shown in fig. 6 is not limiting and may include more or fewer components than shown, or may be combined with certain components, or a different arrangement of components.

The present invention also provides a door body detection apparatus for senior citizen care monitoring, which includes a memory and a processor, the memory storing computer readable instructions which, when executed by the processor, cause the processor to perform the steps of the door body detection method for senior citizen care monitoring in the above embodiments.

The present invention also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, and which may also be a volatile computer readable storage medium, the computer readable storage medium having instructions stored therein, which when executed on a computer, cause the computer to perform the steps of the gate detection method for geriatric care monitoring.

It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a random access memory (random access memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A door detection method for care of elderly people, the method comprising:

2. The door detection method for care monitoring of elderly people according to claim 1, wherein the performing door switching frequency characteristic analysis on the intelligent door magnetic state data to obtain door switching frequency characteristic data comprises:

3. The door detection method for care monitoring of elderly people according to claim 1, wherein the step of performing activity duration feature analysis on an elderly user according to the time series data and the intelligent door magnetic state data to obtain activity duration feature data comprises the steps of:

4. The door detection method for care monitoring of elderly people according to claim 1, wherein the performing feature value mapping on the door switching frequency feature data to generate a switching feature vector, performing feature value mapping on the activity duration feature data to generate a duration feature vector, and performing matrix conversion on the switching feature vector and the duration feature vector to obtain a target feature matrix, includes:

5. The gate detection method for care monitoring of elderly people according to claim 1, wherein the step of inputting the target feature matrix into a preset elderly people activity anomaly analysis model to perform an elderly people activity anomaly analysis to obtain an elderly people activity anomaly analysis result comprises:

6. The gate detection method for care monitoring of elderly people according to claim 5, wherein the matching the corresponding target monitoring policy according to the analysis result of the activity anomaly of the elderly people, generating corresponding intelligent reminding information according to the target monitoring policy, and intelligently reminding the elderly people through the target gate body comprises:

7. The door detection method for geriatric care monitoring of claim 1, further comprising:

8. A gate detection system for care of elderly people, the gate detection system for care of elderly people comprising:

9. A door body detection apparatus for care of the elderly, the door body detection apparatus for care of the elderly comprising: a memory and at least one processor, the memory having instructions stored therein;

the at least one processor invokes the instructions in the memory to cause the gate detection device for geriatric care monitoring to perform the gate detection method for geriatric care monitoring of any one of claims 1-7.

10. A computer readable storage medium having instructions stored thereon, which when executed by a processor, implement the gate detection method for geriatric care monitoring of any one of claims 1-7.