Computer supervision system based on edge computing model

Abstract

The present invention discloses a computer supervision system based on an edge computing model, which includes a household information monitoring unit, an edge computing model processing unit, a household management model design unit, and an Internet of Things regulation unit. By monitoring the household environment information, it realizes the determination and analysis of abnormal situations of household members, and then regulates and manages household devices. By analyzing the abnormal parts of the household environment information, it not only realizes the immediate supervision of abnormal situations, but also realizes the risk warning and early regulation of abnormal situations. It is controlled based on the Internet of Things, which is convenient for the care work of household service personnel and the emergency handling of abnormal situations, and is also convenient for guardians to monitor and trace the work of domestic service personnel. Through the edge analysis and storage of the household management plan by the edge computing model, it reduces the need for data transmission on the network and ensures the real-time, reliable, and secure supervision of the household situation by the application computer.

Description

Technical Field

[0001] The present invention relates to the technical field of computer supervision, and particularly to a computer supervision system based on an edge computing model. Background Art

[0002] Computer systems are becoming increasingly complex. When using a computer for home supervision, there are problems such as poor real-time performance, weak pertinence, and lack of comprehensiveness. It may not be able to detect and respond to newly emerging threats or abnormal situations in real time, resulting in delays or missed sudden abnormal events. It is difficult to predict abnormal risks based on the development trend of abnormal situations, unable to perform abnormal emergency handling and advance regulation, and difficult to avoid the occurrence of abnormal situations, thus unable to ensure the stable operation of family life.

[0003] In view of the above technical deficiencies, a solution is now proposed. Summary of the Invention

[0004] The purpose of the present invention is to solve the problems of comprehensive supervision and traceability of abnormal situations of family members, as well as the problems of poor real-time performance, weak pertinence, and lack of comprehensiveness in computer supervision applications. It facilitates domestic service personnel to respond promptly to abnormal situations of family members, greatly improves the care efficiency for family members, and uses an edge computing model to ensure the immediacy of abnormal handling and advance risk regulation, thereby ensuring the stable operation of family life.

[0005] In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0006] A computer supervision system based on an edge computing model, including a family information monitoring unit, an edge computing model processing unit, a family management model design unit, and an Internet of Things regulation unit. Among them, the family information monitoring unit, the edge computing model processing unit, the family management model design unit, and the Internet of Things regulation unit are signal-connected.

[0007] The family information monitoring unit is used to monitor and collect family environment information in real time.

[0008] The edge computing model processing unit is used to analyze and process family environment information, and then send the information processing result to the family management model design unit. When receiving a warning signal, it performs in-depth analysis and generates a risk prediction result, and sends the risk prediction result to the family management model design unit.

[0009] The home management model design unit constructs a home management model, generates a real-time response plan by receiving the information processing result, then analyzes the abnormal risk level, generates a warning signal through threshold comparison, and feeds the warning signal back to the edge computing model processing unit. Then, it receives the risk prediction result and generates a risk prediction plan. By combining the real-time response plan and the risk prediction plan, it generates a home management plan and sends it to the Internet of Things control unit;

[0010] The Internet of Things control unit is used to receive the home management plan and perform corresponding management and control operations. After the operations are completed, it generates a prompt signal and feeds it back to the home information monitoring unit.

[0011] Furthermore, the edge computing model processing unit applies the edge computing model for distributed data processing and analysis. The home management model design unit constructs a home management model and conducts abnormal risk analysis and determination. The specific process of the interaction between the edge computing model processing unit and the home management model design unit is as follows:

[0012] A1: The edge computing model processing unit establishes an information processing model. First, it analyzes and processes the home environment information, substitutes the home environment information into the information processing model, and outputs the information processing result to the home management model design unit;

[0013] A2: The home management model design unit establishes a home management model, generates a real-time response plan by substituting the received information processing result into the home management model;

[0014] A3: The home management model design unit then analyzes the abnormal risk level, sets a risk threshold, and through threshold comparison, when the risk threshold is exceeded, it generates a warning signal and feeds the warning signal back to the edge computing model processing unit;

[0015] A4: The edge computing model processing unit establishes a deep analysis model, deeply analyzes the information processing result, and outputs the risk prediction result to the home management model design unit;

[0016] A5: The home management model design unit then receives the risk prediction result and substitutes it into the home management model, generates a risk prediction plan, and generates a home management plan by combining the real-time response plan and the risk prediction plan.

[0017] Furthermore, the specific process of establishing the information processing model is as follows:

[0018] B1: Construct an information processing model, analyze the home environment information. Among them, the home environment information includes n indicators. Mark the data of any one indicator as indicator i, input indicator i, and output the information processing result corresponding to indicator i;

[0019] B1-1: Set the information collection period Tc. Taking the value of index i as the Y-axis and the information collection period Tc as the X-axis, establish a dynamic curve graph Si of index i - information collection period Tc;

[0020] B1-2: Preset the abnormal threshold a of index i. When the ordinate of the dynamic curve Si exceeds the abnormal threshold a, extract the corresponding curve segment and mark it as an abnormal curve segment. Presume there are m abnormal curve segments, and mark any one of the abnormal curve segments as Sj;

[0021] B1-3: Establish an abnormal analysis model for the curve segment, conduct abnormal analysis on the abnormal curve segment Sj, and obtain the abnormal evaluation coefficient YC. The specific process is as follows:

[0022] B1-31: Obtain the abnormal time difference τj between the two endpoints of the abnormal curve segment Sj, and then measure and obtain the slope of the two endpoints of the abnormal curve segment Sj. Among them, when the slope is greater than 0, mark it as the rising slope Kp; when the slope is less than 0, mark it as the falling slope Kq;

[0023] B1-32: Measure and extract the peak points and valley points of the abnormal curve segment Sj: Accumulate the number of all peak points and mark it as the number of peak points Df, accumulate the number of all valley points and mark it as the number of valley points Dg, measure the average value of the ordinates of all peak points and mark it as the peak average value Jf, and measure the average value of the ordinates of all valley points and mark it as the valley average value Jg;

[0024] B1-33: Establish a formula through the rising slope Kp, falling slope Kq, number of peak points Df, peak average value Jf, number of valley points Dg, and valley average value Jg to obtain the abnormal evaluation coefficient YC;

[0025] B1-4: When index i exceeds the preset abnormal threshold, extract the abnormal curve segment of index i in the current time period and substitute it into the abnormal analysis model to output the abnormal evaluation coefficient YCi; when index i does not exceed the preset abnormal threshold, output a no-abnormality signal and mark the abnormal evaluation coefficient YCi = 0;

[0026] B2: When all n indicators of the home environment information have been analyzed and processed, integrate the abnormal evaluation coefficients corresponding to the n indicators and mark them as the information processing result, and send them to the home management model design unit.

[0027] Furthermore, the specific process of establishing the in-depth analysis model is as follows:

[0028] C1: When receiving a warning signal, establish an in-depth analysis model and output a risk prediction result;

[0029] C1-1: First, conduct an integrated analysis on the rising slopes Kp of the m abnormal curve segments of index i:

[0030] Build a dynamic curve graph Skp of the rising slope Kp. First, obtain the average value of m rising slopes Kp and label it as the wave rising increase rate Kzp. Then, obtain the standard deviation of the curve Skp and label it as the wave rising coefficient σkp. Next, measure the differences between adjacent points on the curve Skp in sequence, label the difference between any two adjacent points as the rising slope fluctuation difference Ckp, build a dynamic curve graph Scp of the rising slope fluctuation difference Ckp, extract all the peak points and their corresponding peak point coordinates on the curve Scp, calculate the average of the ordinates of all the peak point values, and obtain the rising slope fluctuation extreme value εcp;

[0031] C1-2: Through the wave rising increase rate Kzp, the wave rising coefficient σkp, the rising slope fluctuation difference Ckp, and the rising slope fluctuation extreme value εcp, establish a formula to obtain the abnormal rising slope coefficient Kyp;

[0032] C1-3: Similarly, according to the above operations, perform an integrated analysis on the falling slopes Kq of m abnormal curve segments to obtain the abnormal falling slope coefficient Kyq;

[0033] C1-4: Obtain the coordinates, slope, and prediction time period value of the end point of the curve Si, and combine the abnormal rising slope coefficient Kyp and the abnormal falling slope coefficient Kyq to analyze the dynamic development trend of the curve Si, and establish a formula to obtain the risk prediction coefficient FYi of the index i

[0034] C2: After performing an integrated analysis on the abnormal curve segments of n indicators of the family environment information, output the risk prediction coefficients of n indicators, integrate and label them as the risk prediction results, and send them to the family management model design unit.

[0035] Further, the specific process of building the family management model is as follows:

[0036] D1: The specific process of building the management analysis model is as follows:

[0037] D1-1: The family management model design unit receives the information processing results, and the information processing results include the abnormal evaluation coefficients of n indicators. Take any information processing result YCi as the result parameter and input it into the management analysis model to analyze the abnormal situation of the current family environment information. Establish a formula through the information processing results to comprehensively evaluate n indicators and obtain the abnormal management coefficient GL;

[0038] D1-2: Presuppose that there are W types of abnormal situations in the family environment information, label any one of the abnormal situations as w, and presuppose the abnormal management coefficient corresponding to the occurrence of the abnormal situation w as GLw. Then, through the ratio of the actually measured abnormal management coefficient GL to the preset abnormal management coefficient GLw, calculate the occurrence probability Gw corresponding to the abnormal situation w;

[0039] D1-3: Sort the W probability values in descending order and set a probability threshold c. When the occurrence probability Gw of the abnormal situation w exceeds the probability threshold c, extract the abnormal situation w and generate a corresponding control instruction. When the occurrence probability Gw of the abnormal situation w does not exceed the probability threshold c, no processing is performed;

[0040] D1-4: After comparing the probabilities of the W abnormal situations with the threshold, preset to extract W0 abnormal situations that exceed the probability threshold, then generate W0 control instructions, and integrate the W0 control instructions into a management plan;

[0041] D2: The specific process of generating a real-time response plan is as follows:

[0042] D2-1: Input the abnormal evaluation coefficients of the n indicators of the information processing result into the management analysis model. First, output the probabilities corresponding to the W abnormal situations, generate corresponding control instructions and a management plan, and then mark this management plan as a real-time response plan;

[0043] Then analyze the abnormal risk level and generate a risk prediction plan.

[0044] Furthermore, the specific process of analyzing the abnormal risk level and generating a risk prediction plan is as follows:

[0045] D3: The specific process of analyzing the abnormal risk level is as follows:

[0046] D3-1: Analyze the abnormal risk level of the current home environment information, establish a formula through the information processing result to comprehensively evaluate the n indicators, and obtain the abnormal risk coefficient FX;

[0047] D3-2: Then preset an abnormal risk threshold b for the abnormal risk coefficient FX. When the abnormal risk coefficient FX exceeds the abnormal risk threshold b, generate a warning signal and feedback it to the edge computing model processing unit;

[0048] D4: The specific process of generating a risk prediction plan is as follows:

[0049] D4-1: Input the abnormal evaluation coefficients of the n indicators of the risk prediction result into the management analysis model. First, output the probabilities corresponding to the W abnormal situations, generate corresponding control instructions and a management plan, and then mark this management plan as a risk prediction plan;

[0050] D5: Generate a home management plan by combining the real-time response plan and the risk prediction plan.

[0051] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are:

[0052] Through the monitoring of home environment information, the present invention realizes the determination and analysis of abnormal situations of family members, and then conducts regulation and management of home appliances. By analyzing the abnormal parts of home environment information, real-time response solutions and risk prediction solutions are gradually obtained, and a home management plan is comprehensively generated, which not only realizes the immediate supervision of abnormal situations, but also realizes the risk warning and early regulation of abnormal situations. Based on the Internet of Things, the corresponding home appliances are controlled, which is convenient for the care work of home service personnel and the emergency handling of abnormal situations, and is also convenient for the guardians to monitor and trace the work of domestic service personnel;

[0053] Among them, through the edge computing model, data processing and storage are pushed to the edge of the computer processor, and edge analysis and storage are performed on the home management plan, reducing the need for data transmission on the network and ensuring the advantages of real-time, reliable, secure, and privacy-protected supervision of the home situation by the application computer. Brief Description of the Drawings

[0054] Figure 1 Shows a schematic diagram of the modules of the present invention;

[0055] Figure 2 Shows a schematic diagram of the process of the present invention. Detailed Embodiments

[0056] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention. Embodiment 1

[0057] As Figure 1-2 shown, a computer supervision system based on an edge computing model includes a home information monitoring unit, an edge computing model processing unit, a home management model design unit, and an Internet of Things regulation unit. Among them, the home information monitoring unit, the edge computing model processing unit, the home management model design unit, and the Internet of Things regulation unit are signal-connected;

[0058] The working steps are as follows:

[0059] S1: The home information monitoring unit monitors and collects home environment information in real time;

[0060] The home environment information includes n indicators. Mark the data of any one indicator as indicator i, input indicator i, and output the information processing result corresponding to indicator i. Set the information collection period Tc to collect the n indicators in the home environment information at regular intervals;

[0061] Among them, by monitoring the home environment information, the abnormal situations of family members are analyzed and judged, and then the home appliances are managed and controlled, so as to achieve the purpose of supervising the home situation based on the edge computing model and applying the computer system.

[0062] The home environment information includes sound parameters, and the sound parameters include sound decibel values and sound frequency values. The family members include patients with inconvenient mobility, the elderly and infants in need of care. The home appliances include air conditioners, water dispensers, etc. For example, a movable portable sound monitoring device is set beside the infant to collect sound parameters continuously for 24 hours.

[0063] S2: The edge computing model processing unit establishes an information processing model:

[0064] The edge computing model is a distributed computing architecture. Through the edge computing model, data processing and storage are pushed to the edge of the computer processor to perform edge analysis and storage on the home management solution.

[0065] S2-1: Analyze and process the home environment information through the information processing model, substitute the home environment information into the information processing model, and output the information processing result. The specific process is as follows:

[0066] S2-11: Taking the value of index i as the Y-axis and the information collection period Tc as the X-axis, establish a dynamic curve graph Si of index i - information collection period Tc.

[0067] Preset the abnormal threshold a of index i. When the ordinate of the dynamic curve Si exceeds the abnormal threshold a, extract the corresponding curve segment and mark it as an abnormal curve segment. There are m abnormal curve segments in total, and any one of the abnormal curve segments is marked as Sj.

[0068] S2-12: Establish an abnormal analysis model for the curve segment, perform abnormal analysis on the abnormal curve segment Sj, and obtain the abnormal evaluation coefficient YC:

[0069] Obtain the abnormal time difference τj between the two endpoints of the abnormal curve segment Sj, and then measure and obtain the slope of the two endpoints of the abnormal curve segment Sj. Among them, when the slope is greater than 0, it means the curve shows an upward trend, and it is marked as the rising slope Kp; when the slope is less than 0, it means the curve shows a downward trend, and it is marked as the falling slope Kq.

[0070] Measure and mark all the points with a slope of 0 on the abnormal curve segment Sj as fluctuation points, then extract the next point closest to the fluctuation point and mark it as an adjacent point, measure the slope of the adjacent point. If the slope of the adjacent point is negative, it is determined that the fluctuation point is a peak point; if the slope of the adjacent point is positive, it is determined that the fluctuation point is a valley point.

[0071] Accumulate the number of all peak points and label it as the number of peak points Df, accumulate the number of all valley points and label it as the number of valley points Dg, measure the average value of the vertical coordinates of all peak points and label it as the peak average value Jf, and measure the average value of the vertical coordinates of all valley points and label it as the valley average value Jg;

[0072] Based on the rising slope Kp, falling slope Kq, number of peak points Df, peak average value Jf, number of valley points Dg, and valley average value Jg, establish a formula to obtain the anomaly evaluation coefficient YC: Among them, λ1, λ2, and λ3 are weight factor coefficients, and λ1, λ2, and λ3 are all greater than 0;

[0073] Combine the rising slope and the falling slope to form a coefficient for the degree of abnormal fluctuation change. The higher the sum of the squares of the rising slope and the falling slope, the higher the degree of abnormal fluctuation change, and the abnormal change shows a trend of sharp increase and sharp decrease; Multiply the number of peak points by the peak average value to form a peak coefficient, multiply the number of valley points by the valley average value to form a valley coefficient, and add the peak coefficient and the valley coefficient to form a coefficient for the degree of fluctuation. The higher the coefficient for the degree of fluctuation, the more fluctuations there are in the abnormal curve; Mark the difference between the peak average value and the valley average value as the coefficient for the amplitude of fluctuation. The higher the coefficient for the amplitude of fluctuation, the higher the overall amplitude of fluctuation of the abnormal curve and the more obvious the anomaly; Combine the coefficient for the degree of fluctuation and the coefficient for the amplitude of fluctuation to evaluate the abnormal fluctuation of the abnormal curve; The higher the abnormal time difference τj, the longer the duration of the abnormal fluctuation and the more serious the anomaly;

[0074] S2-13: When the index i exceeds the preset anomaly threshold, extract the abnormal curve segment of the index i in the current time period and substitute it into the anomaly analysis model to output the anomaly evaluation coefficient YCi; When the index i does not exceed the preset anomaly threshold, output a no-anomaly signal and label the anomaly evaluation coefficient YCi = 0;

[0075] S2-2: When all n indexes of the home environment information have been analyzed and processed, integrate the anomaly evaluation coefficients corresponding to the n indexes and label them as the information processing result, and send them to the home management model design unit;

[0076] For example, input home environment information, where the home environment information includes the sound decibel value and the sound frequency value. Substitute the data of the 2 indexes of the home environment information into the information processing model respectively, output the anomaly evaluation coefficients corresponding to the 2 indexes, and integrate them into the information processing result and send it to the home management model design unit; The influence situations of the n indexes of the home environment information are different, and the corresponding preset parameters, including the anomaly threshold and the weight factor coefficients, also need to change accordingly. Preset the values of the parameters through a large amount of data measurement;

[0077] S3: The home management model design unit constructs a home management model. The specific process is as follows:

[0078] S3-1: Generate a real-time response plan based on the received information processing result:

[0079] S3-11: First, establish a management analysis model:

[0080] The family management model design unit receives the information processing result, which includes the abnormal evaluation coefficients of n indicators. Take any information processing result YCi as the result parameter and input it into the management analysis model;

[0081] Analyze the abnormal situation of the current family environment information, establish a formula to comprehensively evaluate the n indicators, and obtain the abnormal management coefficient GL: , where μ is the weight factor coefficient corresponding to indicator i and μ > 0. Since the influence situations of the n indicators are different, the preset weight factor coefficients are different;

[0082] There are W kinds of preset family environment information abnormal situations. Mark any one abnormal situation as w, and preset the abnormal management coefficient corresponding to the occurrence of abnormal situation w as GLw. Then, through the ratio of the actually measured abnormal management coefficient GL to the preset abnormal management coefficient GLw, calculate the occurrence probability Gw corresponding to abnormal situation w: , for different abnormal situations, the values of the preset weight factor coefficient μ are different, and the values of the preset abnormal management coefficient GLw corresponding to the occurrence of abnormal situation w are also different;

[0083] Thus, calculate the probabilities corresponding to the W kinds of abnormal situations, sort the W probability values in descending order, and set a probability threshold c. When the occurrence probability Gw corresponding to abnormal situation w exceeds the probability threshold c, extract this abnormal situation w and generate a corresponding control instruction. When the occurrence probability Gw corresponding to abnormal situation w does not exceed the probability threshold c, no processing is performed;

[0084] After comparing the probabilities of the W kinds of abnormal situations with the threshold, preset to extract W0 kinds of abnormal situations that exceed the probability threshold, then generate W0 kinds of control instructions, and integrate the W0 kinds of control instructions into a management plan;

[0085] Among them, the family environment information collects voice-related parameters. For example, if there are abnormal voice situations such as a baby crying at home, it may include physical discomfort caused by hunger, inappropriate environmental temperature, pain, etc. Through a large number of data experiments, obtain the preset weight factor coefficients corresponding to the corresponding abnormal situations, and continuously correct and optimize the preset values. Thus, determine the abnormal situation of the baby by collecting voice-related parameters and make an immediate response, and control the corresponding devices based on the Internet of Things, which is convenient for service personnel to take care of the baby and for guardians to monitor and trace the baby;

[0086] S3-12: Then generate a real-time response plan:

[0087] After inputting the abnormal evaluation coefficients of n indicators of the information processing results into the management analysis model, first output the probabilities corresponding to W types of abnormal situations, generate corresponding control instructions and management plans, and then mark the management plan as a real-time response plan;

[0088] Among them, through the real-time response plan for timely management. For example, if it is calculated and analyzed that there is an abnormal risk in the baby's crying, and the probability of the abnormal situation caused by hunger resulting in discomfort exceeds the preset probability threshold, then in order to solve the hunger problem, a control instruction for the milk powder mixer is immediately generated, so as to prepare milk products at an appropriate temperature for the management service personnel to feed immediately;

[0089] S3-2: Then analyze the degree of abnormal risk, generate a warning signal through threshold comparison, and feedback the warning signal to the edge computing model processing unit. The specific process of analyzing the degree of abnormal risk is as follows:

[0090] S3-21: Analyze the degree of abnormal risk of the current home environment information, establish a formula to comprehensively evaluate n indicators, and obtain the abnormal risk coefficient FX: , where ω is a conversion coefficient and ω > 0, then ωi is the conversion coefficient of indicator i. Since the influence situations of the indicators are different, the preset values of the conversion coefficients change accordingly;

[0091] S3-22: Then preset the abnormal risk threshold b of the abnormal risk coefficient FX. When the abnormal risk coefficient FX exceeds the abnormal risk threshold b, a warning signal is generated and feedback to the edge computing model processing unit;

[0092] By evaluating the degree of abnormal risk, it is determined whether it is necessary to conduct risk prediction on the current home environment information. Since the prediction time period is a fixed value, if the degree of abnormal risk at the current time is low, it means that the probability of an abnormal situation occurring within a short prediction time period is low. Therefore, risk prediction is not carried out to reduce data processing work;

[0093] S4: The edge computing model processing unit conducts in-depth analysis: When receiving the warning signal, in-depth analysis is carried out and a risk prediction result is generated, and the risk prediction result is sent to the home management model design unit;

[0094] S4-1: When receiving the warning signal, then establish an in-depth analysis model and output the risk prediction result. The specific process of establishing the in-depth analysis model is as follows:

[0095] S4-11: First, conduct an integrated analysis of the rising slopes Kp of m abnormal curve segments of indicator i:

[0096] Build a dynamic curve graph Skp of the rising slope Kp. First, obtain the average value of m rising slopes Kp and mark it as the wave rising increase rate Kzp. Then, obtain the standard deviation of the curve Skp and mark it as the wave rising coefficient σkp. Next, measure the differences between adjacent points on the curve Skp in sequence, mark the difference between any two adjacent points as the rising slope fluctuation difference Ckp, build a dynamic curve graph Scp of the rising slope fluctuation difference Ckp, extract all peak points of the curve Scp and their corresponding peak point coordinates, average the ordinates of all peak point values to obtain the rising slope fluctuation extreme value εcp;

[0097] Build a formula to obtain the abnormal rising slope coefficient Kyp: When the wave rising increase rate Kzp, the wave rising coefficient σkp, and the rising slope fluctuation extreme value εcp are higher, the abnormal rising risk of the dynamic curve Si will be higher. Since the rising slope fluctuation difference Ckp is the difference between any two adjacent points on the curve Skp, and there are m points of the rising slope Kp on the curve Skp, therefore, there are a total of (m - 1) rising slope fluctuation differences Ckp. Sum the (m - 1) rising slope fluctuation differences Ckp to obtain the overall fluctuation amplitude of the rising slope. When the overall fluctuation amplitude of the rising slope is higher, it indicates that the abnormal rising risk of the dynamic curve Si is higher; when the abnormal rising slope coefficient Kyp is higher, it means that the abnormal rising risk of the dynamic curve Si is higher;

[0098] S4 - 12: Similarly, according to the above operations, integrate and analyze the falling slope Kq of m abnormal curve segments to obtain the abnormal falling slope coefficient Kyq. When the abnormal falling slope coefficient Kyq is higher, it means that the abnormal falling risk of the dynamic curve Si is higher;

[0099] S4 - 13: Combine the abnormal rising slope coefficient Kyp and the abnormal falling slope coefficient Kyq to analyze the dynamic development trend of the curve Si, and build a formula to obtain the risk prediction coefficient FYi of the index i: Among them, α1 and α2 are the weight factor coefficients of the abnormal rising slope coefficient Kyp and the abnormal falling slope coefficient Kyq respectively, K0 is the slope of the end point of the curve Si, Yi0 is the ordinate value of the end point of the curve Si, and τ0 is the prediction time period; as the slope K0 of the end point increases, the abnormal probability of the curve Si will increase; as the prediction time period increases, the number of times of abnormal fluctuation of the curve Si will increase, which will lead to an increase in the abnormal probability; when the abnormal rising slope coefficient Kyp and the abnormal falling slope coefficient Kyq are higher, it means that the risk of the abnormal trend is higher;

[0100] S4 - 2: Substitute the abnormal curve segments of n indicators of the family environment information into the in - depth analysis model, output the risk prediction coefficients of the n indicators, integrate and mark them as the risk prediction results, and send them to the family management model design unit;

[0101] Among them, by analyzing the abnormal part of the index parameters, the fluctuation trend of the obtained curve is deeply analyzed, so as to predict the development risk of the index parameters. For example, by performing volatility analysis on the curve of the baby's voice frequency value in historical data, due to the different physical conditions and voice conditions of the babies, the preset abnormal thresholds are different. If the current baby's voice frequency value shows frequent abnormal fluctuations in the historical situation, then the baby may have the characteristic that its physical condition is vulnerable to environmental changes, and the risk of abnormal occurrence in the future time period will also be relatively high;

[0102] S5: The home management model design unit generates a home management plan:

[0103] S5-1: First, generate a risk prediction plan by receiving the risk prediction result. The specific process is as follows:

[0104] After inputting the abnormal evaluation coefficients of the n indicators of the risk prediction result into the management analysis model, first output the probabilities corresponding to W abnormal situations, generate the corresponding control instructions and management plans, and then mark this management plan as the risk prediction plan;

[0105] Among them, through the risk prediction plan, management is carried out in advance. For example, it is calculated and analyzed that there is an abnormal risk trend in the baby's cry, and it is analyzed that the occurrence probability of the abnormal situation caused by the cold environment exceeds the preset probability threshold. Then, in order to avoid the cold risk of the baby, a control instruction for the air conditioner is generated in advance to control the air conditioner temperature to rise. By predicting risks in advance, it is convenient to contain unknown abnormal situations;

[0106] S5-2: Combine the real-time response plan with the risk prediction plan to generate a home management plan and send it to the Internet of Things control unit;

[0107] Through the home management plan, the abnormal situations in the home environment are comprehensively supervised. The home management plan includes the control signals for home appliances. The control signals are divided into immediate control instructions and reservation control instructions, so as to immediately process and pre-control the abnormal situations respectively. Among them, immediate processing can reduce the deterioration of abnormal situations, and pre-control can avoid the occurrence of abnormal risks. And through the application of computer devices and systems for feedback loop supervision work, the analysis and processing are made intelligent and self-growing, so as to continuously improve the accuracy of analysis and processing and risk monitoring;

[0108] S6: The Internet of Things control unit performs management and control operations and feedback loop:

[0109] By receiving the control signals in the home management plan, timely control and pre-control operations are performed on the corresponding devices. After the operations are completed, a prompt signal is generated and fed back to the home information monitoring unit;

[0110] Among them, the Internet of Things control unit receives control signals to perform corresponding control and management on home appliances. For example, when abnormal sounds of a crying baby caused by environmental temperature are detected, the air conditioner is controlled; when a crying baby due to hunger is detected, the milk powder dispenser is controlled; when a crying baby due to illness is detected, the infrared thermometer is controlled to collect and obtain the baby's body temperature information and monitor and verify the body temperature situation; as well as other abnormal reasons analyzed and the corresponding home control devices;

[0111] In addition, the home information monitoring unit receives prompt signals, performs information monitoring and processing again, then obtains the actual processing results through the edge computing model processing unit, compares and analyzes the risk prediction results and the actual processing results, makes certain adjustments to the preset parameters in the home management model, and thus realizes the automatic optimization and upgrade of the home management model.

[0112] To sum up, through the monitoring of home environment information, the determination and analysis of abnormal situations of family members are realized, and then the control and management of home appliances are carried out. By analyzing the abnormal parts of home environment information, real-time response plans and risk prediction plans are gradually obtained, and a home management plan is comprehensively generated, which not only realizes the immediate supervision of abnormal situations, but also realizes the risk warning and early control of abnormal situations. Based on the Internet of Things, the corresponding home appliances are controlled, which is convenient for the care work of home service personnel and the emergency handling of abnormal situations, and is also convenient for the guardian to monitor and trace the work of domestic service personnel;

[0113] Among them, through the edge computing model, data processing and storage are pushed to the edge of the computer processor, and edge analysis and storage are performed on the home management plan, reducing the need for data transmission on the network and ensuring the advantages of real-time, reliable, secure, and privacy protection for the application computer to monitor the home situation.

[0114] The settings of the intervals and the magnitudes of the thresholds are for the convenience of comparison. Regarding the magnitudes of the thresholds, they depend on the amount of sample data and the base quantity set by those skilled in the art for each group of sample data; as long as the proportional relationship between the parameters and the quantified values is not affected.

[0115] The above formulas are all dimensionless and take their numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain a formula closest to the real situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation;

[0116] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A computer monitoring system based on an edge computing model, characterized by: It includes a family information monitoring unit, an edge computing model processing unit, a family management model design unit and an Internet of Things control unit, wherein the family information monitoring unit, the edge computing model processing unit, the family management model design unit and the Internet of Things control unit are signal connected; The home information monitoring unit is used to monitor and collect home environment information in real time; The edge computing model processing unit is used to analyze and process the home environment information, and then send the information processing results to the home management model design unit. When an early warning signal is received, it performs in-depth analysis and generates risk prediction results, and sends the risk prediction results to the home management model design unit; The family management model design unit constructs a family management model, generates a real-time response plan by receiving information processing results, analyzes the abnormal risk level, and generates a warning signal by threshold comparison. The warning signal is fed back to the edge computing model processing unit, and then receives the risk prediction results to generate a risk prediction plan. By combining the real-time response plan with the risk prediction plan, a family management plan is generated and sent to the IoT control unit; The Internet of Things control unit is used to receive the family management plan and perform corresponding management and control operations. After the operation is completed, a prompt signal is generated and fed back to the family information monitoring unit; The edge computing model processing unit applies the edge computing model to perform distributed data processing and analysis. The family management model design unit builds a family management model and performs abnormal risk analysis and judgment. The specific process of the interaction between the edge computing model processing unit and the family management model design unit is as follows: A1: The edge computing model processing unit establishes an information processing model, first analyzes and processes the home environment information, substitutes the home environment information into the information processing model, and outputs the information processing results to the home management model design unit; The specific process of establishing an information processing model is as follows: B1: Construct an information processing model to analyze the family environment information, where the family environment information includes n indicators, any indicator data is marked as indicator i, indicator i is input, and the information processing result corresponding to indicator i is output; B1-1: Set the information collection cycle Tc, take the value of index i as the Y-axis, take the information collection cycle Tc as the X-axis, and establish a dynamic curve graph Si of index i-information collection cycle Tc; B1-2: The abnormal threshold a of the preset index i is set. When the ordinate of the dynamic curve Si exceeds the abnormal threshold a, the corresponding curve segment is extracted and marked as an abnormal curve segment. There are m abnormal curve segments preset, and any abnormal curve segment is marked as Sj; B1-3: Establish an abnormal analysis model for the curve segment, perform abnormal analysis on the abnormal curve segment Sj, and obtain the abnormal evaluation coefficient YC. The specific process is as follows: B1-31: Obtain the abnormal time difference τj between the two endpoints of the abnormal curve segment Sj, and then calculate the slopes of the two endpoints of the abnormal curve segment Sj. When the slope is greater than 0, it is marked as an increasing slope Kp, and when the slope is less than 0, it is marked as a decreasing slope Kq; B1-32: Calculate and extract the peak points and valley points of the abnormal curve segment Sj: accumulate the number of all peak points and mark them as the number of peak points Df, accumulate the number of all valley points and mark them as the number of valley points Dg, calculate the average value of the ordinates of all peak points and mark them as the peak average value Jf, calculate the average value of the ordinates of all valley points and mark them as the valley average value Jg; B1-33: Establish a formula to obtain the abnormality assessment coefficient YC through the rising slope Kp, falling slope Kq, peak point number Df, peak mean value Jf, valley point number Dg and valley mean value Jg; B1-4: When the indicator i exceeds the preset abnormal threshold, the abnormal curve segment of the indicator i in the current time period is extracted and substituted into the abnormal analysis model, and the abnormal evaluation coefficient YCi is output; when the indicator i does not exceed the preset abnormal threshold, a no abnormal signal is output and the abnormal evaluation coefficient YCi=0 is marked; B2: When all n indicators of family environment information are analyzed and processed, the abnormal evaluation coefficients corresponding to the n indicators are integrated and marked as information processing results, and sent to the family management model design unit; A2: The family management model design unit establishes a family management model, receives information processing results and substitutes them into the family management model to generate a real-time response plan; A3: The family management model design unit analyzes the abnormal risk level, sets the risk threshold, and generates a warning signal when the risk threshold is exceeded through threshold comparison. The warning signal is fed back to the edge computing model processing unit. A4: The edge computing model processing unit establishes a deep analysis model, deeply analyzes the information processing results, and outputs the risk prediction results to the family management model design unit; A5: The family management model design unit then receives the risk prediction results and substitutes them into the family management model to generate a risk prediction plan. By combining the real-time response plan with the risk prediction plan, a family management plan is generated.

2. The computer monitoring system based on the edge computing model according to claim 1 is characterized in that: The specific process of establishing a deep analysis model is as follows: C1: When receiving the warning signal, a deep analysis model is established to output the risk prediction results; C1-1: First, integrate and analyze the rising slope Kp of the m abnormal curve segments of index i: Establish a dynamic curve graph Skp of the rising slope Kp, first obtain the average value of m rising slopes Kp and mark it as the wave rise rate Kzp, then obtain the standard deviation of the curve Skp and mark it as the wave rise coefficient σkp, then measure the difference between two adjacent points on the curve Skp in turn, mark the difference between any two adjacent points as the rising slope fluctuation difference Ckp, establish a dynamic curve graph Scp of the rising slope fluctuation difference Ckp, extract all the peak points of the curve Scp and the corresponding peak point coordinates, average the ordinates of all the peak point values, and obtain the rising slope fluctuation extreme value εcp; C1-2: Establish a formula to obtain the abnormal rise slope coefficient Kyp through the wave rise rate Kzp, wave rise coefficient σkp, rise slope fluctuation difference Ckp and rise slope fluctuation extreme value εcp; C1-3: Following the above operation, integrate and analyze the descending slopes Kq of the m abnormal curve segments to obtain the abnormal descending slope coefficient Kyq; C1-4: Obtain the coordinates, slope and predicted time period value of the end point of curve Si, and analyze the dynamic development trend of curve Si by combining the abnormal rising slope coefficient Kyp and the abnormal falling slope coefficient Kyq, and establish a formula to obtain the risk prediction coefficient FYi of indicator i C2: After integrating and analyzing the abnormal curve segments of n indicators of family environment information, the risk prediction coefficients of n indicators are output, and the integrated labels are marked as risk prediction results and sent to the family management model design unit.

3. The computer monitoring system based on the edge computing model according to claim 2 is characterized in that: The specific process of establishing a family management model is: D1: The specific process of establishing a management analysis model is: D1-1: The family management model design unit receives the information processing results, which include the abnormal evaluation coefficients of n indicators. Any information processing result YCi is used as a result parameter and input into the management analysis model to analyze the abnormal situation of the current family environment information. A formula is established through the information processing results to comprehensively evaluate the n indicators and obtain the abnormal management coefficient GL. D1-2: There are W kinds of abnormal situations of home environment information, and any abnormal situation is marked as w. The abnormal management coefficient corresponding to the abnormal situation w is preset as GLw. Then, the probability of occurrence Gw corresponding to the abnormal situation w is calculated by the ratio of the actually calculated abnormal management coefficient GL to the preset abnormal management coefficient GLw; D1-3: Sort the W probability values ​​in descending order and set a probability threshold c. When the probability Gw corresponding to the abnormal situation w exceeds the probability threshold c, the abnormal situation w is extracted and a corresponding control instruction is generated. When the probability Gw corresponding to the abnormal situation w does not exceed the probability threshold c, no processing is performed. D1-4: After the probabilities of W abnormal situations are compared with the thresholds, W0 abnormal situations exceeding the probability thresholds are extracted, and W0 control instructions are generated, and the W0 control instructions are integrated into a management plan; D2: The specific process of generating a real-time response plan is as follows: D2-1: After the abnormal evaluation coefficients of the n indicators of the information processing results are input into the management analysis model, the probabilities corresponding to W abnormal situations are output first, and the corresponding control instructions and management plans are generated, and then the management plan is marked as a real-time response plan; Then analyze the abnormal risk level and generate a risk prediction plan.

4. The computer monitoring system based on the edge computing model according to claim 3 is characterized in that: The specific process of analyzing the abnormal risk level and generating a risk prediction plan is as follows: D3: The specific process of analyzing the abnormal risk level is: D3-1: Analyze the abnormal risk level of the current family environment information, establish a formula based on the information processing results to comprehensively evaluate n indicators, and obtain the abnormal risk coefficient FX; D3-2: Preset an abnormal risk threshold b of the abnormal risk coefficient FX. When the abnormal risk coefficient FX exceeds the abnormal risk threshold b, generate a warning signal and feed it back to the edge computing model processing unit; D4: The specific process of generating risk prediction scheme is as follows: D4-1: After the abnormal evaluation coefficients of the n indicators of the risk prediction results are input into the management analysis model, the probabilities corresponding to W abnormal situations are output first, and the corresponding control instructions and management plans are generated, and then the management plan is marked as a risk prediction plan; D5: Generate a family management plan by combining the real-time response plan with the risk prediction plan.

Images