Electric bed quality evaluation system based on big data

Abstract

The invention discloses an electric bed quality evaluation system based on big data. The electric bed quality evaluation system comprises a data acquisition layer, an edge cloud layer, an AI analysis model and a visualization module, the data acquisition layer is used for acquiring original data of the electric bed; the edge cloud layer is used for receiving original data and storing and preprocessing the original data; the AI analysis model is used for analyzing the processing data after the edge cloud layer preprocessing; the visualization module integrates a broken line diagram, a box diagram, a quality evaluation block diagram and life prediction data on a digital twinborn model; the state of the electric bed can be monitored in real time, faults can be effectively predicted, the service life of equipment is prolonged, and the operation convenience of a user and a manager is improved through data visualization; the method has high application value, and particularly has important significance in the aspects of improving the reliability of the electric bed, reducing the maintenance cost and improving the user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of quality assessment, and in particular to an electric bed quality assessment system based on big data. Background Art

[0002] The bed frame can electrically control the height of the head or foot of the bed board and freely adjust the bending degree of the bed board. It is divided into two categories: slat frame and bed board type. The slat frame is called European electric bed, while the bed board type is called American electric bed. It may be equipped with vibration or music function according to needs. Nowadays, it can also be connected to mobile devices for remote control. It is generally common in home or medical use.

[0003] Big data technology has made breakthrough progress in many industries in recent years. Especially in the field of Internet of Things (IoT) devices and smart homes, a large amount of sensor data, usage behavior data, and user feedback information have become new resources for evaluating product performance. By collecting and analyzing various real-time data from electric beds, the quality and performance of electric beds can be evaluated more accurately. These data are usually large-scale, complex, and dynamic, and traditional data analysis methods are difficult to process efficiently.

[0004] As electric beds become more intelligent, more and more electric beds are equipped with a variety of sensors that can monitor bed status, usage, environmental conditions and other data in real time. Big data technology can help effectively process these massive, heterogeneous sensor data and extract valuable information for electric bed quality assessment. Through data mining and machine learning technology, we can discover the potential correlation between user usage patterns and bed quality, predict bed maintenance needs, and even optimize products based on user personalized needs.

[0005] By introducing big data technology, the electric bed quality assessment system will be able to achieve more accurate, real-time and personalized quality management. In the future, it will not only improve the manufacturing quality monitoring of electric beds, but also provide strong data support for later after-sales service, product maintenance and user experience optimization. In addition, through data accumulation and continuous optimization of intelligent algorithms, the quality assessment system can self-learn and evolve, gradually improve the accuracy and efficiency of its assessment, and provide consumers with better products and services.

[0006] Although the quality assessment method based on big data has great potential, it also faces a series of challenges. First, it is impossible to quickly identify anomalies in the large amount of collected data, resulting in slow data processing and difficulty in efficient processing. In addition, it is impossible to predict the life cycle of the electric bed after abnormal analysis and processing, and it is not convenient to quickly perform quality assessment calculations. The analyzed data is not easy to visualize and display.

[0007] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0008] In view of the problems in the related technology, the present invention proposes an electric bed quality assessment system based on big data to overcome the above-mentioned technical problems existing in the existing related technology.

[0009] To this end, the specific technical solution adopted by the present invention is as follows:

[0010] An electric bed quality assessment system based on big data, including a data acquisition layer, an edge cloud layer, an AI analysis model and a visualization module;

[0011] The data acquisition layer is used to collect raw data of the electric bed through sensors, including noise data, vibration data, circuit data and movement accuracy data;

[0012] The edge cloud layer is used to receive the raw data collected by the data collection layer, and store and pre-process the raw data;

[0013] The AI ​​analysis model is used to analyze the processed data after preprocessing by the edge cloud layer. The AI ​​analysis model first generates a visual line graph for the processed data, and obtains the abnormal information of the electric bed according to the line graph, wherein the abnormal information includes the abnormal duration and the abnormal value. The abnormal duration is the interval between the end time of the abnormal information and the start time of the first use of the electric bed, and a relationship is established based on the abnormal duration and the abnormal value. Then, a box plot is established based on all the abnormal durations and abnormal values ​​to observe the trend of the abnormal duration and the abnormal value. Then, the operation status of the electric bed is predicted through the long short-term memory network, that is, the number of faults detected in the box plot is The data is input into the long short-term memory network to realize the prediction of the service life of the electric bed and obtain the life prediction data; and the Pearson correlation coefficient is used to realize the correlation calculation between the original data to judge the correlation between the original data, and the correlation judgment of the fault of the electric bed is realized through the correlation between the original data, and the support vector machine is iterated for many times, and the first iteration is divided into good quality and poor quality. In the second iteration, the data of good quality and poor quality are respectively classified into two categories, and the data of good quality is divided into quality level one and quality level two, and the data of poor quality is divided into quality level three and quality level four, and then the statistical quality evaluation is used to generate a quality evaluation block diagram;

[0014] The visualization module is used to receive the line graph, box plot and quality assessment block diagram generated by the AI ​​analysis model, and to establish a digital twin model of the electric bed through the production data of the electric bed, and then integrate the line graph, box plot, quality assessment block diagram and life prediction data into the digital twin model.

[0015] As a preferred embodiment, the data acquisition layer includes sensors for collecting data, and the sensors include noise sensors, vibration sensors, infrared sensors, and voltage and current detection circuits;

[0016] The noise sensor is used to collect the movement noise of the electric bed and obtain noise data. The vibration sensor is used to collect the movement vibration of the electric bed and obtain vibration data. The infrared sensor is used to collect the movement position of the electric bed and obtain movement accuracy data. The voltage and current detection circuit is used to collect the voltage and current of the electric bed and obtain circuit data.

[0017] As a preferred embodiment, the noise data is used to analyze and process whether high noise is generated during the control of the electric bed, the vibration data is used to analyze and process whether large-scale shaking is generated during the control of the electric bed, the movement accuracy data is used to analyze and process the position deviation during the control of the electric bed, and the circuit data is used to analyze and process the motor performance of the electric bed.

[0018] As a preferred embodiment, the Internet of Things technology is used for data transmission between the data acquisition layer and the edge cloud layer, the edge cloud layer and the AI ​​analysis model are used for data transmission through the 5G network, the AI ​​analysis model and the visualization module constitute a platform layer, and the platform layer also performs 5G network data transmission with the user's mobile device. The platform layer analyzes and processes the collected data, and transmits the line chart, box plot, quality assessment block diagram, life prediction data and digital twin model displayed by the visualization module to the user's mobile device.

[0019] As a preferred embodiment, the edge cloud layer stores the original data transmitted by the data acquisition layer, the edge cloud layer adopts distributed storage, and the edge cloud layer also pre-processes the original data to obtain processed data. The pre-processing of the edge cloud layer includes data cleaning, deduplication, filling missing values ​​and unifying the format of the original data.

[0020] As a preferred embodiment, the abnormal information of the electric bed obtained by the line graph is calculated as follows:

[0021] First, calculate the Z-Score value. For each time point, calculate the standardized deviation of the treatment data from the mean and use the Z-Score formula: ,

[0022] in, is the processing data at the current time point, is the average value of the processed data, is the standard deviation of the processed data, represents the standardized deviation of the treated data from the mean;

[0023] And when Z exceeds the set threshold, it is marked as an outlier, that is ;

[0024] At this time, Merge into the line chart, and generate two horizontal lines in the line chart. The two horizontal lines are , and Y is represented as the Y axis in the line graph, and represents the collected processing data, the X axis in the line graph represents time, and then the line graph exceeds The values ​​are marked and abnormal information is displayed in real time, and the abnormal information is expressed as abnormal duration and abnormal value.

[0025] As a preferred embodiment, the abnormal duration and the abnormal value are the abnormal values ​​exceeding The data is extracted, and the abnormal duration is the time interval between the end time of the abnormal value and the start time of the electric bed;

[0026] The date and time format is yyyy-mm-dd HH:MM:SS;

[0027] The abnormal duration is converted:

[0028] Convert between different units:

[0029] Duration hours: Duration = (end time − start time) × 24;

[0030] Duration in minutes: Duration = (end time − start time) × 1440;

[0031] Duration in seconds: Duration = (end time − start time) × 86400;

[0032] All the abnormal durations and the abnormal values ​​are associated to form a data vector.

[0033] As a preferred embodiment, the calculation and analysis steps of the box plot are as follows:

[0034] Sorting data: arranging the abnormal duration and the abnormal value according to the abnormal duration;

[0035] Calculate the quartiles: the first quartile Q1 is the lower quarter of the data vector, that is, the 25th percentile; the median Q2 is the median of the data vector, that is, the 50th percentile; the third quartile Q3 is the upper quarter of the data vector, that is, the 75th percentile;

[0036] Calculate the interquartile range: The interquartile range is the gap between Q3 and Q1, which is used to measure the dispersion of the data: ,in, represents the interquartile range;

[0037] Determine the range of whiskers: The whiskers of a box plot represent the valid range of the data and are usually defined as:

[0038] Upper mustache: ;

[0039] Lower part: ;

[0040] Data judgment: Data exceeding the upper whisker and below the lower whisker are judged as fault data;

[0041] Steps to calculate quartiles:

[0042] Median Q2: Arrange the data vectors in time series and find the position of the median. If the number of data vectors is odd, take the middle number; if the number of data vectors is even, take the average of the two middle numbers.

[0043] First quartile Q1: Find the median of the data vector to the left of the median Q2; if the number of data vectors is odd, the first quartile Q1 is the middle number; if it is even, the first quartile Q1 is the average of the two middle numbers;

[0044] Third quartile Q3: Find the median of the data vector to the right of the median Q2. The calculation method is the same as the first quartile Q1.

[0045] As a preferred embodiment, the calculation formula of the long short-term memory network is as follows:

[0046] Forget Gate: ,

[0047] in, Represented as the output of the forget gate, the value is between 0 and 1, 1 means that the information of the previous moment is completely retained, and 0 means that it is completely forgotten; Represented as the weight matrix of the forget gate; Represented as the bias term of the forget gate; Represented as Sigmoid activation function; Represented as the hidden state of the previous moment, that is, the output of the previous moment; It is represented as the input at the current moment, which is the fault data detected in the box plot;

[0048] Input Gate: ,

[0049] in, Represented as the output of the input gate, the value is between 0 and 1, indicating the importance of the input at the current moment; Represented as the weight matrix of the input gate; Represented as the bias term of the input gate; Represented as candidate cell states; Represented as a weight matrix, responsible for connecting the hidden state of the previous moment and the current input Perform linear transformation; Represented as a bias term, it is used to adjust the calculation process of candidate cell states; It is represented as a hyperbolic tangent activation function, which maps the linear combination of candidate cell states to the range of [-1,1];

[0050] Update cell status: ,

[0051] in, Represents the cell state at the current moment; Represents the cell state at the last moment; Represented as the output of the forget gate; The output of the input gate; Represented as candidate cell states;

[0052] Output Gate: ,

[0053] The hidden state is calculated as: ,

[0054] in, Represented as the output of the output gate; Represented as the hidden state at the current moment, that is, the output of LSTM, which will be used as the input at the next moment;

[0055] The calculation formula of the Pearson correlation coefficient is as follows: ,

[0056] in, and are the data points of the i-th observation value of two different fault data in the original data; and is the mean of X and Y; ∑ represents the sum of all fault data;

[0057] Among them, r=1 indicates a completely positive linear correlation; that is, when one fault data increases, the other fault data will also increase at a fixed ratio;

[0058] r=-1 indicates a completely negative linear correlation; that is, when one fault data is present, the other fault data will decrease at a fixed ratio;

[0059] r=0 means there is no linear correlation; that is, there is no linear relationship between the two fault data;

[0060] r>0 indicates positive correlation, that is, when one fault data increases, the other fault data tends to increase;

[0061] r<0 indicates negative correlation, that is, when one fault data increases, the other fault data tends to decrease;

[0062] Correlation strength: 0.1<|r|<0.3 indicates weak correlation; 0.3<|r|<0.5 indicates moderate correlation; |r|>0.5 indicates strong correlation.

[0063] As a preferred embodiment, the calculation of the binary classification processing of the support vector machine is as follows: ,

[0064] in, is the normal vector, i.e. the normal to the hyperplane; is the bias term; is the input sample, i.e., the abnormal duration and abnormal value;

[0065] For positive samples , the constraints are: ;

[0066] For negative samples , the constraints are: ;

[0067] It is used to find a hyperplane in a given abnormal duration and abnormal value. The hyperplane can separate the samples of abnormal duration and abnormal value as much as possible.

[0068] The beneficial effects of the present invention are:

[0069] 1. When in use, the present invention collects raw data from the electric bed through a large number of sensors, mainly noise data, vibration data, circuit data and movement accuracy data of the electric bed, which can detect the noise, vibration, movement accuracy and servo motor or servo cylinder during the movement of the electric bed, and detect the movement state of the electric bed. The collected raw data is transmitted through the Internet of Things, and the raw data is transmitted to the edge cloud layer, so as to realize rapid data processing, reduce the computing pressure of the platform layer and improve the computing efficiency;

[0070] 2. When the present invention is used, the data is transmitted to the edge cloud layer, and the edge cloud layer stores and pre-processes the transmitted raw data. The edge cloud layer stores the raw data through distributed storage to ensure efficient processing of data streams, and reduces latency through real-time data stream processing. The edge cloud layer pre-processes the raw data to clean, remove duplicates, fill in missing values ​​and unify the format of the raw data, and perform preliminary processing on the raw data, preventing the platform layer from processing the raw data, causing the calculation and analysis to take up a lot of time;

[0071] 3. When the present invention is in use, the AI ​​analysis model analyzes the processed data, and performs outlier detection on the line graph of the processed data by integrating the Z-Score value on the line graph, so as to obtain the abnormal duration and abnormal value in the processed data, and re-establish a box graph for the abnormal duration and abnormal value, so as to observe the trend of the abnormal duration and abnormal value; and input the fault data detected in the box graph into the long short-term memory network, so as to predict the service life of the electric bed; and perform correlation analysis on the noise data, vibration data, circuit data and movement accuracy data, that is, the correlation calculation between different data is realized through the Pearson correlation coefficient, so as to judge the correlation between different data, and then realize the correlation judgment of the fault of the electric bed; finally, the abnormal duration and abnormal value are subjected to binary classification processing by the support vector machine, and multiple iterations are adopted to facilitate the quality evaluation of the electric bed;

[0072] 4. When the present invention is used, a digital twin model of the electric bed is established through the production data of the electric bed, and then a line graph, a box plot, a quality assessment block diagram and life prediction data are integrated into the digital twin model, so that the information of the electric bed can be visualized, which is convenient for background analysis and processing and for users to watch;

[0073] In summary, the present invention uses a large number of sensors to monitor the noise, vibration, circuit data and movement accuracy of the electric bed in real time, and transmits the original data to the edge cloud layer through the Internet of Things for rapid processing, thereby reducing the computing pressure of the platform layer; the edge cloud layer is responsible for storing and preprocessing the original data, including data cleaning, deduplication, filling missing values ​​and unifying the format, to ensure efficient processing and reduce delays; an AI analysis model is used to detect anomalies in the processed data, and the service life of the electric bed is predicted by combining line graphs, box plots and long short-term memory networks; at the same time, the Pearson correlation coefficient is used to perform multidimensional data association analysis to help identify electric bed faults, and a support vector machine is used to perform binary classification of fault data to ultimately achieve quality assessment; a digital twin model is established through the production data of the electric bed, and information such as line graphs, box plots, quality assessment and life prediction are integrated into the digital twin model to achieve information visualization, which is convenient for background analysis and user viewing;

[0074] It can not only monitor the status of electric beds in real time, but also effectively predict faults, extend the life of equipment, and improve the operational convenience of users and managers through data visualization. It has high application value, especially in improving the reliability of electric beds, reducing maintenance costs and improving user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0076] Figure 1 It is a structural schematic diagram of an electric bed quality assessment system based on big data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0077] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in the field should be able to understand other possible implementations and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0078] According to an embodiment of the present invention, a motor-assisted bed quality assessment system based on big data is provided.

[0079] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1As shown, a quality assessment system for an electric bed based on big data according to an embodiment of the present invention includes a data acquisition layer, an edge cloud layer, an AI analysis model and a visualization module;

[0080] The data acquisition layer is used to collect the original data of the electric bed through sensors, including noise data, vibration data, circuit data and movement accuracy data;

[0081] The edge cloud layer is used to receive the raw data collected by the data collection layer, and store and pre-process the raw data;

[0082] The AI ​​analysis model is used to analyze the processed data after edge cloud preprocessing. The AI ​​analysis model first generates a visual line graph for the processed data, and obtains the abnormal information of the electric bed based on the line graph, wherein the abnormal information includes the abnormal duration and the abnormal value. The abnormal duration is the interval between the end time of the abnormal information and the start time of the first use of the electric bed, and a relationship is established based on the abnormal duration and the abnormal value. Then, a box plot is established based on all the abnormal durations and abnormal values ​​to observe the trend of the abnormal duration and the abnormal value. The operating status of the electric bed is predicted through the long short-term memory network, that is, the fault data detected in the box plot is input into the long short-term memory network to predict the service life of the electric bed. , obtain life prediction data; and also use the Pearson correlation coefficient to calculate the correlation between the original data, judge the correlation between the original data, and use the correlation between the original data to make a correlation judgment on the fault of the electric bed. Finally, the AI ​​analysis model uses the support vector machine to perform binary classification processing on the abnormal duration and abnormal value to determine the quality of the electric bed, and perform multiple iterations on the support vector machine, dividing the first iteration into good quality and poor quality, and the second iteration, respectively, classifying the data of good quality and poor quality, dividing the data of good quality into quality level 1 and quality level 2, and dividing the data of poor quality into quality level 3 and quality level 4, and then the statistical quality evaluation is used to generate a quality evaluation block diagram;

[0083] The visualization module is used to receive the line graph, box graph and quality assessment block diagram generated by the AI ​​analysis model, and to establish a digital twin model of the electric bed through the production data of the electric bed, and then integrate the line graph, box graph, quality assessment block diagram and life prediction data into the digital twin model.

[0084] In this embodiment, preferably, the data acquisition layer includes sensors for collecting data, and the sensors include noise sensors, vibration sensors, infrared sensors, and voltage and current detection circuits;

[0085] The noise sensor is used to collect the movement noise of the electric bed and obtain noise data. The vibration sensor is used to collect the movement vibration of the electric bed and obtain vibration data. The infrared sensor is used to collect the movement position of the electric bed and obtain movement accuracy data. The voltage and current detection circuit is used to collect the voltage and current of the electric bed and obtain circuit data.

[0086] It should be noted that the noise sensor is used to monitor the noise level of the bed, reflecting the smoothness of the bed's movement and the operating status of the motor; the vibration sensor is used to detect the vibration of the electric bed, and then evaluate the stability and movement accuracy of the bed; the infrared sensor is used to obtain the movement accuracy data of the electric bed, and can monitor the movement trajectory and positioning accuracy of the electric bed in real time; the voltage and current detection circuit is used to monitor the circuit status of the electric bed, ensure that the power supply of the electric bed is normal, and can reflect the working status of the motor and circuit system; through the data collection of these sensors, the movement status, stability and circuit health of the electric bed can be fully monitored, providing comprehensive basic data support for subsequent data processing and fault diagnosis.

[0087] In this embodiment, preferably, the noise data is used to analyze whether high noise is generated during the control process of the electric bed, the vibration data is used to analyze whether large-amplitude jitter is generated during the control process of the electric bed, the movement accuracy data is used to analyze the position deviation during the control process of the electric bed, and the circuit data is used to analyze the motor performance of the electric bed;

[0088] It should be noted that the noise data is used to analyze whether the electric bed generates high noise during movement. If the noise exceeds the set threshold, it may mean that the control system or drive system of the electric bed has an abnormality, such as motor failure, excessive mechanical friction, etc. This analysis helps to promptly discover and solve the noise problem of the electric bed during the control process.

[0089] Vibration data is used to analyze whether the electric bed has a large vibration during movement; large vibrations may be caused by loose mechanical parts, unstable structure or abnormal control system; this analysis helps to determine the structural stability of the electric bed and its motion control accuracy, ensuring its stability and safety during use;

[0090] The movement accuracy data is used to analyze whether there is position deviation in the control process of the electric bed; position deviation may cause the electric bed to fail to accurately reach the predetermined position, affecting the user experience and accuracy requirements; by analyzing and processing the position accuracy, it can be determined whether there is an error in the positioning system of the electric bed to ensure the accuracy of its motion trajectory;

[0091] Circuit data is used to analyze the motor performance of the electric bed. Fluctuations or abnormalities in current and voltage may reflect excessive motor load, power supply problems, or failure of the motor itself. By analyzing circuit data, potential problems in the operation of the electric bed motor, such as overload and overheating, can be identified in a timely manner, thereby avoiding failures and improving the safety and service life of the electric bed.

[0092] In this embodiment, preferably, the Internet of Things technology is used for data transmission between the data collection layer and the edge cloud layer, and the edge cloud layer and the AI ​​analysis model are used for data transmission through the 5G network. The AI ​​analysis model and the visualization module constitute the platform layer, and the platform layer also performs 5G network data transmission with the user's mobile device. The platform layer analyzes and processes the collected data, and transmits the line chart, box plot, quality assessment block diagram, life prediction data and digital twin model displayed by the visualization module to the user's mobile device;

[0093] It should be noted that the Internet of Things ensures smooth transmission of data between the data collection layer and the edge cloud layer, while the 5G network ensures efficient data exchange between the edge cloud layer and the AI ​​analysis model, and between the platform layer and user devices; the AI ​​model conducts in-depth analysis of the collected data and presents it through a visualization module to help users view the operating status, quality assessment, life prediction, and digital twin models of the electric bed in real time, ensuring that users can understand the equipment performance in a timely manner and make corresponding adjustments, and it is convenient for the background to obtain quality information of the electric bed, so as to facilitate subsequent production improvements.

[0094] In this embodiment, preferably, the edge cloud layer stores the original data transmitted by the data collection layer, and the edge cloud layer adopts distributed storage. The edge cloud layer also pre-processes the original data to obtain processed data. The pre-processing of the edge cloud layer includes data cleaning, deduplication, filling missing values ​​and unifying the format of the original data;

[0095] It should be noted that the edge cloud layer uses distributed storage technology to store the original data transmitted from the data collection layer; distributed storage has good scalability and fault tolerance, can ensure the reliable storage of large amounts of data, and support high-concurrency data access;

[0096] The edge cloud layer preprocesses the raw data to provide clearer and more standardized data for subsequent analysis; the preprocessing steps include:

[0097] Data cleaning: remove noise data or irrelevant information to ensure data accuracy;

[0098] Deduplication: If there are duplicate records in the original data, the edge cloud layer will perform deduplication processing to prevent redundant data from affecting the analysis results;

[0099] Filling missing values: For missing values ​​during the collection process, the edge cloud layer uses appropriate methods (such as mean filling, interpolation, etc.) to fill the missing original data to maintain the integrity of the original data;

[0100] Unified format: In order to enable unified processing and analysis of raw data, the edge cloud layer will standardize the format of raw data from different sources to make it conform to the predetermined data format requirements;

[0101] Data cleaning is used to smooth the original data and remove noise. The following calculation formula is used to process the original data: ,

[0102] in, Represented as the original data corresponding to time point t in the original data, Represented as time points in the original data The corresponding smoothing value is is a smoothing factor used to assign weights to new raw data and historical raw data, and Represents the original data at scale and location The transformation structure on is the conjugate of the wavelet transform, is the scale parameter, is the translation parameter;

[0103] The missing values ​​are filled with the mean, and the calculation formula of the mean is as follows: ;

[0104] in, is the calculated mean of the original data, is the weight of each raw data. Usually the latest raw data has a larger weight. is the original data in the time series, is the noise error, is the moving average coefficient of the weight, is the total amount of original data;

[0105] The unified format is to normalize the original data and map the original data between [0,1], and the normalization formula is as follows: ,in, Represented as raw data, Represented as the minimum value in the original data, Represented as the maximum value in the original data, Represented as normalized raw data.

[0106] In this embodiment, preferably, the abnormal information of the electric bed obtained by the line graph is calculated as follows:

[0107] First, calculate the Z-Score value. For each time point, calculate the standardized deviation of the treatment data from the mean and use the Z-Score formula:

[0108] ,

[0109] in, is the processing data at the current time point, is the average value of the processed data, is the standard deviation of the processed data, represents the standardized deviation of the treated data from the mean;

[0110] And when Z exceeds the set threshold, it is marked as an outlier, that is ;

[0111] At this time, Merge into the line chart, and generate two horizontal lines in the line chart. The two horizontal lines are , and Y is represented as the Y axis in the line graph, and represents the collected processing data, the X axis in the line graph represents time, and then the line graph exceeds The values ​​are marked and abnormal information is displayed in real time, and the abnormal information is expressed as abnormal duration and abnormal value;

[0112] It should be noted that in the line chart, when the original data value exceeds the set abnormal threshold, the corresponding abnormal data point will be marked so that the user can quickly identify the abnormal situation; two horizontal lines will be generated in the line chart, namely the upper threshold and the lower threshold; these two lines are used to calibrate the abnormal threshold of the Z-Score; when the Z-Score of the data point exceeds these two horizontal lines, it is marked as abnormal; the line chart that can display abnormal information in real time is helpful to monitor the status changes of the electric bed and promptly discover potential faults or performance problems.

[0113] In this embodiment, preferably, the abnormal duration and abnormal value are the abnormal The data is extracted, and the abnormal duration is the time interval between the end time of the abnormal value and the start time of the electric bed use;

[0114] The date and time format is yyyy-mm-dd HH:MM:SS;

[0115] Convert abnormal duration:

[0116] Convert between different units:

[0117] Duration hours: Duration = (end time − start time) × 24;

[0118] Duration in minutes: Duration = (end time − start time) × 1440;

[0119] Duration in seconds: Duration = (end time − start time) × 86400;

[0120] Associate all abnormal durations and abnormal values ​​to form a data vector;

[0121] It should be noted that the abnormal duration and abnormal value are data that are beyond the normal range extracted from the line chart, and these abnormal information are further calculated and associated. The abnormal duration refers to the time period when the abnormal value appears. The calculation process includes time format processing, unit conversion, and finally associating all abnormal information to form a data vector, which is convenient for subsequent data analysis and processing.

[0122] In this embodiment, preferably, the calculation and analysis steps of the box plot are as follows:

[0123] Sorting data: Arrange the abnormal duration and abnormal value according to the abnormal duration;

[0124] Calculate the quartiles: the first quartile Q1 is the lower quarter of the data vector, that is, the 25th percentile; the median Q2 is the median of the data vector, that is, the 50th percentile; the third quartile Q3 is the upper quarter of the data vector, that is, the 75th percentile;

[0125] Calculate the interquartile range: The interquartile range is the gap between Q3 and Q1, which is used to measure the dispersion of the data: ,in, represents the interquartile range;

[0126] Determine the range of whiskers: The whiskers of a box plot represent the valid range of the data and are usually defined as:

[0127] Upper mustache: ;

[0128] Lower part: ;

[0129] Data judgment: Data exceeding the upper whisker and below the lower whisker are judged as fault data;

[0130] Steps to calculate quartiles:

[0131] Median Q2: Arrange the data vectors in time series and find the position of the median. If the number of data vectors is odd, take the middle number; if the number of data vectors is even, take the average of the two middle numbers.

[0132] First quartile Q1: Find the median of the data vector to the left of the median Q2; if the number of data vectors is odd, the first quartile Q1 is the middle number; if it is even, the first quartile Q1 is the average of the two middle numbers;

[0133] Third quartile Q3: Find the median of the data vector to the right of the median Q2. The calculation method is the same as the first quartile Q1.

[0134] It should be noted that the use of box plot analysis can help monitoring personnel quickly identify abnormal information. The box plot provides a very intuitive visualization method, which is convenient for understanding and interpreting the distribution of data. By identifying data that exceeds the upper and lower whiskers, potential equipment failures or system abnormalities can be effectively detected. It can greatly improve the efficiency of abnormality detection and fault diagnosis, and avoid the impact of potential failures on the system.

[0135] In this embodiment, preferably, the calculation formula of the long short-term memory network is as follows:

[0136] Forget Gate: ,

[0137] in, Represented as the output of the forget gate, the value is between 0 and 1, 1 means that the information of the previous moment is completely retained, and 0 means that it is completely forgotten; Represented as the weight matrix of the forget gate; Represented as the bias term of the forget gate; Represented as Sigmoid activation function; Represented as the hidden state of the previous moment, that is, the output of the previous moment; It is represented as the input at the current moment, which is the fault data detected in the box plot;

[0138] Input Gate: ,

[0139] in, Represented as the output of the input gate, the value is between 0 and 1, indicating the importance of the input at the current moment; Represented as the weight matrix of the input gate; Represented as the bias term of the input gate; Represented as candidate cell states; Represented as a weight matrix, responsible for connecting the hidden state of the previous moment and the current input Perform linear transformation; Represented as a bias term, it is used to adjust the calculation process of candidate cell states; It is represented as a hyperbolic tangent activation function, which maps the linear combination of candidate cell states to the range of [-1,1];

[0140] Update cell status: ,

[0141] in, Represents the cell state at the current moment; Represents the cell state at the last moment; Represented as the output of the forget gate; The output of the input gate; Represented as candidate cell states;

[0142] Output Gate: ,

[0143] The hidden state is calculated as: ,

[0144] in, Represented as the output of the output gate; Represented as the hidden state at the current moment, that is, the output of LSTM, will be used as the input at the next moment;

[0145] The calculation formula of Pearson correlation coefficient is as follows: ,

[0146] in, and are the data points of the i-th observation value of two different fault data in the original data; and is the mean of X and Y; ∑ represents the sum of all fault data;

[0147] Among them, r=1 indicates a completely positive linear correlation; that is, when one fault data increases, the other fault data will also increase at a fixed ratio;

[0148] r=-1 indicates a completely negative linear correlation; that is, when one fault data is present, the other fault data will decrease at a fixed ratio;

[0149] r=0 means there is no linear correlation; that is, there is no linear relationship between the two fault data;

[0150] r>0 indicates positive correlation, that is, when one fault data increases, the other fault data tends to increase;

[0151] r<0 indicates negative correlation, that is, when one fault data increases, the other fault data tends to decrease;

[0152] Correlation strength: 0.1<|r|<0.3 indicates weak correlation; 0.3<|r|<0.5 indicates moderate correlation; |r|>0.5 indicates strong correlation;

[0153] It should be noted that the fault data in the box plot uses time series prediction to predict the fault or performance degradation of the electric bed and provide early warning, that is, the service life of the electric bed is predicted through the long short-term memory network. The long short-term memory network is used to capture the long-term dependencies in the time series, predict the state changes of the electric bed in the future, and update the calculation through the update gate;

[0154] The Pearson correlation coefficient is used to measure the linear relationship between two types of data. The value ranges from -1 to 1. The closer the value is to 1 or -1, the stronger the linear relationship is, and the closer the value is to 0, the almost no linear relationship is. The Pearson correlation coefficient can be used to quickly understand the correlation between data sets.

[0155] In this embodiment, preferably, the calculation of the binary classification processing of the support vector machine is as follows: ,

[0156] in, is the normal vector, i.e. the normal to the hyperplane; is the bias term; is the input sample, i.e., the abnormal duration and abnormal value;

[0157] For positive samples , the constraints are: ;

[0158] For negative samples , the constraints are: ;

[0159] It is used to find a hyperplane in a given abnormal duration and abnormal value. The hyperplane can separate the samples of abnormal duration and abnormal value as much as possible.

[0160] It should be noted that the data is classified through the binary classification processing of the support vector machine. The data is classified through a hyperplane. The fewer the fault data and the smaller the fault data value, the better the quality of the electric bed. That is, the input sample is composed of abnormal duration and abnormal value. A hyperplane can be found from these samples to separate the positive class (normal sample) and the negative class (abnormal sample). The classification result will help us identify and diagnose fault data, thereby improving the monitoring and fault detection capabilities of the system. By optimizing the objective function and constraints, the support vector machine can accurately segment samples of abnormal duration and abnormal value, thereby improving the efficiency and accuracy of anomaly detection.

[0161] The specific operation process of this application:

[0162] When in use, the data acquisition layer collects the original data of the electric bed through sensors, including noise data, vibration data, circuit data and movement accuracy data; the data acquisition layer and the edge cloud layer use Internet of Things technology for data transmission, the edge cloud layer receives the original data collected by the data acquisition layer, and stores and preprocesses the original data; the edge cloud layer and the AI ​​analysis model transmit data through the 5G network, and the AI ​​analysis model analyzes the processed data preprocessed by the edge cloud layer. The AI ​​analysis model first generates a visual line graph for the processed data, and obtains abnormal information of the electric bed based on the line graph. The abnormal information includes the abnormal duration and abnormal value. The abnormal duration is the interval between the end time of the abnormal information and the start time of the first use of the electric bed. The abnormal duration and the abnormal value are related, and then a box plot is established for all the abnormal durations and abnormal values ​​to observe the trend of the abnormal duration and abnormal value. The operating state of the electric bed is predicted through the long short-term memory network, that is, the fault data detected in the box plot is input into the long short-term memory network to predict the service life of the electric bed. The Pearson correlation coefficient is also used to realize the correlation calculation between the original data. The correlation between the original data is judged, and the correlation judgment of the electric bed fault is realized through the correlation between the original data. Finally, the AI ​​analysis model realizes binary classification of abnormal duration and abnormal value through support vector machine to judge the quality of the electric bed, and performs multiple iterations of support vector machine. The first iteration is divided into good quality and poor quality. The second iteration is to perform binary classification on the data of good quality and poor quality respectively, and the data of good quality is divided into quality level 1 and quality level 2, and the data of poor quality is divided into quality level 3 and quality level 4. Then the statistical quality evaluation is used to generate quality The visualization module is used to generate the line graph, box graph and quality assessment block diagram generated by the AI ​​analysis model, and establish the digital twin model of the electric bed through the production data of the electric bed, and then integrate the line graph, box graph, quality assessment block diagram and life prediction data into the digital twin model; the AI ​​analysis model and the visualization module constitute the platform layer, and the platform layer also transmits data with the user's mobile device over the 5G network. The platform layer analyzes and processes the collected data, and transmits the line graph, box graph, quality assessment block diagram, life prediction data and digital twin model displayed by the visualization module to the user's mobile device.

[0163] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An electric bed quality assessment system based on big data, characterized in that: Includes data collection layer, edge cloud layer, AI analysis model and visualization module; The data acquisition layer is used to collect raw data of the electric bed through sensors, including noise data, vibration data, circuit data and movement accuracy data; The edge cloud layer is used to receive the raw data collected by the data collection layer, and store and pre-process the raw data; The AI ​​analysis model is used to analyze the processed data after preprocessing by the edge cloud layer. The AI ​​analysis model first generates a visual line graph for the processed data, and obtains the abnormal information of the electric bed according to the line graph, wherein the abnormal information includes the abnormal duration and the abnormal value. The abnormal duration is the interval between the end time of the abnormal information and the start time of the first use of the electric bed, and a relationship is established according to the abnormal duration and the abnormal value, and then a box plot is established according to all the abnormal durations and abnormal values ​​to realize the observation of the trend of the abnormal duration and the abnormal value; and then the operation status of the electric bed is predicted through the long short-term memory network, that is, the fault data detected in the box plot is input into the long short-term memory network to realize the prediction of the service life of the electric bed. The life prediction data is obtained by measuring the original data through the Pearson correlation coefficient, and the correlation between the original data is judged. The correlation judgment of the fault of the electric bed is realized through the correlation between the original data. Finally, the AI ​​analysis model realizes binary classification of abnormal duration and abnormal value through the support vector machine to realize the judgment of the quality of the electric bed, and the support vector machine is iterated for multiple times, and the first iteration is divided into good quality and poor quality. In the second iteration, the data of good quality and poor quality are respectively classified into two categories, and the data of good quality is divided into quality level 1 and quality level 2, and the data of poor quality is divided into quality level 3 and quality level 4. Then, the statistical quality evaluation is used to generate a quality evaluation block diagram; The visualization module is used to receive the line graph, box plot and quality assessment block diagram generated by the AI ​​analysis model, and to establish a digital twin model of the electric bed through the production data of the electric bed, and then integrate the line graph, box plot, quality assessment block diagram and life prediction data into the digital twin model.

2. The electric bed quality assessment system based on big data according to claim 1 is characterized in that: The data acquisition layer includes sensors for collecting data, and the sensors include noise sensors, vibration sensors, infrared sensors and voltage and current detection circuits; The noise sensor is used to collect the movement noise of the electric bed and obtain noise data. The vibration sensor is used to collect the movement vibration of the electric bed and obtain vibration data. The infrared sensor is used to collect the movement position of the electric bed and obtain movement accuracy data. The voltage and current detection circuit is used to collect the voltage and current of the electric bed and obtain circuit data.

3. The electric bed quality assessment system based on big data according to claim 2 is characterized in that: The noise data is used to analyze and process whether high noise is generated during the control process of the electric bed, the vibration data is used to analyze and process whether large-scale jitter is generated during the control process of the electric bed, the movement accuracy data is used to analyze and process the position deviation during the control process of the electric bed, and the circuit data is used to analyze and process the motor performance of the electric bed.

4. The electric bed quality assessment system based on big data according to claim 1, characterized in that: The data acquisition layer and the edge cloud layer use Internet of Things technology for data transmission, the edge cloud layer and the AI ​​analysis model use 5G network for data transmission, the AI ​​analysis model and the visualization module form a platform layer, and the platform layer also performs 5G network data transmission with the user's mobile device. The platform layer analyzes and processes the collected data, and transmits the line chart, box plot, quality assessment block diagram, life prediction data and digital twin model displayed by the visualization module to the user's mobile device.

5. The electric bed quality assessment system based on big data according to claim 1, characterized in that: The edge cloud layer stores the original data transmitted by the data acquisition layer. The edge cloud layer adopts distributed storage. The edge cloud layer also pre-processes the original data to obtain processed data. The pre-processing of the edge cloud layer includes data cleaning, deduplication, filling missing values ​​and unifying the format of the original data.

6. The electric bed quality assessment system based on big data according to claim 1, characterized in that: The abnormal information of the electric bed obtained by the line graph is calculated as follows: First, calculate the Z-Score value. For each time point, calculate the standardized deviation of the treatment data from the mean and use the Z-Score formula: ， in, is the processing data at the current time point, is the average value of the processed data, is the standard deviation of the processed data, represents the standardized deviation of the treated data from the mean; And when Z exceeds the set threshold, it is marked as an outlier, that is ; At this time, Merge into the line chart, and generate two horizontal lines in the line chart. The two horizontal lines are , and Y is represented as the Y axis in the line graph, and represents the collected processing data, the X axis in the line graph represents time, and then the line graph exceeds The values ​​are marked and abnormal information is displayed in real time, and the abnormal information is expressed as abnormal duration and abnormal value.

7. The electric bed quality assessment system based on big data according to claim 6, characterized in that: The abnormal duration and the abnormal value are the abnormal values ​​in the line graph. The data is extracted, and the abnormal duration is the time interval between the end time of the abnormal value and the start time of the electric bed; The date and time format is yyyy-mm-dd HH:MM:SS; The abnormal duration is converted: Convert between different units: Duration hours: Duration = (end time − start time) × 24; Duration in minutes: Duration = (end time − start time) × 1440; Duration in seconds: Duration = (end time − start time) × 86400; All the abnormal durations and the abnormal values ​​are associated to form a data vector.

8. The electric bed quality assessment system based on big data according to claim 7, characterized in that: The calculation and analysis steps of the box plot are as follows: Sorting data: arranging the abnormal duration and the abnormal value according to the abnormal duration; Calculate the quartiles: the first quartile Q1 is the lower quarter of the data vector, that is, the 25th percentile; the median Q2 is the median of the data vector, that is, the 50th percentile; the third quartile Q3 is the upper quarter of the data vector, that is, the 75th percentile; Calculate the interquartile range: The interquartile range is the gap between Q3 and Q1, which is used to measure the dispersion of the data: ,in, represents the interquartile range; Determine the range of whiskers: The whiskers of a box plot represent the valid range of the data and are usually defined as: Upper mustache: ; Lower part: ; Data judgment: Data exceeding the upper whisker and below the lower whisker are judged as fault data; Steps to calculate quartiles: Median Q2: Arrange the data vectors in time series and find the position of the median. If the number of data vectors is odd, take the middle number; if the number of data vectors is even, take the average of the two middle numbers. First quartile Q1: Find the median of the data vector to the left of the median Q2; if the number of data vectors is odd, the first quartile Q1 is the middle number; if it is even, the first quartile Q1 is the average of the two middle numbers; Third quartile Q3: Find the median of the data vector to the right of the median Q2. The calculation method is the same as the first quartile Q1.

9. The electric bed quality assessment system based on big data according to claim 6, characterized in that: The calculation formula of the long short-term memory network is as follows: Forget Gate: ， in, Represented as the output of the forget gate, the value is between 0 and 1, 1 means that the information of the previous moment is completely retained, and 0 means that it is completely forgotten; Represented as the weight matrix of the forget gate; Represented as the bias term of the forget gate; Represented as Sigmoid activation function; Represented as the hidden state of the previous moment, that is, the output of the previous moment; It is represented as the input at the current moment, which is the fault data detected in the box plot; Input Gate: ， in, Represented as the output of the input gate, the value is between 0 and 1, indicating the importance of the input at the current moment; Represented as the weight matrix of the input gate; Represented as the bias term of the input gate; Represented as candidate cell states; Represented as a weight matrix, responsible for connecting the hidden state of the previous moment and the current input Perform linear transformation; Represented as a bias term, it is used to adjust the calculation process of candidate cell states; It is represented as a hyperbolic tangent activation function, which maps the linear combination of candidate cell states to the range of [-1,1]; Update cell status: ， in, Represents the cell state at the current moment; Represents the cell state at the previous moment; Represented as the output of the forget gate; The output of the input gate; Represented as candidate cell states; Output Gate: ， The hidden state is calculated as: ， in, Represented as the output of the output gate; Represented as the hidden state at the current moment, that is, the output of LSTM, will be used as the input at the next moment; The calculation formula of the Pearson correlation coefficient is as follows: ， in, and are the data points of the i-th observation value of two different fault data in the original data; and is the mean of X and Y; ∑ represents the sum of all fault data; Among them, r=1 indicates a completely positive linear correlation; that is, when one fault data increases, the other fault data will also increase at a fixed ratio; r=-1 indicates a completely negative linear correlation; that is, when one fault data is present, the other fault data will decrease at a fixed ratio; r=0 means there is no linear correlation; that is, there is no linear relationship between the two fault data; r>0 indicates positive correlation, that is, when one fault data increases, the other fault data tends to increase; r<0 indicates negative correlation, that is, when one fault data increases, the other fault data tends to decrease; Correlation strength: 0.1<|r|<0.3 indicates weak correlation; 0.3<|r|<0.5 indicates moderate correlation; |r|>0.5 indicates strong correlation.

10. The electric bed quality assessment system based on big data according to claim 1, characterized in that: The calculation of the binary classification processing of the support vector machine is as follows: ， in, is the normal vector, i.e. the normal to the hyperplane; is the bias term; is the input sample, i.e., the abnormal duration and abnormal value; For positive samples , the constraints are: ; For negative samples , the constraints are: ; It is used to find a hyperplane given the abnormal duration and abnormal value.

Images