CN108836012B - Hardness-adjustable intelligent mattress with human muscle state detection function and control method thereof - Google Patents
Hardness-adjustable intelligent mattress with human muscle state detection function and control method thereof Download PDFInfo
- Publication number
- CN108836012B CN108836012B CN201810990599.6A CN201810990599A CN108836012B CN 108836012 B CN108836012 B CN 108836012B CN 201810990599 A CN201810990599 A CN 201810990599A CN 108836012 B CN108836012 B CN 108836012B
- Authority
- CN
- China
- Prior art keywords
- muscle
- action
- human
- layer
- human body
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
- A47C—CHAIRS; SOFAS; BEDS
- A47C27/00—Spring, stuffed or fluid mattresses or cushions specially adapted for chairs, beds or sofas
- A47C27/08—Fluid mattresses or cushions
- A47C27/10—Fluid mattresses or cushions with two or more independently-fillable chambers
-
- A—HUMAN NECESSITIES
- A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
- A47C—CHAIRS; SOFAS; BEDS
- A47C31/00—Details or accessories for chairs, beds, or the like, not provided for in other groups of this subclass, e.g. upholstery fasteners, mattress protectors, stretching devices for mattress nets
Landscapes
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses a hardness-adjustable intelligent mattress for detecting muscle state of human body and a control method thereof, which can collect muscle electric signals on the surface of the human body in a non-contact coupling mode through a specific functional layer of the hardness-adjustable intelligent mattress, and carry out effective interference shielding, anti-aliasing filtering, amplification and digital sampling on the weak and mixed muscle electric signals; the invention extracts frequency domain characteristic quantity by utilizing the muscle electric signal, further integrates a signal characteristic group of whole body muscle action, realizes the determination of action mode type by utilizing a BP neural network, further analyzes the sleep quality of a human body, can realize high-reliability detection and analysis aiming at human body fine action which is difficult to be accurately monitored in the prior art, more accurately judges the action type of the whole body of the human body, further ensures that the adjustment of the intelligent mattress with adjustable hardness is matched with the sleep state of the human body, and plays a role in obviously improving the sleep quality.
Description
Technical Field
The invention relates to the technical field of intelligent home furnishing, in particular to a hardness-adjustable intelligent mattress with human muscle state detection function and a control method thereof.
Background
The sleep is an important physiological activity required by everyone, the sleep can help the human body to recover fatigue and relieve emotion, and sufficient sleep is very necessary for the normal life of the human body. In modern society, people who suffer from insomnia, somnolence and other symptoms are rare, and often have great influence on the daytime lives of people, resulting in other mental and physical problems. Meanwhile, the health condition of the human body directly influences and determines the sleep quality, and a plurality of health problems such as muscle pain, frequent micturition and the like cause insomnia or low sleep quality.
The mattress is an important tool for sleeping and is closely related to the sleeping quality. The types of mattresses used by modern people tend to be diversified gradually, and mainly comprise: spring mattresses, palm mattresses, latex mattresses, water mattresses, magnetic mattresses, and the like. In recent years, once the intelligent mattress with adjustable hardness is released, the intelligent mattress is popular with consumers. As the name suggests, the hardness of the intelligent mattress with adjustable hardness can be adjusted, so that the intelligent mattress can adapt to different physiological curves, sleeping postures and habits of each person, and provides a personalized comfortable sleeping environment for each person. The structure of the intelligent mattress with adjustable hardness generally comprises a hardness adjusting layer, a comfortable contact layer and one or more functional layers; wherein the hardness adjusting layer adjusts the hardness of the mattress by adopting the principle of changing the air inflation quantity; the comfortable contact layer is contacted with the human body, and provides more comfortable contact touch feeling; and each functional layer can realize various functions related to intellectualization, such as pressure state detection, temperature regulation and the like.
The human body also makes various actions during sleeping, and particularly a series of actions are also unconsciously generated after sleeping. The frequency and amplitude of the action in the light sleep state are relatively large, and the frequency and amplitude of the action in the deep sleep state are relatively small. Meanwhile, under the conditions of uncomfortable body feeling, urine holding, uncomfortable body posture and the like caused by overhigh or overlow temperature, the human body can be stressed to act. The intelligent mattress can continuously monitor the human body action in the sleeping state, can know the sleeping depth rule of the user, analyze the reasons influencing the sleeping quality of the user, and give personalized solutions, including adaptive control on a soft and hard adjusting layer and a functional layer of the intelligent mattress.
At present, a pressure sensor can be arranged on a functional layer of the intelligent mattress with adjustable hardness for monitoring the actions of a human body in a sleeping state, the pressure change caused by the actions of the human body is measured, and the amplitude and the frequency of the actions of the human body are represented by the amplitude and the frequency of the pressure value change. However, the human body motion monitoring accuracy based on pressure measurement is low, the monitoring accuracy and real-time performance for human body fine motion are not sufficient, and it is difficult to effectively judge the human body motion position and motion mode.
The various actions of the human body are realized by muscle traction, the muscle actions can generate fine electric signals on the surface of the human body, and the muscle electric signals can generate difference along with different positions and modes of the human body actions and different amplitudes and frequencies of the actions. And, the intelligent mattress closely contacts and area of contact is big with the human body among the sleep process, provides convenience for the extraction of muscle signal of telecommunication, consequently, if can utilize this signal of telecommunication to monitor human muscle state among the sleep process, and then human sleep quality of analysis, compare prior art and can play obviously more excellent technological effect. However, the muscle electrical signal is weak and is mixed with a lot of noise, and the extraction and identification of the electrical signal have difficulties to be solved.
Disclosure of Invention
In view of this, the invention provides a hardness-adjustable intelligent mattress for detecting human muscle state and a control method thereof.
The invention provides a hardness-adjustable intelligent mattress for detecting human muscle state, which is characterized by comprising: a soft or hard adjusting layer, a comfortable contact layer and at least one functional layer; the hardness adjusting layer comprises a plurality of inflatable chambers which can be inflated and deflated independently, and the hardness of the corresponding mattress area of each inflatable chamber is adjusted by inflating or deflating each inflatable chamber to change the internal air pressure value of the chamber; the comfortable contact layer is arranged on the soft and hard adjusting layer and is directly contacted with the human body; the functional layers are embedded into the comfortable contact layer or the surface layer of the comfortable contact layer is laid, and each functional layer comprises a plurality of functional units and flexible wires connected with the functional units; the functional unit of the muscle state detection functional layer comprises a coupling capacitor patch electrode, an interference shielding shell, a signal amplification module, a demixing and filtering module and an analog-to-digital conversion module; the coupling capacitor patch electrode is coupled with the surface of human muscle to form a capacitor, senses a potential difference signal generated by muscle action, and transmits the potential difference signal to the signal amplification module; the signal amplification module is used for amplifying potential difference signals generated by muscle action and transmitting the amplified potential difference signals to the de-mixing filtering module; the de-mixing filtering module filters the amplified potential difference signal to eliminate the mixed signal and then transmits the mixed signal to the analog-to-digital conversion module; the analog-to-digital conversion module performs analog-to-digital conversion to generate digital signals representing muscle actions and outputs the digital signals through the flexible lead connected with each functional unit; the interference shielding shell is used for shielding external interference signals;
the adjustable intelligent mattress of softness still includes: the human body action recognition unit is used for receiving digital signals which are output by each functional unit of the muscle state detection functional layer and represent muscle actions, and recognizing human body action modes according to all the digital signals;
the sleep state analysis unit is used for analyzing the sleep state of the human body according to the human body action mode;
and the mattress adjusting unit is used for adapting to the sleeping state of the human body and adjusting the hardness adjusting layer and other functional layers of the intelligent mattress with adjustable hardness.
Preferably, the human motion recognition unit specifically includes: the muscle action signal characteristic extraction module is used for extracting characteristic quantities from the digital signals which are output by each functional unit and represent muscle actions, and integrating all the characteristic quantities to form a whole body muscle action signal characteristic group; and the action mode classification neural network is used for training by utilizing a training sample for recording the corresponding relation between the whole body muscle action signal characteristic group and the human body action mode type, and then judging the current human body action mode type according to the whole body muscle action signal characteristic group input in real time.
Further preferably, the muscle action signal feature extraction module is configured to calculate the feature quantity of the digital signal representing the muscle action output by each functional unit as follows:
wherein XPiCharacteristic quantity P representing digital signal representing muscle action output from the i-th functional uniti(f) A power spectrum representing the digital signal representing the muscle action output by the ith functional unit; wherein the value range of i is 1-n; the muscle action signal feature extraction module combines the feature values of all the functional units to generate a whole-body muscle action signal feature group XP＝{XP1，XP2…XPi…XPn}。
It is further preferred that the action pattern classification neural network comprises an input layer, a hidden layer and an output layer; the input layer corresponds to the input whole body muscle action signal characteristic group; the hidden layer is used for training by using the training sample to obtain the corresponding relation between the whole body muscle action signal characteristic group and the human body action mode type, and further determining the human body action mode type according to the real-time input whole body muscle action signal characteristic group; the output layer is used for outputting the human body action mode types.
Further preferably, the action pattern classification neural network uses training samples to input the samples of the whole body muscle action signal feature set and the corresponding human body action pattern types as learning samples into the BP neural network; performing forward conduction calculations: substituting the learning sample into a BP neural network, and sequentially calculating the numerical values of a hidden layer and an output layer; judging whether the deviation of the current round is less than or equal to a preset allowable deviation epsilon, if so, stopping iteration, and if not, continuing to execute reverse calculation, and changing the weight according to the learning rate; repeatedly learning, and continuously adjusting the weight values among the neurons of the BP neural network until the deviation reaches less than or equal to the preset allowable deviation, so that the training of the BP neural network is finished; and inputting the current real-time acquired whole body muscle action signal feature group serving as an input vector to the trained BP neural network, so that the BP neural network outputs the current human body action mode type.
Preferably, the sleep state analysis unit calculates a degree of deviation from the distribution of the human motion pattern types in the ideal sleep state, based on the distribution of the human motion pattern types within a predetermined statistical time period.
Further preferably, the sleep state analysis unit calculates the deviation degree according to a difference between the occurrence frequency of each human body action mode type in the statistical time duration and the occurrence frequency in an ideal sleep state, in combination with a normalized weight parameter of each human body action mode.
Further preferably, the mattress adjusting unit determines adjusting parameters for adjusting the soft and hard adjusting layers and other functional layers according to the deviation degree.
Preferably, the interference shielding case includes an insulating layer, a metal shielding layer and a silica gel outer cover layer in sequence from inside to outside.
Preferably, the other functional layers include a temperature adjustment layer, and the mattress adjustment unit adjusts the temperature of the temperature adjustment layer in response to the sleep state of the human body.
Furthermore, the invention provides a method for controlling the intelligent mattress with adjustable hardness based on human muscle state detection, which is characterized by comprising the following steps of:
coupling capacitor patch electrodes included in functional units of the hardness-adjustable intelligent mattress functional layer are coupled with the surface of human muscle to form a capacitor, and potential difference signals generated by muscle action are sensed;
amplifying potential difference signals generated by muscle action, filtering the amplified potential difference signals to eliminate mixed signals, and then performing analog-to-digital conversion to generate digital signals representing the muscle action;
receiving digital signals which are output by each functional unit and represent muscle actions, and identifying human action modes according to all the digital signals;
the human sleep state analysis module is used for analyzing the human sleep state according to the human action mode;
the intelligent mattress is suitable for the sleeping state of the human body, and adjusts the hardness adjusting layer and other functional layers of the hardness adjustable intelligent mattress.
Preferably, the recognizing the human motion pattern specifically includes: extracting characteristic quantities from the digital signals representing muscle actions output by each functional unit, and integrating all the characteristic quantities to form a whole body muscle action signal characteristic group; and after training the BP neural network by using a training sample for recording the corresponding relation between the whole body muscle action signal characteristic group and the human body action mode type, judging the current human body action mode type according to the whole body muscle action signal characteristic group input in real time.
Further preferably, the digital signal representing the muscle action output to each functional unit is calculated as follows:
wherein XPiCharacteristic quantity P representing digital signal representing muscle action output from the i-th functional uniti(f) A power spectrum representing the digital signal representing the muscle action output by the ith functional unit; wherein the value range of i is 1-n; and, the feature quantities of all the functional units are combined to generate a whole body muscle action signal feature group XP＝{XP1，XP2…XPi…XPn}。
Further preferably, the samples of the whole body muscle action signal characteristic group and the corresponding human body action pattern types are used as learning samples to be input into the BP neural network; performing forward conduction calculations: substituting the learning sample into a BP neural network, and sequentially calculating the numerical values of a hidden layer and an output layer; judging whether the deviation of the current round is less than or equal to a preset allowable deviation epsilon, if so, stopping iteration, and if not, continuing to execute reverse calculation, and changing the weight according to the learning rate; repeatedly learning, and continuously adjusting the weight values among the neurons of the BP neural network until the deviation reaches less than or equal to the preset allowable deviation, so that the training of the BP neural network is finished; and inputting the current real-time acquired whole body muscle action signal feature group serving as an input vector to the trained BP neural network, so that the BP neural network outputs the current human body action mode type.
Preferably, the analyzing the sleep state of the human body according to the human body action pattern specifically includes: and calculating the deviation degree of the human motion pattern type distribution in the ideal sleep state according to the distribution condition of the human motion pattern type in the preset statistical time length.
Further preferably, the deviation degree is calculated according to a difference between the occurrence frequency of each human body action mode type in the statistical time length and the occurrence frequency in an ideal sleep state, in combination with a normalized weight parameter of each human body action mode.
Further preferably, the adjusting of the hardness adjusting layer and other functional layers of the hardness adjustable intelligent mattress specifically comprises: and determining the adjusting parameters for adjusting the soft and hard adjusting layer and other functional layers according to the deviation degree.
Therefore, the muscle electrical signals on the surface of the human body can be collected in a non-contact coupling mode through the specific functional layer of the intelligent mattress with adjustable hardness in the sleeping process, and effective interference shielding, anti-aliasing filtering, amplification and digital sampling are carried out on the weak and mixed muscle electrical signals; the invention extracts frequency domain characteristic quantity by using the muscle electric signal, further integrates a signal characteristic group of whole body muscle action, realizes the determination of action mode type by using a BP neural network, and further analyzes the sleep quality of a human body.
The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
fig. 1 is a schematic diagram illustrating a general architecture of a soft and hard adjustable mattress for detecting human muscle state according to an embodiment of the present invention;
FIG. 2 is a schematic structural diagram of a soft and hard adjustable mattress body according to an embodiment of the invention;
FIG. 3 is a schematic diagram illustrating a functional layer detection grid structure of a mattress with adjustable hardness according to an embodiment of the invention;
fig. 4 shows a schematic structural diagram of a functional unit of a muscle state detection functional layer of a soft and hard adjustable mattress according to an embodiment of the present invention;
FIG. 5 is a circuit block diagram of a functional unit of a muscle status detection function layer according to an embodiment of the present invention;
fig. 6 shows a block diagram of a human body motion recognition unit according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
The embodiment of the invention provides an intelligent mattress with adjustable hardness and a human muscle state detection function. Referring to fig. 1, a general architecture diagram of the system is shown, and the system comprises a soft and hard adjustable mattress body 1, a human body action recognition unit 2, a sleep state analysis unit 3 and a mattress adjustment unit 4.
The hardness-adjustable mattress body 1 comprises a hardness-adjustable layer 101, a comfortable contact layer 102 and one or more functional layers 103, as shown in fig. 2. Wherein, the soft or hard adjusting layer 101 adjusts the soft or hard degree of the mattress by changing the air inflation amount. Specifically, the hardness adjusting layer is divided into a plurality of inflation chambers 101A which can be inflated and deflated independently, each inflation chamber 101A is provided with an air inlet and outlet valve, the valves are communicated with an inflation pump through air ducts, the internal air pressure of the inflation chambers 101A can be increased through inflation, so that the hardness of the mattress is increased, and the internal air pressure of the inflation chambers 101A can be reduced through deflation, so that the hardness of the mattress is reduced; since each chamber 101A can be independently adjusted in firmness, different positions of the entire mattress can have different firmness, and thus can be adapted to the physiological curve and sensory preference of the human body. The comfortable contact layer 102 is disposed on the softness adjusting layer 101, and comprises a multi-layer structure of latex, cotton wadding, surface skin-friendly fabric and the like, and the layer is in contact with the human body to provide a relatively comfortable touch feeling. One or more functional layers 103 may be embedded within the comfort contact layer, each of said functional layers 103 comprising a functional unit 103A and a flexible conductor 103B connecting the respective functional unit 103A. As shown in fig. 3, the functional layer 103 of the soft and hard adjustable mattress body 1 is divided into a plurality of detection grids 103C distributed in a matrix, and one functional unit 103A is disposed in each detection grid 103C. The flexible lead 103B is used for connecting the functional unit 103A, and realizes input and output of signals of the functional unit 103A. For example, the functional layer 103 may be a temperature adjustment functional layer, the functional unit 103A of the functional layer is an electric heater, and the functional unit 103A adjusts its own heating temperature according to a temperature adjustment signal transmitted by the flexible wire 103B, thereby creating a sleep temperature environment suitable for the user's body feeling.
Moreover, the functional layer 103 of the soft and hard adjustable mattress body at least comprises a muscle state detection functional layer, and the functional layer can be laid on the surface layer of the comfortable contact layer. The functional unit 103A of the functional layer for detecting muscle status is configured as shown in fig. 4, and includes a coupling capacitor patch electrode 103A1, an interference shielding case 103A2, and a circuit board 103A 3. As shown in fig. 5, the circuit board 103A3 is provided with a signal amplification module 103a4, a downmix filter module 103a5, and an analog-to-digital conversion module 103a 6. The coupling capacitance patch electrode 103A1 forms capacitance by coupling with the surface of the muscle of the human body, senses a potential difference signal generated by the action of the muscle and transmits the potential difference signal to the signal amplification module 103A 4; the coupling capacitor patch electrode 103a1 does not need to be directly attached to the skin of the human body, and can be coupled to the surface of the human body through clothing to form a capacitor body, thereby sensing an electric signal on the surface of the human body. The electric signals generated by the muscle movement of the human body are weak electric signals, are mixed with various noise electric signals, are easy to be interfered, and have low signal-to-noise ratio, the frequency range is generally between 50Hz and 2KHz, and the signal amplitude is 0-2 mV. The signal amplifying module 103a4 is configured to amplify the potential difference signal generated by the muscle action, and transmit the amplified potential difference signal to the downmix filtering module 103a 5. The anti-aliasing filtering module 103a5 adopts a low-pass filter with a cut-off frequency of about 2KHz, filters the amplified potential difference signal to remove aliasing signals, and then transmits the aliasing signals to the analog-to-digital conversion module 103a 5. The analog-to-digital conversion module 103a5 performs analog-to-digital conversion on the filtered potential difference signal by using a high-precision AD converter to generate a digital signal representing muscle action, and outputs the digital signal to the human body action recognition unit 2 through the flexible lead 103B connected to each functional unit. The interference shielding shell 103A2 is used for shielding external interference signals and ensuring the accuracy of electric signals generated by muscle actions, and the interference shielding shell 103A2 sequentially comprises an insulating layer, a metal shielding layer and a silica gel outer cover layer from inside to outside.
The human body motion recognition unit 2 is configured to receive the digital signal indicating the muscle motion output from each functional unit 103A of the muscle state detection functional layer, and recognize the human body motion pattern from all the digital signals. As shown in fig. 6, the human motion recognition unit 2 specifically includes: a muscle action signal feature extraction module 201 and an action pattern classification neural network 202.
The muscle action signal feature extraction module 201 is configured to extract a feature quantity from the digital signal representing muscle action output by each functional unit 103A; n functional units 103A are provided, and the muscle action signal feature extraction module 201 calculates feature quantities of the digital signals representing muscle actions output by each functional unit as follows:
wherein XPiCharacteristic quantity P representing digital signal representing muscle action output from the i-th functional uniti(f) A power spectrum representing the digital signal representing the muscle action output by the ith functional unit; wherein the value range of i is 1-n. Further, the muscle action informationThe number feature extraction module 201 integrates feature quantities of all n functional units to form a whole body muscle action signal feature group XP＝{XP1，XP2…XPi…XPn}。
The action pattern classification neural network 202 is a BP neural network including an input layer, a hidden layer, and an output layer; the input layer corresponds to the input whole body muscle action signal characteristic group XP(ii) a The hidden layer is used for training by using the training sample to obtain the corresponding relation between the whole body muscle action signal characteristic group and the human body action mode type, and further according to the real-time input whole body muscle action signal characteristic group XPDetermining the type of the human body action mode; the output layer is used for outputting the human body action mode types. The training sample records the corresponding relationship between the whole body muscle action signal feature group and the human body action pattern type, for example, in the design stage of the mattress, by performing actual measurement, a tester executes various human body action pattern types, such as whole body action types of changing from horizontal to lateral, left to right, lateral to horizontal, lower limb body action types of kicking legs, upper limb body action types of swing arms, micro-motion body action types of slight leg tremor, and the like, and records the whole body muscle action signal feature group generated by each human body action pattern type, thereby generating corresponding relationship items between a human body action pattern type and the whole body muscle action signal feature group, and the training sample is formed by accumulating a sufficient number of the corresponding relationship items, such as 10000 items. And then inputting the whole body muscle action signal feature group in the accumulated training sample into a BP neural network, repeatedly adjusting and training the weight and deviation among all neurons of the BP neural network, enabling the type of the human action pattern output by the BP neural network to be as close as possible to the type of the human action pattern recorded by the training sample, and when the degree of the proximity reaches the expectation, considering that the algorithm training is finished, and storing the corresponding weight and deviation. The trained BP neural network can be used for acquiring a whole body muscle action signal characteristic group X in real timeP＝{XP1，XP2…XPi…XPnAnd outputting the corresponding human body action mode type. More specifically, the process of training the motion pattern classification neural network 202 and performing human motion pattern type recognition includes the steps of:
(1) using training sample data, and inputting samples of a whole body muscle action signal characteristic group and corresponding human body action mode types into the BP neural network as learning samples; wherein, the inputted whole body muscle action signal characteristic group Xp＝{xp1...xpNIn which xp1,xp2,……xpNAs one dimension of the input vector, respectively corresponding to the characteristic quantity of the digital signal which is output by the ith functional unit and represents the muscle action; the expected output value corresponding to the feature set of the input is tpmThe expected output value is the type of human action pattern corresponding to the feature set in the sample, tpmThe different values of (a) represent the human action pattern types.
(2) Performing forward conduction calculations: the BP neural network input layer is provided with N input neurons, the hidden layer is provided with K hidden layer neurons, the output layer is provided with M output neurons, and then the numerical values of the hidden layer and the output layer are sequentially calculated as follows:
wherein w1nkIs the weight between the nth neuron of the input layer and the kth neuron of the hidden layer, O1pkIs the output of the k neuron of the hidden layer; w2kmIs the weight between the k neuron of the hidden layer and the m neuron of the output layer, O2pmIs the output, activation function of the mth output layer neuron
i represents the ith round of training;
(3) performing deviation calculation:
judging whether the deviation of the current round (the ith round) is less than or equal to a preset allowable deviation epsilon, if so, stopping iteration, and if not, continuing the following process;
(4) performing a reverse calculation:
wherein the learning rate is mu, the learning rate is,
-δpm(i)＝(tpm-O2pm(i))O2pm(i)(1-O2pm(i))，
the weight is changed as follows:
w1nk(i+1)＝w1nk(i)+Δw1nk(i+1)
w2km(i+1)＝w2km(i)+Δw2km(i+1)
(5) returning to the step (2), and repeating the (i + 1) th round of learning.
And repeatedly learning, and continuously adjusting the weight values among the neurons until the deviation reaches less than or equal to the preset allowable deviation epsilon, so that the training of the BP neural network is finished. Thus, for the current real-time acquired full-body muscle action signal feature set XP＝{XP1，XP2...XPi...XPnAnd inputting the input vector into the trained BP neural network, and taking the output of the neural network as the current human body action mode type of the user.
Therefore, a detection time point may be set at every predetermined unit time (for example, 5 seconds), and the digital signal representing the muscle action is collected and outputted once by each functional unit 103A of the muscle state detection functional layer at the detection time point, and the human body action pattern type at the detection time point is identified by the human body action identifying unit 2.
The sleep state analysis unit 3 receives the human body motion pattern type output by the human body motion recognition unit 2 at each detection time point; further, the sleep state analysis unit 3 counts the count value of the number of occurrences of each human motion pattern type for a predetermined statistical time period (for example, within 5 minutes from the current time), thereby determining the distribution of the human motion pattern types. For example, a general body motion type of 1, a lower limb body motion type of 15, an upper limb body motion type of 2, and a micromotion body motion type of 38 occur within a statistical time period. The sleep state analysis unit 3 calculates the degree of deviation between the distribution and the distribution of the human body action mode types in the ideal sleep state; that is, the sleep state analysis unit 3 calculates the deviation degree according to the difference between the occurrence frequency of each human body action mode type in the statistical time length and the occurrence frequency in the ideal sleep state, and by combining the normalized weight parameter of each human body action mode. For example, presetting the general human body action type 0 time, the lower limb human body action type 5 times, the upper limb human body action type 5 times, the inching human body action type 10 times in the ideal sleep state, the difference between the occurrence frequency of each human body action mode type in the counting time length and the occurrence frequency in the ideal sleep state is respectively a general human body action type +1, a lower limb human body action type +10, an upper limb human body action type-3 and a jogging human body action type +28, the difference is further combined with the normalized weight parameter of each human body action mode type, for example, the weighting parameter for the general body motion type is 10, the weighting parameter for the lower limb body motion type and the upper limb body motion type is 5, the weighting parameter for the inching body motion type is 1, the calculated degree of deviation is (+1 × 10) + (+10 × 5) + (-3 × 5) + (28 × 1) ═ 73.
And the mattress adjusting unit 4 determines adjusting parameters for adjusting the soft and hard adjusting layers and other functional layers according to the deviation degree, and then adjusts the soft and hard adjusting layers and other functional layers according to the adjusting parameters. For example, the higher the deviation value is, the more the number of actions and the larger the amplitude of the actions in the sleeping process of the user are, which indicates that the sleeping depth state of the user is not good, the mattress adjusting unit 4 can reduce the air inflation amount of the soft and hard adjusting layer, so that the mattress is softer, which is beneficial for the user to enter the deep sleeping state, and the reduction amplitude of the air inflation amount is proportional to the deviation degree, that is, the higher the deviation degree is, the larger the reduction amplitude of the air inflation amount is, and the mattress becomes softer; conversely, if the deviation degree is lower, the reduction amplitude of the inflation quantity is smaller; if the degree of deviation is 0 or a negative value, the adjustment of the air inflation amount of the soft and hard adjustment layers may not be performed. The other functional layers comprise temperature adjusting layers, the mattress adjusting unit can adjust the temperature value of the temperature adjusting layers according to the deviation degree, and similarly, the higher the deviation degree is, the lower the sleep depth state of the user is, and the larger the change amount of the temperature is; conversely, the lower the deviation degree is, the smaller the temperature change amount is; if the degree of deviation is 0 or a negative value, no adjustment may be performed on the temperature.
Therefore, the muscle electrical signals on the surface of the human body can be collected in a non-contact coupling mode through the specific functional layer of the intelligent mattress with adjustable hardness in the sleeping process, and effective interference shielding, anti-aliasing filtering, amplification and digital sampling are carried out on the weak and mixed muscle electrical signals; the invention extracts frequency domain characteristic quantity by using the muscle electric signal, further integrates a signal characteristic group of whole body muscle action, realizes the determination of action mode type by using a BP neural network, and further analyzes the sleep quality of a human body.
The various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that the implementation may be in practice using a microprocessor or a Digital Signal Processor (DSP). The present invention may also be embodied as apparatus or device programs (e.g., computer programs and computer program products) for performing a portion or all of the methods described herein. Such programs implementing the present invention may be stored on computer-readable media or may be in the form of one or more signals. Such a signal may be downloaded from an internet website or provided on a carrier signal or in any other form.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names.
Claims (5)
1. The utility model provides an adjustable intelligent mattress of hardness of human muscle state detection which characterized in that includes: a soft or hard adjusting layer, a comfortable contact layer and at least one functional layer; the hardness adjusting layer comprises a plurality of inflatable chambers which can be inflated and deflated independently, and the hardness of the corresponding mattress area of each inflatable chamber is adjusted by inflating or deflating each inflatable chamber to change the internal air pressure value of the chamber; the comfortable contact layer is arranged on the soft and hard adjusting layer and is directly contacted with the human body; the functional layers are embedded into the comfortable contact layer or the surface layer of the comfortable contact layer is laid, and each functional layer comprises a plurality of functional units and flexible wires connected with the functional units;
the functional unit of the muscle state detection functional layer comprises a coupling capacitor patch electrode, an interference shielding shell, a signal amplification module, a demixing and filtering module and an analog-to-digital conversion module; the coupling capacitor patch electrode is coupled with the surface of human muscle to form a capacitor, senses a potential difference signal generated by muscle action, and transmits the potential difference signal to the signal amplification module; the interference shielding shell comprises an insulating layer, a metal shielding layer and a silica gel outer cover layer from inside to outside in sequence, and the signal amplification module is used for amplifying potential difference signals generated by muscle action and transmitting the amplified potential difference signals to the de-mixing filtering module; the de-mixing filtering module filters the amplified potential difference signal to eliminate the mixed signal and then transmits the mixed signal to the analog-to-digital conversion module; the analog-to-digital conversion module performs analog-to-digital conversion to generate digital signals representing muscle actions and outputs the digital signals through the flexible lead connected with each functional unit; the interference shielding shell is used for shielding external interference signals;
the adjustable intelligent mattress of softness still includes: the human body action recognition unit is used for receiving digital signals which are output by each functional unit of the muscle state detection functional layer and represent muscle actions, and recognizing human body action modes according to all the digital signals; the human body action recognition unit specifically comprises: the muscle action signal characteristic extraction module is used for extracting characteristic quantities from the digital signals which are output by each functional unit and represent muscle actions, and integrating all the characteristic quantities to form a whole body muscle action signal characteristic group; the action mode classification neural network is used for judging the current human action mode type according to a real-time input whole body muscle action signal characteristic group after training by utilizing a training sample for recording the corresponding relation between the whole body muscle action signal characteristic group and the human action mode type;
the sleep state analysis unit is used for analyzing the sleep state of the human body according to the human body action mode; the sleep state analysis unit calculates the deviation degree between the distribution condition of the human motion mode types in the preset statistical time and the distribution of the human motion mode types in the ideal sleep state;
and the mattress adjusting unit is used for adapting to the sleeping state of the human body and adjusting the hardness adjusting layer and other functional layers of the hardness-adjustable intelligent mattress, and the mattress adjusting unit determines adjusting parameters for adjusting the hardness adjusting layer and other functional layers according to the deviation degree and then adjusts the hardness adjusting layer and other functional layers according to the adjusting parameters.
2. The intelligent mattress for detecting muscle state according to claim 1, wherein the muscle action signal feature extraction module is configured to calculate the feature quantity of the digital signal representing muscle action output by each functional unit as follows:
wherein XPiCharacteristic quantity P representing digital signal representing muscle action output from the i-th functional uniti(f) A power spectrum representing the digital signal representing the muscle action output by the ith functional unit; wherein the value range of i is 1-n; the muscle action signal feature extraction module combines the feature values of all the functional units to generate a whole-body muscle action signal feature group
XP＝{XP1，XP2...XPi...XPn}。
3. The human muscle state detection smart mattress of claim 2, wherein the motion pattern classification neural network comprises an input layer, a hidden layer and an output layer; the input layer corresponds to the input whole body muscle action signal characteristic group; the hidden layer is used for training by using the training sample to obtain the corresponding relation between the whole body muscle action signal characteristic group and the human body action mode type, and further determining the human body action mode type according to the real-time input whole body muscle action signal characteristic group; the output layer is used for outputting the human body action mode types.
4. A hardness-adjustable intelligent mattress control method based on human muscle state detection is characterized by comprising the following steps:
coupling capacitor patch electrodes included in functional units of the hardness-adjustable intelligent mattress functional layer are coupled with the surface of human muscle to form a capacitor, and potential difference signals generated by muscle action are sensed;
when a potential difference signal generated by muscle action is induced by using the coupling capacitor patch electrode, an external interference signal is shielded;
amplifying potential difference signals generated by muscle action, filtering the amplified potential difference signals to eliminate mixed signals, and then performing analog-to-digital conversion to generate digital signals representing the muscle action;
receiving digital signals which are output by each functional unit and represent muscle actions, and identifying human action modes according to all the digital signals;
the human sleep state analysis module is used for analyzing the human sleep state according to the human action mode;
the method for analyzing the human sleep state according to the human action mode specifically comprises the following steps: calculating the deviation degree of the human motion mode type distribution in the ideal sleep state according to the distribution condition of the human motion mode type in the preset statistical time;
the identification of the human action mode specifically comprises the following steps: extracting characteristic quantities from the digital signals representing muscle actions output by each functional unit, and integrating all the characteristic quantities to form a whole body muscle action signal characteristic group; after training the BP neural network by using a training sample recording the corresponding relation between the whole body muscle action signal characteristic group and the human body action mode type, judging the current human body action mode type according to the whole body muscle action signal characteristic group input in real time;
the intelligent mattress is suitable for the sleeping state of the human body, and adjusts the hardness adjusting layer and other functional layers of the hardness adjustable intelligent mattress.
5. The intelligent mattress control method based on human muscle state detection and adjustable in hardness as claimed in claim 4, wherein the digital signal representing muscle action output to each functional unit is characterized by calculating characteristic quantity as follows:
wherein XPiCharacteristic quantity P representing digital signal representing muscle action output from the i-th functional uniti(f) A power spectrum representing the digital signal representing the muscle action output by the ith functional unit; wherein the value range of i is 1-n; and, the feature quantities of all the functional units are combined to generate a whole body muscle action signal feature group XP＝{XP1，XP2...XPi...XPn}。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810990599.6A CN108836012B (en) | 2018-08-28 | 2018-08-28 | Hardness-adjustable intelligent mattress with human muscle state detection function and control method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810990599.6A CN108836012B (en) | 2018-08-28 | 2018-08-28 | Hardness-adjustable intelligent mattress with human muscle state detection function and control method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108836012A CN108836012A (en) | 2018-11-20 |
| CN108836012B true CN108836012B (en) | 2021-09-07 |
Family
ID=64189738
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810990599.6A Active CN108836012B (en) | 2018-08-28 | 2018-08-28 | Hardness-adjustable intelligent mattress with human muscle state detection function and control method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108836012B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102458239A (en) * | 2009-06-05 | 2012-05-16 | 皇家飞利浦电子股份有限公司 | Capacitive sensing system |
| CN105310681A (en) * | 2014-06-23 | 2016-02-10 | 三星电子株式会社 | Bioelectrode, and method and apparatus for processing biosignal using the same |
| CN106068097A (en) * | 2014-02-20 | 2016-11-02 | 佛吉亚汽车座椅有限责任公司 | Integrate the seat having sensor |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102247063A (en) * | 2010-05-21 | 2011-11-23 | 许汉忠 | Active supporting device of healthy bed and method thereof |
| TWM402038U (en) * | 2010-10-25 | 2011-04-21 | Jing Zhan Biotechnology Co Ltd | Pneumatic device for modularized air cushion |
| CH707087A1 (en) * | 2012-10-10 | 2014-04-15 | Raffael Kuster | Agitation pad. |
| CN104257159B (en) * | 2014-09-05 | 2017-06-16 | 昆山合才智能科技有限公司 | Gather method, sping mattress and the system of physiology signal |
| CN204218446U (en) * | 2014-09-05 | 2015-03-25 | 杨松 | Gather mat and the system of physiology signal |
| CN104732250B (en) * | 2015-03-25 | 2018-09-25 | 安徽农业大学 | The sleeping position and sleep behavior of a kind of noiseless sleep test recognition methods |
| CN106388819A (en) * | 2016-10-28 | 2017-02-15 | 许昌学院 | Human upper limb muscle state monitoring system based on surface electromyogram signals and judging method thereof |
| CN106667435A (en) * | 2016-12-17 | 2017-05-17 | 复旦大学 | Intelligent sensing mattress for monitoring sleep |
| CN107126005A (en) * | 2017-07-03 | 2017-09-05 | 广东葆康信息科技有限公司 | Biomethanics mattress and system |
| CN108420228A (en) * | 2018-04-13 | 2018-08-21 | 浙江想能云软件股份有限公司 | A kind of soft or hard adjustable bed mattess of intelligence and its monitoring method of sleep state monitoring |
-
2018
- 2018-08-28 CN CN201810990599.6A patent/CN108836012B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102458239A (en) * | 2009-06-05 | 2012-05-16 | 皇家飞利浦电子股份有限公司 | Capacitive sensing system |
| CN106068097A (en) * | 2014-02-20 | 2016-11-02 | 佛吉亚汽车座椅有限责任公司 | Integrate the seat having sensor |
| CN105310681A (en) * | 2014-06-23 | 2016-02-10 | 三星电子株式会社 | Bioelectrode, and method and apparatus for processing biosignal using the same |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108836012A (en) | 2018-11-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109157194B (en) | System and method for acquiring and analyzing human health data of mattress with adjustable hardness | |
| CN108042108B (en) | Sleep quality monitoring method and system based on body vibration signals | |
| Lee et al. | Estimation of body postures on bed using unconstrained ECG measurements | |
| US10973344B2 (en) | Bedding system with a CNN based machine vision process | |
| CN107137071B (en) | Method for calculating short-term heart rate value by analyzing heart attack signal | |
| CN108420228A (en) | A kind of soft or hard adjustable bed mattess of intelligence and its monitoring method of sleep state monitoring | |
| US20210106256A1 (en) | Posture determination apparatus | |
| JP3543688B2 (en) | Environmental control equipment and care homes | |
| CN107049699A (en) | A kind of measuring method of the intelligent chaise longue pad of hypnosis and its heart rate and respiratory wave | |
| US11406304B2 (en) | Systems and methods for physiological sign analysis | |
| KR101857572B1 (en) | Smart pillow | |
| CN108549834A (en) | A kind of human body sitting posture recognition methods and its system based on flexible sensor | |
| CN109528159A (en) | A kind of sleep quality and respiration monitoring system and method based on bed body | |
| JP3849152B2 (en) | Biological signal detector | |
| CN108836012B (en) | Hardness-adjustable intelligent mattress with human muscle state detection function and control method thereof | |
| Febriana et al. | Sleep monitoring system based on body posture movement using Microsoft Kinect sensor | |
| TWI629049B (en) | A method for analyzing a heart shock signal for calculating a short-term heart rate value | |
| CN210990210U (en) | Non-inductive sleep monitoring device and monitoring system | |
| JP2008080071A (en) | Evaluation device for quality of sleep | |
| CN209677917U (en) | Intelligent electric bed | |
| CN116839764A (en) | Method and device for detecting use state of piezoelectric cloth mattress and intelligent terminal | |
| CN114937341B (en) | Wheelchair fall risk monitoring method and system based on multidimensional force | |
| JP4172193B2 (en) | Massage machine | |
| Kimura et al. | Development of an unobtrusive vital signs detection system using conductive fiber sensors | |
| JP4017915B2 (en) | Massage equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |