Intelligent device and method for judging animal emotion based on multiple signals

By combining bone conduction microphone and brain reaction monitoring technology, combined with skin electrical and myoelectric signals, the problem of insufficient judgment of single factors in the existing smart dog chain technology is solved, and accurate monitoring and control of dog emotions is achieved, ensuring humans Safety.

CN118948280APending Publication Date: 2024-11-15NORTHEAST AGRICULTURAL UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411056440.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The existing intelligent dog chain technology only judges the dog's emotions through a single factor such as dog barking or physiological indicators. It has insufficient accuracy, especially in complex environments, and it is difficult to denoise signal, so it is impossible to achieve accurate mood monitoring and control.

Method used

Combining bone conduction microphone and brain-based electroskin and electromyography monitoring, emotions are judged through respiratory heart rate changes, and factor adaptability algorithm is designed to judge emotional state using skin electroskin and electromyography reactions to improve the accuracy of emotional monitoring.

Benefits of technology

It realizes accurate judgment and monitoring of dog emotions in complex environments, improves the accuracy and stability of emotional monitoring, and takes timely measures to prevent dogs from attacking humans.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118948280A_ABST
    Figure CN118948280A_ABST
Patent Text Reader

Abstract

The invention discloses an intelligent device and method for judging animal emotion based on multiple signals, and belongs to the technical field of animal health monitoring and feeding. The device comprises a first signal collecting and processing module, a second signal collecting and processing module, an arithmetic unit module, an actuator module and a recall function module. According to the method, firstly, four specific physiological indexes are collected and imported into a processor, then fuzzy distribution of input quantity and output quantity is defined, a triangular membership function is established, a fuzzy rule is established, a first emotional anomaly degree and a second emotional anomaly degree are solved, and finally, the total emotional anomaly degree of a dog is calculated. According to the invention, a bone conduction microphone is combined with skin electricity monitoring and myoelectricity monitoring based on brain response, bark signals are collected by using a bone conduction voice technology, emotion changes are judged according to the change degree of breathing heart rate, and meanwhile, skin electricity response and myoelectricity response are added into the judgment of the emotional state, so that the judgment accuracy is improved. And a corresponding factor adaptive degree algorithm is designed, so that the function of theoretically judging whether the emotion is changed or not is realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention discloses an intelligent device and method for judging animal emotions based on multiple signals, belonging to the technical field of animal health monitoring and breeding. Background Art

[0002] Pet keeping has become an indispensable part of many families. However, abnormal emotions of pets can bring potential safety hazards to society and affect the physical and mental health of pets themselves. Therefore, pets often need to wear collars and / or chains. Taking dogs as an example, collars and dog chains have almost become a necessity for dog owners. This is not only a guarantee for the health of the dogs themselves, but also a guarantee for the stability of the social environment and the harmonious coexistence of people and dogs. However, ordinary dog ​​chains only have the function of tying dogs to one's side for care, and cannot monitor the dog's emotions and physiological conditions. Monitoring the dog's emotions can not only protect the dog's own physical and psychological conditions, but also guarantee the safety of life and property of people around it. Therefore, it is of great significance to accurately monitor the dog's mental state.

[0003] Smart dog chains can detect physical indicators or monitor a dog's psychological state. Existing smart dog chain technology only judges health status through physical indicators and monitors the dog's emotions based on a single factor, namely, the dog's barking. The accuracy of dog emotion monitoring needs to be improved, and precise control cannot be achieved when the dog's emotions are abnormal.

[0004] The existing technologies and products can be roughly divided into the following aspects:

[0005] First, use voice recognition technology to determine the dog’s emotional state

[0006] Dogs express their emotional state through barking. Therefore, by building a database of barking sounds of different types of dogs, analyzing the differences in the tones of dog barking under different emotions, and designing and implementing a dedicated algorithm to distinguish dog emotions, the dog's emotional state can be determined. In 2017, PetpulsLab established a database of barking sounds of 10,000 dogs from 50 breeds, and tried to develop an algorithm to analyze dog emotions. In 2020, Petpuls designed a new smart collar using speech recognition technology and AI translation. The collar uses an air conduction microphone to monitor dog barking to detect five canine emotions: happiness, relaxation, anxiety, anger, and sadness, and tracks the dog's physical activity and rest.

[0007] However, judging emotions based solely on dog barking has its limitations. In a dog's daily behavior, high-pitched barking represents emotions such as happiness and curiosity, while low-pitched barking represents a warning of danger. When a dog is in a more excited state, the precursors of his offensive behavior become high-frequency gnashing of teeth and high-pitched loud scolding. In addition, studies have shown that the voice of each dog is unique, and different breeds of dogs have different personalities and habits, and the way of emotional translation should also be different. Therefore, it is inaccurate to judge emotional state only by sound.

[0008] In addition, the air conduction microphones used in traditional smart collars have great limitations in collecting sound. When there are many dogs around and the environment is noisy, using air conduction microphones to collect sound will cause problems such as complex and difficult signal denoising.

[0009] Second, physiological indicators such as heart rate and breathing are used to detect the dog's physical condition

[0010] The pet smartwatch iPetWear uses low-power physiological detection radar technology to accurately obtain its cardiac shock waves, respiratory waves and body movement waves by scanning the dog's body surface at high speed. It can not only accurately detect the dog's heart rate and respiratory rate, but also analyze physiological disease indicators such as heart rate variability (HRV), respiratory variability (BRV), and apnea through AI data mining. Among them, the displacement resolution of heart rate and respiratory rate measurement data reaches 0.02mm, and the heart rate detection range can reach more than 165 times.

[0011] When an animal begins to feel nervous, its heartbeat and breathing rate will speed up. This change in rhythm is triggered by sympathetic nerve activity. Therefore, monitoring the two physiological indicators of breathing and heart rate can also be applied to emotion detection.

[0012] Through the analysis of existing technologies, we can know that:

[0013] Bone conduction microphones are more effective in collecting dog barking, thereby enabling remote monitoring of the owner and accurate collection of dog barking signals. In addition, the air conduction voice microphones used in traditional smart collars have great limitations in collecting sound. When there are many dogs around and the environment is noisy, using air conduction voice microphones to collect sound will result in complex and difficult signal denoising. The bone conduction voice device used in this product does not use air as the medium for collecting sound, but uses a highly sensitive vibration sensor to collect bone vibrations, which are then converted into audio signals, shielding the noise from the sound source and preventing the transmission of environmental noise at both ends of the communication system from the source. Therefore, even if the dog barking signal is collected in a strong noise environment, it can be transmitted clearly, so that the dog barking sound collected and fed back to the owner is clear and the collected dog barking signal is accurate.

[0014] Skin electricity monitoring and myoelectric monitoring based on brain reactions make it more accurate to judge the dog's emotions. The emotional state is judged by the four-fold detection indicators of skin electricity, myoelectricity, breathing frequency, and barking tone, and combined with the owner's subjective judgment through screen monitoring, the dog can be recalled and emergency paralysis treatment can be carried out in abnormal emotional states, ensuring real-time monitoring and management of the dog in a safe state.

[0015] If the above two-phase technology can be combined and an algorithm that can achieve 1+1>2 can be developed, it will be helpful to theoretically determine whether the dog's mood has changed. Summary of the invention

[0016] In order to achieve the above objectives, the present invention combines a bone conduction microphone with skin electricity monitoring and myoelectric monitoring based on brain reactions, uses bone conduction voice technology to collect dog barking signals, and judges emotional changes by the degree of changes in breathing and heart rate. At the same time, the skin electricity reaction and myoelectric reaction are innovatively added to the judgment of emotional state, and a corresponding factor adaptation degree algorithm is designed to achieve the function of theoretically judging whether the emotion has changed.

[0017] The object of the present invention is achieved in that:

[0018] An intelligent device for judging animal emotions based on multiple signals, comprising a signal acquisition and processing module 1, a signal acquisition and processing module 2, an operator module, an actuator module and a recall function module;

[0019] The signal acquisition module 1 includes a bone conduction voice submodule, a sound processing submodule, a heart rate monitoring submodule and a breathing monitoring submodule; it is used to collect animal bone vibration signals and convert them into sound signals, and judge and analyze animal emotions by monitoring heart rate and breathing;

[0020] The signal acquisition module 2 includes a skin conductance submodule and a muscle conductance submodule; the skin conductance submodule is used to measure the change of skin resistance caused by the activity of the autonomic nervous system and the change of sweat gland secretion; the muscle conductance submodule is used to measure the change of muscle potential caused by brain excitement;

[0021] The operator processing module includes a data acquisition submodule, a function operation submodule and a numerical judgment submodule; the data acquisition submodule is used to collect animal physiological index signals; the function operation submodule is used to define the fuzzy distribution of input and output quantities and establish a triangular membership function; the numerical judgment submodule is used to judge the degree of emotional abnormality;

[0022] The actuator module includes a scene acquisition module and a medical processing module;

[0023] The recall function module includes a GPS positioning submodule and a recall function speaker submodule; the scene acquisition module is used to collect environmental scenes within a radius of 50 meters around the animal's location; the medical processing module is used to install an anesthesia device and determine the timing of starting anesthesia.

[0024] In the above-mentioned intelligent device for judging animal emotions based on multiple signals, the skin conductance submodule is installed at a 1 / 4 position between the front and the back of the animal's neck close to the front; the muscle conductance submodule is installed at a 1 / 4 position between the front and the back of the animal's neck close to the front.

[0025] In the above-mentioned intelligent device for judging animal emotions based on multiple signals, a microcurrent measurement circuit is installed between the skin conductance submodule and the muscle conductance submodule.

[0026] In the above-mentioned intelligent device for judging animal emotions based on multiple signals, the bone conduction voice submodule is installed in the front part of the animal's neck to conduct the mechanical vibration of the bones; the heart rate monitoring submodule and the breathing monitoring submodule are installed at the 1 / 4 of the right side behind the animal's neck.

[0027] The above-mentioned intelligent device for judging animal emotions based on multiple signals, the device scene acquisition module and medical processing module have an automatic start function, and a green LED and a photosensitive element are installed inside the device as photoelectric sensors for detecting the position of animal blood vessels.

[0028] The above-mentioned intelligent device for judging animal emotions based on multiple signals is connected to one or more remote control terminals, and the remote control terminals synchronize the physiological indicator data of the animals.

[0029] The physiological indicator data include the dog's name, gender, owner's name, contact information, whether all three shots of vaccine have been received, whether rabies vaccine has been injected on time, medical history, and precautions.

[0030] An intelligent method for judging animal emotions based on multiple signals, comprising the following steps:

[0031] Step a: Collect four specific physiological indicators and import them into the processor

[0032] The four specific physiological indicators are: skin electricity factor x1, myoelectric factor x2, respiratory heart rate change factor x3 and voice pitch change factor x4;

[0033] Step b: Define the fuzzy distribution of input and output quantities and establish the triangular membership function

[0034] The input is the four specific physiological indicators, and the output is the first abnormal emotion degree u1 and the second abnormal emotion degree u2;

[0035] Let the skin electrical factor x1∈[0,2], unit μs, the fuzzy subsets of the skin electrical factor x1 are the small range of skin conductance difference SSCD, the medium range of skin conductance difference MSCD and the large range of skin conductance difference LSCD; the fuzzy subsets of the myoelectric factor x2 are the small range of myoelectric conductance difference SMCD, the medium range of myoelectric conductance difference MMCD and the large range of myoelectric conductance difference LMCD;

[0036] The membership function of the skin electrical triangle is:

[0037] SSCD=-2x1+1,,x1∈[0,0.5]

[0038]

[0039] The membership function of the myoelectric triangle is:

[0040] SMCD=-2x2+1,,x2∈[0,0.5]

[0041]

[0042] The first abnormal emotion degree u1 is obtained by the skin electricity factor x1 and the myoelectric factor x2, and u1∈[0,1], unit %; the second abnormal emotion degree u2 is obtained by the respiratory heart rate change factor x3 and the voice tone change factor x4, and u2∈[0,1], unit %;

[0043] The fuzzy subsets of the first abnormal emotion level u1 and the second abnormal emotion level u2 both include normal emotion VNE, relatively normal emotion NE, abnormal emotion SAE, relatively abnormal emotion AE and very abnormal emotion VAE; the triangular membership function is:

[0044] VN i =1-5u i ,u i ∈[0,0.2]

[0045]

[0046] i=1,2

[0047] Step c: Establish fuzzy rules

[0048] Fuzzy rule list

[0049]

[0050] Each fuzzy control rule has an implication relation R i, where i=1,2,...,9, the total implication relation of the system is R=R1∪R2∪...∪R9;

[0051] Step d: Calculate the first emotion abnormality level u1 and the second emotion abnormality level u2 according to the Mamdani algorithm

[0052]

[0053] in, For input x j , j=1,2,3,4 corresponding to the three membership functions SSCD, MSCD and LSCD, the mapping of the membership function expression with the largest function value on the image, R j is the membership function expression of the corresponding abnormal emotion degree obtained in the fuzzy rule distribution column, ∪ is the union, is matrix multiplication;

[0054] Step e: Calculate the dog's total emotional abnormality u

[0055] The average proportion coefficient of the first abnormal emotion level u1 group is:

[0056]

[0057] The average proportion coefficient of the second abnormal emotion level u2 group is:

[0058]

[0059] Among them, r(a,b) represents the Person correlation coefficient of set a and set b, and the calculation method is:

[0060]

[0061] The Person-Mamdani algorithm is used to calculate the total abnormal emotion level u:

[0062]

[0063] The degree of abnormality of the dog's mood is given by u.

[0064] The beneficial effects of the intelligent device and method for judging animal emotions based on multiple signals of the present invention are: combining a bone conduction microphone with skin electricity monitoring and myoelectric monitoring based on brain response, using bone conduction voice technology to collect dog barking signals, judging emotional changes by the degree of changes in breathing and heart rate, and innovatively adding skin electricity response and myoelectric response to the judgment of emotional state, and designing a corresponding factor adaptation degree algorithm to achieve the function of theoretically judging whether the emotion has changed. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a schematic diagram of the product structure of the intelligent device for judging animal emotions based on multiple signals of the present invention;

[0066] Figure 2 The technical roadmap of the present invention for intelligent devices and methods for determining animal conditions based on multiple signals;

[0067] Figure 3 It is a schematic diagram of human-computer interaction of an intelligent device for judging animal conditions based on multiple signals of the present invention;

[0068] Figure 4 It is a schematic diagram of a processing submodule of an intelligent device for judging an animal's condition based on multiple signals according to the present invention;

[0069] Figure 5 This is a schematic diagram of the signal collection points of the bone conduction microphone;

[0070] Figure 6 It is a flow chart of the comprehensive emotion judgment algorithm. DETAILED DESCRIPTION

[0071] The specific implementation modes of the present invention will be further described in detail below with reference to the accompanying drawings.

[0072] Method 1

[0073] The following is a specific implementation of the intelligent device for judging animal emotions based on multiple signals of the present invention.

[0074] The smart device based on multiple signals to judge animal emotions in this specific implementation mode has a product structure diagram as shown in FIG. Figure 1 As shown, it includes a signal acquisition and processing module 1, a signal acquisition and processing module 2, an operator module, an actuator module and a recall function module;

[0075] The signal acquisition module 1 includes a bone conduction voice submodule, a sound processing submodule, a heart rate monitoring submodule and a breathing monitoring submodule; it is used to collect animal bone vibration signals and convert them into sound signals, and judge and analyze animal emotions by monitoring heart rate and breathing;

[0076] The signal acquisition module 2 includes a skin conductance submodule and a muscle conductance submodule; the skin conductance submodule is used to measure the change of skin resistance caused by the activity of the autonomic nervous system and the change of sweat gland secretion; the muscle conductance submodule is used to measure the change of muscle potential caused by brain excitement;

[0077] The operator processing module includes a data acquisition submodule, a function operation submodule and a numerical judgment submodule; the data acquisition submodule is used to collect animal physiological index signals; the function operation submodule is used to define the fuzzy distribution of input and output quantities and establish a triangular membership function; the numerical judgment submodule is used to judge the degree of emotional abnormality;

[0078] The actuator module includes a scene acquisition module and a medical processing module;

[0079] The recall function module includes a GPS positioning submodule and a recall function speaker submodule; the scene acquisition module is used to collect environmental scenes within a radius of 50 meters around the animal's location; the medical processing module is used to install an anesthesia device and determine the timing of starting anesthesia.

[0080] The installation position of the skin conductance submodule is at the 1 / 4 position close to the front between the front and the back of the animal's neck; the installation position of the muscle conductance submodule is at the 1 / 4 position close to the front between the front and the back of the animal's neck.

[0081] A microcurrent measurement circuit is installed between the skin conductivity submodule and the muscle conductivity submodule.

[0082] The bone conduction voice submodule is installed in the front part of the animal's neck to conduct the mechanical vibration of the bones; the heart rate monitoring submodule and the breathing monitoring submodule are installed at the 1 / 4 of the right side behind the animal's neck.

[0083] The scene acquisition module and the medical processing module of the device have an automatic start function. A green LED and a photosensitive element are installed inside the device as photoelectric sensors for detecting the position of animal blood vessels.

[0084] One or more remote control terminals are connected, and the remote control terminals synchronize the physiological index data of the animals.

[0085] The physiological indicator data include the dog's name, gender, owner's name, contact information, whether all three shots of vaccine have been received, whether rabies vaccine has been injected on time, medical history, and precautions.

[0086] Method 2

[0087] The following is a specific implementation of the intelligent method for judging animal emotions based on multiple signals of the present invention.

[0088] The intelligent method for judging animal emotions based on multiple signals in this specific implementation comprises the following steps:

[0089] Step a: Collect four specific physiological indicators and import them into the processor

[0090] The four specific physiological indicators are: skin electricity factor x1, myoelectric factor x2, respiratory heart rate change factor x3 and voice pitch change factor x4;

[0091] Step b: Define the fuzzy distribution of input and output quantities and establish the triangular membership function

[0092] The input is the four specific physiological indicators, and the output is the first abnormal emotion degree u1 and the second abnormal emotion degree u2;

[0093] Let the skin electrical factor x1∈[0,2], unit μs, the fuzzy subsets of the skin electrical factor x1 are the small range of skin conductance difference SSCD, the medium range of skin conductance difference MSCD and the large range of skin conductance difference LSCD; the fuzzy subsets of the myoelectric factor x2 are the small range of myoelectric conductance difference SMCD, the medium range of myoelectric conductance difference MMCD and the large range of myoelectric conductance difference LMCD;

[0094] The membership function of the skin electrical triangle is:

[0095] SSCD=-2x1+1,,x1∈[0,0.5]

[0096]

[0097] The membership function of the myoelectric triangle is:

[0098] SMCD=-2x2+1,,x2∈[0,0.5]

[0099]

[0100] The first abnormal emotion degree u1 is obtained by the skin electricity factor x1 and the myoelectric factor x2, and u1∈[0,1], unit %; the second abnormal emotion degree u2 is obtained by the respiratory heart rate change factor x3 and the voice tone change factor x4, and u2∈[0,1], unit %;

[0101] The fuzzy subsets of the first abnormal emotion level u1 and the second abnormal emotion level u2 both include normal emotion VNE, relatively normal emotion NE, abnormal emotion SAE, relatively abnormal emotion AE and very abnormal emotion VAE; the triangular membership function is:

[0102] VN i =1-5u i ,u i ∈[0,0.2]

[0103]

[0104] i=1,2

[0105] Step c: Establish fuzzy rules

[0106] Fuzzy rule list

[0107]

[0108] Each fuzzy control rule has an implication relation R i , where i=1,2,...,9, the total implication relation of the system is R=R1∪R2∪...∪R9;

[0109] Step d: Calculate the first emotion abnormality level u1 and the second emotion abnormality level u2 according to the Mamdani algorithm

[0110]

[0111] in, For input x j , j=1,2,3,4 corresponding to the three membership functions SSCD, MSCD and LSCD, the mapping of the membership function expression with the largest function value on the image, R j is the membership function expression of the corresponding abnormal emotion degree obtained in the fuzzy rule distribution column, ∪ is the union, is matrix multiplication;

[0112] Step e: Calculate the dog's total emotional abnormality u

[0113] The average proportion coefficient of the first abnormal emotion level u1 group is:

[0114]

[0115] The average proportion coefficient of the second abnormal emotion level u2 group is:

[0116]

[0117] Among them, r(a,b) represents the Person correlation coefficient of set a and set b, and the calculation method is:

[0118]

[0119] The Person-Mamdani algorithm is used to calculate the total abnormal emotion level u:

[0120]

[0121] The degree of abnormality of the dog's mood is given by u.

[0122] Method 3

[0123] The following is a specific implementation of the present invention in which an intelligent device and method for judging animal emotions based on multiple signals are combined.

[0124] The technical route of the present invention is as follows Figure 2 As shown, it is divided into three parts:

[0125] (I) Signal acquisition part

[0126] 1. Measure skin conductance

[0127] It is installed in the right front part of the dog's neck, where blood vessels are densely distributed, to measure changes in skin resistance caused by changes in autonomic nervous system activity and sweat gland secretion.

[0128] 2. Measuring muscle conductance

[0129] Installed in the right front of the dog's neck, where blood vessels are dense, it is used to measure changes in muscle potential caused by brain excitement. Because it is similar to the measurement method of skin conductance physiological indicators, a microcurrent measurement circuit can be installed to achieve the same.

[0130] 3. Measure breathing and heart rate

[0131] It is installed at the back of the neck for measurement. Its range of action is close to the brain stem, which is convenient for measuring physiological indicators such as breathing and heart rate. A green LED and a photosensitive element (photodiode) are installed inside the collar as photoelectric sensors. When the dog wears the collar, the LED will emit green light and shine on the skin near the brain stem. Part of the light will be absorbed by the skin, while the other part will be reflected back to the bracelet. The photosensitive element will detect the change in the intensity of the reflected light to determine the respiratory heart rate value, and also determine the location of the blood vessels.

[0132] 4. Measuring the pitch of a dog’s bark

[0133] Installed in front of the dog's neck, it is used to conduct tiny mechanical vibrations of the bones. This vibration can be captured by the contact sensor and converted into an electrical signal, which is then amplified, filtered, and digitized to generate an audio signal, which can collect the sound and determine the change in the tone of the dog's barking.

[0134] (II) Algorithm judgment part

[0135] The data processed by the comprehensive emotion judgment algorithm and the dog's physiological index data will be analyzed as follows: Figure 3 The data is reflected on the App side in the form of

[0136] (III) Actuator part

[0137] 1. When the algorithm determines that the dog's mood is abnormal, the front camera and bone conduction microphone will automatically turn on

[0138] The front camera is placed in front of the dog's collar to illuminate the dog's surroundings. It can also help the owner see the target of attack when the dog is in an abnormal mood, observe the situation, and determine whether it is necessary to take forced paralysis measures. The bone conduction microphone is installed in the same way as the device for measuring the dog's barking tone.

[0139] 1.1. The owner judges that his emotions are abnormal

[0140] It is installed at the back of the neck and measured close to the brain stem. When the emotional comprehensive detection system detects that the dog is in an abnormal state and the owner also requests compulsory anesthesia, the subcutaneous injection anesthesia device is activated. The anesthesia system will determine the amount of anesthesia to be injected subcutaneously based on the pet's weight previously entered by the owner, and inject subcutaneously into the blood vessel position detected by the photosensitive element, so that the emotionally abnormal dog will stop attacking and harming humans. The specific internal components of the device are as follows Figure 4 shown.

[0141] 1.2. The owner judges that there is no abnormality in his emotions

[0142] If the dog is not in an abnormal mood, it can be recalled at any time. The GPS positioning function allows the owner to find the dog's location at any time, realizing cordless management and preventing the pet from getting lost. The recall function speaker is a recall voice player set by the owner in advance, which is convenient for the owner to manage the pet cordlessly.

[0143] 2. When the algorithm determines that the dog’s emotions are normal

[0144] If there is no abnormal situation, the owner can still synchronize the dog's health and physiological indicators at any time, and can also check the dog's surroundings through the camera at any time. At the same time, the information collection QR code on the front of the dog collar can store the dog's personal information, including the dog's name, gender, owner's name, contact information, whether the three shots of vaccine have been completed, whether the rabies vaccine has been injected on time, medical history, precautions and other basic information in the cloud space under the QR code, to help understand the dog's basic situation and the owner's contact information.

[0145] Method 4

[0146] The following is a specific implementation of the present invention in which an intelligent device and method for judging animal emotions based on multiple signals are combined.

[0147] The present invention is an intelligent dog collar that can monitor the physiological indicators of dogs in real time. It uses information from four channels, namely skin conductance difference, myoelectric conductance difference, respiratory heart rate, and voice tone, to determine the degree of abnormal emotions of dogs based on an emotional comprehensive judgment algorithm, thereby solving the problem of dogs' abnormal emotions causing harm to human safety. The originality of the present invention lies in:

[0148] (I) Signal acquisition part

[0149] 1. For the first time, skin galvanic response and myoelectric response were applied to the emotional judgment of animals such as dogs.

[0150] As a basic physiological signal that is easy to obtain, skin electricity can quantify emotional reactions and has been widely used in human emotion recognition research, but its application in animals such as dogs to judge emotions is innovative. Skin electricity response is caused by changes in skin resistance caused by the activity of the autonomic nervous system when the body is stimulated by external factors or changes in emotional state, which causes changes in the dilation and contraction of blood vessels in the skin and the secretion of sweat glands.

[0151] For electromyographic signals, when the brain is excited and transmitted downward, the motor neurons of the central nervous system generate electrical pulses under the stimulation of synapses, causing a series of changes in the muscle fibers, resulting in the contraction of the muscle fibers. At the same time, the propagating electrical signals cause a current field in the dog's soft tissue, and a potential difference is shown between the detection electrodes, namely the electromyographic signal. Therefore, the degree of excitement of the brain can be reflected by monitoring the electromyographic signals, which reflects the psychological and emotional activities.

[0152] Dogs have a large number of sweat glands on their necks, and using two physiological indicators based on brain reactions to judge emotions improves the accuracy of emotional judgment, so it can be well applied to dog emotion detection.

[0153] 2. The first bone conduction microphone to collect dog sounds

[0154] When there are a large number of dogs or the surrounding environment is noisy, using an air-conduction voice microphone to collect sound will result in problems such as difficulty in denoising and the sounds of multiple dogs mixed together and unable to be distinguished. Using air-conduction voice can solve this problem well.

[0155] Bone conduction microphones are different from traditional air conduction. The sound source of bone conduction microphones is not collected from sound sensors, but from the vibration of the skull through highly sensitive vibration sensors, and then converted into audio signals. It shields noise from the sound source and prevents the transmission of environmental noise at both ends of the communication system from the source. Therefore, even in strong noise environments, useful signals can be transmitted clearly. The signal collection points are as follows Figure 5 shown.

[0156] (II) Algorithm judgment part

[0157] For the comprehensive emotion judgment algorithm, this device proposes a fuzzy processing algorithm based on correlation coefficient optimization.

[0158] Different from the multi-factor fuzzy processing method, this algorithm divides algorithms with similar correlation coefficients into one group, calculates multiple groups of emotional abnormality data after fuzzy processing, and combines them with the proportion of correlation coefficients to obtain a more accurate emotional abnormality value.

[0159] Through the early simulation experiments, it can be seen that the Pearson correlation coefficient values ​​of the changes in the conductivity of the skin electrical signal and the electromyographic signal are similar to the abnormal degree of emotion, and the changes in the breathing heart rate and voice tone are similar to the abnormal degree of emotion. Therefore, when making emotional judgments, the four factors are divided into groups of two, and the abnormal degree of emotion obtained after fuzzy processing is calculated respectively. The flow chart is as follows Figure 6 shown.

[0160] Step a: Collect four specific physiological indicators and import them into the processor

[0161] The four specific physiological indicators are: skin electricity factor x1, myoelectric factor x2, respiratory heart rate change factor x3 and voice pitch change factor x4;

[0162] Step b: Define the fuzzy distribution of input and output quantities and establish the triangular membership function

[0163] The input is the four specific physiological indicators, and the output is the first abnormal emotion degree u1 and the second abnormal emotion degree u2;

[0164] Let the skin electrical factor x1∈[0,2], unit μs, the fuzzy subsets of the skin electrical factor x1 are the small skin conductance difference SSCD (Small Skin Conductance Difference), the middle skin conductance difference MSCD (Middle Skin Conductance Difference) and the large skin conductance difference LSCD (Large Skin Conductance Difference); the fuzzy subsets of the myoelectric factor x2 are the small muscle conductance difference SMCD (Small Muscle Conductance Difference), the middle muscle conductance difference MMCD (Middle Muscle Conductance Difference) and the large muscle conductance difference LMCD (Large Muscle Conductance Difference);

[0165] The membership function of the skin electrical triangle is:

[0166] SSCD=-2x1+1,,x1∈[0,0.5]

[0167]

[0168] The membership function of the myoelectric triangle is:

[0169] SMCD=-2x2+1,,x2∈[0,0.5]

[0170]

[0171] The first abnormal emotion degree u1 is obtained by the skin electricity factor x1 and the myoelectric factor x2, and u1∈[0,1], unit %; the second abnormal emotion degree u2 is obtained by the respiratory heart rate change factor x3 and the voice tone change factor x4, and u2∈[0,1], unit %;

[0172] The fuzzy subsets of the first abnormal emotion level u1 and the second abnormal emotion level u2 both include very normal emotions VNE (Very Normal Emotions), normal emotions NE (Normal Emotions), some abnormal emotions SAE (Some Abnormal Emotions), some abnormal emotions AE (Abnormal Emotions) and very abnormal emotions VAE (Very Abnormal Emotions); the triangular membership function is:

[0173] VN i =1-5u i ,u i ∈[0,0.2]

[0174]

[0175] i=1,2

[0176] Step c: Establish fuzzy rules

[0177] The nine fuzzy control rules summarized from experience are listed as follows:

[0178] Fuzzy rule list

[0179]

[0180] Each fuzzy control rule has an implication relation R i , where i=1,2,...,9, the total implication relation of the system is R=R1∪R2∪...∪R9;

[0181] Step d: Calculate the first emotion abnormality level u1 and the second emotion abnormality level u2 according to the Mamdani algorithm

[0182]

[0183] in, For input x j , j=1,2,3,4 corresponding to the three membership functions SSCD, MSCD and LSCD, the mapping of the membership function expression with the largest function value on the image, R j is the membership function expression of the corresponding abnormal emotion degree obtained in the fuzzy rule distribution column, ∪ is the union, is matrix multiplication;

[0184] Step e: Calculate the dog's total emotional abnormality u

[0185] The average proportion coefficient of the first abnormal emotion level u1 group is:

[0186]

[0187] The average proportion coefficient of the second abnormal emotion level u2 group is:

[0188]

[0189] Among them, r(a,b) represents the Person correlation coefficient of set a and set b, and the calculation method is:

[0190]

[0191] The Person-Mamdani algorithm is used to calculate the total abnormal emotion level u:

[0192]

[0193] The degree of abnormality of the dog's mood is given by u.

[0194] (III) Actuator part

[0195] 1. Set up the front camera

[0196] When the comprehensive emotion judgment algorithm determines that the dog's emotions are in an abnormal state, this product will turn on the front camera and bone conduction microphone to see the dog's environment in real time. Only when the owner's App also determines that the dog is in an abnormal state at this time and has shown aggressive behavior, will the forced anesthesia device be activated. This multi-judgment actuator is not available in traditional smart dog collars. It not only ensures the absolute safety of people when facing dog attacks, but also ensures the accuracy of the forced anesthesia system to prevent unnecessary psychological harm to the dog.

[0197] 2. Anesthesia method

[0198] In traditional smart dog collars, even if the dog's mood is detected to be abnormal, there is no corresponding measure to control this problem and reduce the harm to the surrounding area. This product chooses to use subcutaneous injection to force anesthesia for dogs that attack humans.

[0199] This product fully considers various anesthesia methods such as voltage anesthesia and anesthesia mask gas anesthesia. Given that the subcutaneous fat rate and electric shock resistance of each dog are uncertain, it is not convenient to set a unified algorithm to calculate the anesthesia metering threshold for surface anesthesia such as voltage anesthesia. If the effect is too little, it will be minimal, and if the effect is too much, it will endanger the dog's life. The anesthesia mask odor anesthesia widely used in pet hospitals is a chronic anesthesia, which is not convenient for making the dog lose its offensive ability immediately. Considering the speed and destructive power of the dog's aggressive behavior, this widely used anesthesia method was not selected.

[0200] Subcutaneous anesthesia is an effective anesthesia method. After injection, the dog will immediately lose its ability to attack, and it is only related to its weight, which is convenient for calculating the injection dose of anesthesia. At the same time, by measuring the breathing and heart rate, the green LED light can accurately determine the location of the blood vessels, ensuring the accuracy of subcutaneous injection.

[0201] Finally, combined with the technical innovations, the advantages of the present invention are summarized as follows:

[0202] 1. Summary of key technical points of the present invention and corresponding beneficial effects

[0203] Beneficial effects 1. In terms of signal collection, the present invention innovatively introduces skin electrical signals and electromyographic signals to judge the degree of abnormal emotions of dogs, reflects emotional changes from brain reaction physiological index data, diversifies the monitoring methods of emotions, and improves the accuracy of judgment. At the same time, the present invention uses bone conduction voice microphones to collect sound signals, realizes the collection of dog barking signals and the judgment of tone changes in complex environments, so as to improve the stability of sound signal collection.

[0204] Beneficial effect 2. In terms of algorithm judgment, the present invention proposes a fuzzy processing algorithm based on correlation coefficient optimization, which performs fuzzy processing and calculation on two groups of data with the same correlation coefficients, namely skin electricity, electromyography conductance changes and respiratory heart rate, and sound pitch changes, and finally performs comprehensive calculation with the real-time updated correlation coefficient ratio to improve the accuracy of the degree of emotional abnormality.

[0205] Beneficial effect 3. In terms of actuators, the present invention proposes a series of practical measures when the dog's emotions are abnormal. When the algorithm determines that the dog's emotions are abnormal, the owner's observation and judgment of the environment will be realized through the front camera and bone conduction microphone, so as to achieve the purpose of comprehensive judgment of the dog's emotions and improve the accuracy of judgment. At the same time, when it is determined that the dog has the behavior of attacking humans, forced anesthesia measures are taken, that is, subcutaneous injection is performed to calm it down, reducing the harm to the surrounding environment and human health and safety.

[0206] 2. Key technical points of the present invention

[0207] Key technical point 1: In terms of signal collection, skin electrical signals and electromyography signals are used to judge the dog's emotional changes, so as to achieve the diversity and accuracy of the dog's comprehensive emotional judgment; bone conduction voice microphones are used to realize the collection of dog barking signals in complex environments, thereby improving the stability of the product in sound collection.

[0208] Key technical point 2: In terms of algorithm judgment, a fuzzy processing algorithm based on correlation coefficient optimization is proposed to achieve comprehensive judgment of dog emotions, so as to improve the accuracy of the system's emotion judgment.

[0209] Key technical point 3: In terms of actuators, a series of countermeasures are proposed to prevent dogs’ aggressive behavior and ensure the safety of human life and property; at the same time, the owner’s observation of the surrounding environment achieved on the App side also improves the accuracy of judging whether emotions are abnormal, avoids system misjudgment, and ensures the dog’s mental health.

Claims

1. An intelligent device for judging animal emotions based on multiple signals, characterized in that: It includes a signal acquisition and processing module 1, a signal acquisition and processing module 2, an operator module, an actuator module and a recall function module; The signal acquisition module 1 includes a bone conduction voice submodule, a sound processing submodule, a heart rate monitoring submodule and a breathing monitoring submodule; it is used to collect animal bone vibration signals and convert them into sound signals, and judge and analyze animal emotions by monitoring heart rate and breathing; The signal acquisition module 2 includes a skin conductance submodule and a muscle conductance submodule; the skin conductance submodule is used to measure the change of skin resistance caused by the activity of the autonomic nervous system and the change of sweat gland secretion; The muscle conductance module is used to measure changes in muscle potential caused by brain excitement; The operator processing module includes a data acquisition submodule, a function operation submodule and a numerical judgment submodule; the data acquisition submodule is used to collect animal physiological index signals; The function operation submodule is used to define the fuzzy distribution of input and output quantities and establish a triangular membership function; the numerical judgment submodule is used to judge the degree of abnormal emotion; The actuator module includes a scene acquisition module and a medical processing module; The recall function module includes a GPS positioning submodule and a recall function speaker submodule; the scene acquisition module is used to collect environmental scenes within a radius of 50 meters around the animal's location; the medical processing module is used to install an anesthesia device and determine the timing of starting anesthesia.

2. The intelligent device for judging animal emotions based on multiple signals according to claim 1, characterized in that: The installation position of the skin conductance submodule is at the 1 / 4 position close to the front between the front and the back of the animal's neck; the installation position of the muscle conductance submodule is at the 1 / 4 position close to the front between the front and the back of the animal's neck.

3. The intelligent device for judging animal emotions based on multiple signals according to claim 1 or 2, characterized in that: A microcurrent measurement circuit is installed between the skin conductivity submodule and the muscle conductivity submodule.

4. The intelligent device for judging animal emotions based on multiple signals according to claim 1, characterized in that: The bone conduction voice submodule is installed in the front part of the animal's neck to conduct the mechanical vibration of the bones; the heart rate monitoring submodule and the breathing monitoring submodule are installed at the right 1 / 4 of the rear part of the animal's neck.

5. The intelligent device for judging animal emotions based on multiple signals according to claim 1, characterized in that: The scene acquisition module and the medical processing module of the device have an automatic start function. A green LED and a photosensitive element are installed inside the device as photoelectric sensors for detecting the position of animal blood vessels.

6. The intelligent device for judging animal emotions based on multiple signals according to claim 1, characterized in that: One or more remote control terminals are connected, and the remote control terminals synchronize the physiological index data of the animals.

7. The intelligent device for judging animal emotions based on multiple signals according to claim 6, characterized in that: The physiological indicator data include the dog's name, gender, owner's name, contact information, whether all three shots of vaccine have been received, whether rabies vaccine has been injected on time, medical history, and precautions.

8. An intelligent method for judging animal emotions based on multiple signals, characterized in that: The following steps are involved: Step a: Collect four specific physiological indicators and import them into the processor The four specific physiological indicators are: skin electricity factor x1, myoelectric factor x2, respiratory heart rate change factor x3 and voice pitch change factor x4; Step b: Define the fuzzy distribution of input and output quantities and establish the triangular membership function The input is the four specific physiological indicators, and the output is the first abnormal emotion degree u1 and the second abnormal emotion degree u2; Let the skin electrical factor x1∈[0,2], unit μs, the fuzzy subsets of the skin electrical factor x1 are the small range of skin conductance difference SSCD, the medium range of skin conductance difference MSCD and the large range of skin conductance difference LSCD; the fuzzy subsets of the myoelectric factor x2 are the small range of myoelectric conductance difference SMCD, the medium range of myoelectric conductance difference MMCD and the large range of myoelectric conductance difference LMCD; The membership function of the skin electrical triangle is: SSCD=-2x1+1,,x1∈[0,0.5] The membership function of the myoelectric triangle is: SMCD=-2x2+1,,x2∈[0,0.5] The first abnormal emotion degree u1 is obtained by the skin electricity factor x1 and the myoelectric factor x2, and u1∈[0,1], unit %; the second abnormal emotion degree u2 is obtained by the respiratory heart rate change factor x3 and the voice tone change factor x4, and u2∈[0,1], unit %; The fuzzy subsets of the first abnormal emotion level u1 and the second abnormal emotion level u2 both include normal emotion VNE, relatively normal emotion NE, abnormal emotion SAE, relatively abnormal emotion AE and very abnormal emotion VAE; the triangular membership function is: VNE i =1-5u i ,u i ∈[0,0.2] i=1,2 Step c: Establish fuzzy rules Each fuzzy control rule has an implication relation R i , where i=1,2,...,9, the total implication relation of the system is R=R1∪R2∪...∪R9; Step d: Calculate the first emotion abnormality level u1 and the second emotion abnormality level u2 according to the Mamdani algorithm in, For input x j , j=1,2,3,4 corresponding to the three membership functions SSCD, MSCD and LSCD, the mapping of the membership function expression with the largest function value on the image, R j is the membership function expression of the corresponding abnormal emotion degree obtained in the fuzzy rule distribution column, ∪ is the union, is matrix multiplication; Step e: Calculate the dog's total emotional abnormality u The average proportion coefficient of the first abnormal emotion level u1 group is: The average proportion coefficient of the second abnormal emotion level u2 group is: Among them, r(a,b) represents the Person correlation coefficient of set a and set b, and the calculation method is: The Person-Mamdani algorithm is used to calculate the total abnormal emotion level u: The degree of abnormality of the dog's mood is given by u.