Method for monitoring the health status of a pig and related device

Abstract

The application provides a pig health state monitoring method and related device, and belongs to the technical field of monitoring. The method comprises the following steps: collecting a first sound signal corresponding to a target object in a first frequency spectrum range and collecting a second sound signal corresponding to the target object in a second frequency spectrum range; performing sound spectrum analysis on the first sound signal to obtain a corresponding first frequency spectrum signal and performing sound spectrum analysis on the second sound signal to obtain a corresponding second frequency spectrum signal; analyzing the first frequency spectrum signal according to a preset first sound diagnosis model to obtain a first health state corresponding to the target object; analyzing the second frequency spectrum signal according to a second sound diagnosis model to obtain a second health state corresponding to the target object; and fusing the first health state and the second health state to obtain a target health state corresponding to the target object. The problem that the determination method for the abnormal behavior of the target object in the related art is time-consuming and laborious, the efficiency is not high, and serious economic losses are easily caused is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of monitoring, in particular to a pig health state monitoring method and related device. BACKGROUND

[0002] The breeding industry is an important industry in China's agriculture. For example, breeding pigs plays an important role in ensuring the supply of meat food safety. China's pig industry is changing from traditional pig industry to modern pig industry. Whether it is breeding mode, regional layout or production method, production capacity is changing significantly. Most of the domestic pig industry is large-scale breeding. Pig farms are equipped with several pig houses. The pigs bred by the breeder are evenly distributed in several pig houses. Each pig house is equipped with a feeding port for feeding pigs. The feeding port regularly feeds water and food to the pigs. Each pig's ear is equipped with an electronic ear tag. By identifying the electronic ear tag, the breeder can accurately identify the identity of the pig to facilitate the breeder to manage the pig.

[0003] When the pigs in the pig house are sick, the breeder will not be able to learn about the pig's illness in a timely manner, which will affect the survival rate of the pigs. In related technologies, the behavior of pigs is mainly observed and judged by manual observation method. Not only is it time-consuming and labor-intensive, but also the efficiency is not high. Moreover, due to the time and energy constraints of humans, some abnormal behavior of the breeding process is easily overlooked, resulting in serious economic losses. SUMMARY

[0004] The main purpose of the embodiments of the present application is to provide a pig health state monitoring method and related device, which aims to solve the problem of time-consuming and labor-intensive determination method of abnormal behavior of target object in related technologies, low efficiency and easy to cause serious economic losses.

[0005] In a first aspect, the embodiments of the present application provide a pig health state monitoring method, comprising:

[0006] Collecting a first sound signal corresponding to a target object in a first frequency spectrum range and collecting a second sound signal corresponding to the target object in a second frequency spectrum range;

[0007] Performing sound spectrum analysis on the first sound signal to obtain a corresponding first frequency spectrum signal and performing sound spectrum analysis on the second sound signal to obtain a corresponding second frequency spectrum signal;

[0008] Analyzing the first frequency spectrum signal according to a preset first sound diagnosis model to obtain a first health state corresponding to the target object;

[0009] Analyzing the second frequency spectrum signal according to a preset second sound diagnosis model to obtain a second health state corresponding to the target object;

[0010] fusing the first health state and the second health state to obtain a target health state corresponding to the target object.

[0011] In a second aspect, an embodiment of the present application provides a device for monitoring a health state of a pig, comprising:

[0012] a data collection module, configured to collect a first sound signal corresponding to a target object in a first frequency spectrum range and collect a second sound signal corresponding to the target object in a second frequency spectrum range;

[0013] a signal processing module, configured to perform sound spectrum analysis on the first sound signal to obtain a first frequency spectrum signal corresponding to the target object and perform sound spectrum analysis on the second sound signal to obtain a second frequency spectrum signal corresponding to the target object;

[0014] a first determination module, configured to analyze the first frequency spectrum signal according to a preset first sound diagnosis model to obtain a first health state corresponding to the target object;

[0015] a second determination module, configured to analyze the second frequency spectrum signal according to a preset second sound diagnosis model to obtain a second health state corresponding to the target object;

[0016] a target determination module, configured to fuse the first health state and the second health state to obtain a target health state corresponding to the target object.

[0017] In a third aspect, an embodiment of the present application further provides a terminal device, comprising a processor, a memory, a computer program stored in the memory and executable by the processor, and a data bus for realizing connection communication between the processor and the memory, wherein the computer program is executable by the processor to realize steps of any one of the methods for monitoring a health state of a pig provided in the specification of the present application.

[0018] In a fourth aspect, an embodiment of the present application further provides a storage medium for computer readable storage, characterized in that the storage medium stores one or more programs, and the one or more programs are executable by one or more processors to realize steps of any one of the methods for monitoring a health state of a pig provided in the specification of the present application.

[0019] The embodiment of the present application provides a pig health state monitoring method and a related device, the method comprises the following steps: collecting a first sound signal corresponding to a target object in a first frequency spectrum range and collecting a second sound signal corresponding to the target object in a second frequency spectrum range; performing sound spectrum analysis on the first sound signal to obtain a corresponding first frequency spectrum signal and performing sound spectrum analysis on the second sound signal to obtain a corresponding second frequency spectrum signal; analyzing the first frequency spectrum signal according to a preset first sound diagnosis model to obtain a first health state corresponding to the target object; analyzing the second frequency spectrum signal according to a preset second sound diagnosis model to obtain a second health state corresponding to the target object; and fusing the first health state and the second health state to obtain a target health state corresponding to the target object. The method comprehensively utilizes sound signals in different frequency spectrum ranges and sound spectrum analysis technology, and combines the preset first sound diagnosis model and the second sound diagnosis model, so that the health state of the target object is more comprehensively evaluated. Different frequency spectrum ranges reflect the physical conditions of the target object from different angles, and then through the sound spectrum analysis technology, the complex sound signal can be converted into a quantifiable frequency spectrum signal, and then the preset first sound diagnosis model and the second sound diagnosis model are used for accurate health state analysis. Therefore, by using multiple frequency spectrum signals, more dimensional information can be obtained, which helps to more comprehensively understand the health state of the target object, and finally a more reliable and comprehensive target health state is obtained by fusing the first health state and the second health state in different frequency spectrum ranges, avoiding the limitations of a single evaluation method, and improving the evaluation accuracy and comprehensiveness of the health state of the target object, providing a more reliable basis for assisting veterinarians or breeders to further manage and intervene the pigs. The problem of time-consuming, low efficiency and easy to cause serious economic loss in the related art is also solved. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without any creative labor.

[0021] Figure 1 A flowchart of a pig health state monitoring method provided by the embodiment of the present application is shown in the figure.

[0022] Figure 2 A module structure schematic diagram of a pig health state monitoring device provided by the embodiment of the present application is shown in the figure.

[0023] Figure 3 A structure schematic block diagram of a terminal device provided by the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of the present application.

[0025] The flowcharts shown in the drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor are they necessarily executed in the described order. For example, some operations / steps can be further decomposed, combined or partially merged, so that the actual execution order can be changed according to actual situations.

[0026] It should be understood that the terms used in the present application specification are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the present application specification and the appended claims, the singular forms "a", "an" and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0027] Embodiments of the present application provide a pig health state monitoring method and related device. The pig health state monitoring method can be applied to a terminal device, which can be an electronic device such as a tablet computer, a notebook computer, a desktop computer, a personal digital assistant, and a wearable device. The terminal device can be a server or a server cluster.

[0028] Some embodiments of the present application will be described in detail below with reference to the accompanying drawings. In the case of no conflict, the following embodiments and features in the embodiments can be combined with each other.

[0029] Please refer to Figure 1 , Figure 1 A pig health state monitoring method provided by an embodiment of the present application.

[0030] As Figure 1 shown, the pig health state monitoring method includes steps S101 to S105.

[0031] Step S101, collecting a first sound signal corresponding to a target object in a first frequency spectrum range and collecting a second sound signal corresponding to the target object in a second frequency spectrum range.

[0032] Exemplarily, the target object can be any kind of artificially bred or human-protected animal, such as a pig.

[0033] It should be noted that the specific type of the target object is not limited in the present application, and can be selected as needed.

[0034] Exemplarily, the first frequency spectrum range is a frequency spectrum range that the target object emits and that the outside world can directly hear, and the second frequency spectrum range is a frequency spectrum range that the target object emits and that the outside world needs to use tools to hear or detect.

[0035] Exemplarily, a high-quality microphone is used to collect the sound signals of the target object in the first frequency spectrum range and the second frequency spectrum range, and then the sound interface or audio interface corresponding to the microphone is connected to a computer or other collection equipment. Thus, the collected sound signals are imported into a spectrum analysis software for spectrum analysis, so as to obtain the first sound signal in the first frequency spectrum range and the second sound signal in the second frequency spectrum range.

[0036] In step S102, the first sound signal is subjected to sound spectrum analysis to obtain a corresponding first frequency spectrum signal, and the second sound signal is subjected to sound spectrum analysis to obtain a corresponding second frequency spectrum signal.

[0037] Exemplarily, the types and characteristics of the collected first sound signal and second sound signal are determined to select a sound spectrum analysis method. Common sound spectrum analysis methods include Fourier transform, short-time Fourier transform (STFT), wavelet transform, etc. Thus, the selected sound spectrum analysis method is applied to the collected first sound signal and second sound signal respectively, to obtain the first frequency spectrum signal corresponding to the first sound signal and the second frequency spectrum signal corresponding to the second sound signal.

[0038] For example, for Fourier transform or short-time Fourier transform (STFT), a corresponding library in existing audio processing software or programming language can be used to realize it. For example, the SciPy library in Python or the signal processing toolbox in MATLAB is used to perform sound spectrum analysis on the first sound signal to obtain the first frequency spectrum signal, and to perform sound spectrum analysis on the second sound signal to obtain the second frequency spectrum signal.

[0039] In step S103, the first frequency spectrum signal is analyzed according to a preset first sound diagnosis model to obtain a first health state corresponding to the target object.

[0040] Exemplarily, before the first frequency spectrum signal is input into the first sound diagnosis model, some pretreatment is performed on the first frequency spectrum signal, such as normalization, denoising or feature extraction. This helps to improve the performance and robustness of the model. Then, the pretreated first frequency spectrum signal is input into the preset first sound diagnosis model for classification analysis to obtain the first health state of the target object. The first sound diagnosis model can be a machine learning or deep learning model used to classify signals to infer the health state of the target object.

[0041] Exemplarily, after the preprocessed first frequency spectrum signal is classified by using the first sound diagnosis model, the abnormal probability value and the normal probability value of the first frequency spectrum signal corresponding to the target object are obtained, so as to determine the first health state of the target object according to the abnormal probability value and the normal probability value.

[0042] For example, when the abnormal probability value is greater than or equal to the normal probability value, it is determined that the first health state of the target object is abnormal; when the abnormal probability value is less than the normal probability value, it is determined that the first health state of the target object is normal.

[0043] In some embodiments, the first sound diagnosis model comprises a scaling coefficient calculation layer, an offset determination layer, a signal filtering layer, a distance calculation layer, and a condition determination layer. The first health state of the target object is determined by analyzing the first frequency spectrum signal according to the preset first sound diagnosis model, which comprises: determining a target frequency spectrum signal, which is used to represent a standard signal of the target object in a corresponding health state; calculating a first scaling coefficient corresponding to the first frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model, and calculating a second scaling coefficient corresponding to the target frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model; calculating a ratio of the first scaling coefficient and the second scaling coefficient according to the offset layer of the first sound diagnosis model, to obtain an offset coefficient corresponding to the first frequency spectrum signal; determining a target upper filter limit and a target lower filter limit corresponding to a filter according to the signal filtering layer of the first sound diagnosis model, and performing filter processing on the first frequency spectrum signal according to the target upper filter limit and the target lower filter limit, to obtain a third frequency spectrum signal corresponding to the first frequency spectrum signal; calculating distance information between the third frequency spectrum signal and the target frequency spectrum signal according to the distance calculation layer of the first sound diagnosis model, to obtain a distance result; and comparing the distance result according to the condition determination layer of the first sound diagnosis model, to determine the first health state of the target object.

[0044] Exemplarily, the target spectrum signal is a standard signal of the target object in a corresponding health state. For example, when the health state is normal, the target spectrum signal is a standard signal of the target object in the first spectrum range representing the health state as normal; when the health state is a first abnormality, the target spectrum signal is a standard signal of the target object in the first spectrum range representing the health state as the first abnormality; when the health state is a second abnormality, the target spectrum signal is a standard signal of the target object in the first spectrum range representing the health state as the second abnormality, and so on. The health state as the first abnormality, the second abnormality, and so on, represents that the target object is in different sick or abnormal states, that is, the signals emitted by the target object under different disease conditions are different, thereby indicating that the standard signals under the health state are also different.

[0045] Exemplarily, the target spectrum signal corresponding to different health states of the target object is collected according to historical experience or historical database. For example, the health state includes health, sub-health, and disease, that is, the first abnormality is sub-health, and the second abnormality is disease. Therefore, for different health states, the corresponding sound signals are collected, and the target spectrum signals under the corresponding health states are obtained by extracting the sound signals. In addition, the collected signals can also be subjected to spectrum analysis (such as Fourier transform, short-time Fourier transform, etc.) to obtain the target spectrum signals corresponding to each health state.

[0046] Exemplarily, the first sound diagnosis model includes a scaling coefficient calculation layer, an offset determination layer, a signal filtering layer, a distance calculation layer, and a condition determination layer. The purpose of the scaling coefficient calculation layer is to calculate the scaling coefficient of the given spectrum signal, thereby providing support for the subsequent alignment of the first spectrum signal and the target spectrum signal.

[0047] Exemplarily, the first spectrum signal and the target spectrum signal are respectively input into the scaling coefficient calculation layer in the first sound diagnosis model, thereby obtaining the first scaling coefficient corresponding to the first spectrum signal and the second scaling coefficient corresponding to the target spectrum signal.

[0048] Exemplarily, the offset layer is used to calculate the offset coefficient between the first spectrum signal and the target spectrum signal, so as to realize the alignment between the first spectrum signal and the target spectrum signal.

[0049] Exemplarily, the first scaling coefficient and the second scaling coefficient are input into the offset layer of the first sound diagnosis model, thereby dividing the first scaling coefficient by the second scaling coefficient, thereby obtaining the offset coefficient.

[0050] Exemplarily, the target of the signal filtering layer is to determine the upper limit and the lower limit of the filter according to the offset coefficient, thereby filtering the first spectrum signal, and providing support for the subsequent calculation of the distance between the first spectrum signal and the target spectrum signal.

[0051] Exemplarily, the upper limit and the lower limit of the filter are adjusted by the offset coefficient to obtain a filter upper limit and a filter lower limit, and then the first frequency spectrum signal is filtered according to the filter upper limit and the filter lower limit to generate a third frequency spectrum signal.

[0052] Exemplarily, the distance calculation layer aims to compare the distance between the first frequency spectrum signal and the target frequency spectrum signal to determine their similarity.

[0053] Exemplarily, the distance calculation layer using the first sound diagnosis model calculates the distance between the third frequency spectrum signal and the target frequency spectrum signal to obtain a distance result. Different distance measurement methods can be used, such as Euclidean distance, Manhattan distance, cosine similarity, etc.

[0054] Exemplarily, the condition determination layer determines the first health status of the target object by comparing the distance result. For example, the target frequency spectrum signal includes multiple frequency spectrum signals, and the distance result is obtained by calculating the distance between each of the multiple frequency spectrum signals and the first frequency spectrum signal, and then the health status represented by the target frequency spectrum signal corresponding to the minimum distance result is determined as the first health status of the target object.

[0055] Exemplarily, the smaller the distance result is, the closer the first frequency spectrum signal is to the standard signal, indicating that the first health status is closer to the health status corresponding to the standard signal.

[0056] Specifically, by determining the target frequency spectrum signal, a standardized sound signal model can be established to represent the health status. This standard signal provides a comparable benchmark, making the evaluation of health status more objective and reliable. The scaling coefficient calculation layer can calculate the scaling coefficient according to the difference between the target frequency spectrum signal and the actually collected first frequency spectrum signal. These coefficients provide the proportional relationship between the actual signal and the standard signal, thereby supporting the alignment of the target frequency spectrum signal and the first frequency spectrum signal, ensuring the accuracy of the subsequent distance result calculation. The filter processing determines the upper limit and the lower limit of the filter according to the offset coefficient to adjust the actual signal. This adjustment helps to eliminate noise or abnormal frequencies in the actual signal, making the evaluation more accurate. Then, by calculating the distance result between the third frequency spectrum signal and the target frequency spectrum signal, the first health status of the target object can be accurately determined.

[0057] In some embodiments, the scaling coefficient calculation on the first frequency spectrum signal according to the preset scaling coefficient calculation layer of the first sound diagnosis model obtains a first scaling coefficient corresponding to the first frequency spectrum signal, and the scaling coefficient calculation on the target frequency spectrum signal according to the preset scaling coefficient calculation layer of the first sound diagnosis model obtains a second scaling coefficient corresponding to the target frequency spectrum signal, including: obtaining a first amplitude value corresponding to the first frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model, and determining a first probability value of a first frequency spectrum feature corresponding to the first frequency spectrum signal according to the first amplitude value; obtaining a first frequency value corresponding to the first frequency spectrum feature according to the scaling coefficient calculation layer of the first sound diagnosis model; determining the first scaling coefficient corresponding to the first frequency spectrum signal according to the first probability value and the first frequency value by using the scaling coefficient calculation layer of the first sound diagnosis model; obtaining a second amplitude value corresponding to the target frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model, and determining a second probability value of a second frequency spectrum feature corresponding to the target frequency spectrum signal according to the second amplitude value; obtaining a second frequency value corresponding to the second frequency spectrum feature according to the scaling coefficient calculation layer of the first sound diagnosis model; determining the second scaling coefficient corresponding to the target frequency spectrum signal according to the second probability value and the second frequency value by using the scaling coefficient calculation layer of the first sound diagnosis model; wherein the first scaling coefficient is calculated according to the following formula:

[0058]

[0059]

[0060] Scale1 represents the first scaling coefficient, f 1t represents the first frequency value corresponding to the t-th characteristic value in the first frequency spectrum feature, p 1t represents the first probability value corresponding to the t-th characteristic value in the first frequency spectrum feature, A(1t) represents the first amplitude value corresponding to the t-th characteristic value in the first frequency spectrum feature, and T represents the total number of characteristics corresponding to the first frequency spectrum feature in the first frequency spectrum signal.

[0061] For example, the first amplitude value corresponding to the first frequency spectrum signal is obtained according to the scaling coefficient calculation layer of the first sound diagnosis model, so that the first probability value of the first frequency spectrum feature corresponding to the first frequency spectrum signal is determined according to the first amplitude value, which can be obtained according to the following formula:

[0062]

[0063] wherein p 1tLet A(1t) represent the first probability value corresponding to the t-th dimension feature value in the first spectral feature, let A(1t) represent the first amplitude value corresponding to the t-th dimension feature value in the first spectral feature, and let T represent the total number of features corresponding to the first spectral feature in the first spectral signal.

[0064] For example, the scaling factor calculation layer of the first sound diagnostic model statistically analyzes the first spectral feature to obtain the first frequency value corresponding to the first spectral feature. Then, the scaling factor calculation layer of the first sound diagnostic model determines the first scaling factor corresponding to the first spectral signal according to the following formula using the first probability value and the first frequency value:

[0065]

[0066] Where Scale1 represents the first scaling factor, f 1t Let T represent the first frequency value corresponding to the t-th dimension feature value in the first spectral feature, and let T represent the total number of features corresponding to the first spectral feature in the first spectral signal.

[0067] For example, the second amplitude value corresponding to the target spectral signal is obtained by calculating the scaling factor layer of the first sound diagnostic model, and then the second probability value of the second spectral feature corresponding to the target spectral signal is determined based on the second amplitude value, which can be obtained according to the following formula:

[0068]

[0069] p 2h A(2h) represents the second probability value corresponding to the h-th dimension feature value in the second spectral feature, A(2h) represents the second amplitude value corresponding to the h-th dimension feature value in the second spectral feature, and H represents the total number of features corresponding to the second spectral feature in the second spectral signal.

[0070] For example, the scaling factor calculation layer of the first sound diagnostic model is used to statistically analyze the second spectral features to obtain the second frequency value corresponding to the second spectral features. Then, according to the scaling factor calculation layer of the first sound diagnostic model, the second scaling factor corresponding to the second spectral signal is determined using the second probability value and the second frequency value according to the following formula:

[0071]

[0072] Where Scale2 represents the second scaling factor, f 2h p represents the second frequency value corresponding to the h-th dimension feature value in the second spectral feature. 2h H represents the second probability value corresponding to the h-th dimension feature value in the second spectral feature, and H represents the total number of features corresponding to the second spectral feature in the second spectral signal.

[0073] Specifically, by calculating the amplitude value, frequency value and probability value of the first spectrum signal and the target spectrum signal, the first scaling coefficient corresponding to the first spectrum signal and the second scaling coefficient corresponding to the target spectrum signal are obtained, thereby providing good support for subsequent alignment and distance calculation of the first spectrum signal and the target spectrum signal, and providing more reliable and accurate basis for health status evaluation of the target object.

[0074] In some embodiments, the signal filtering layer according to the first sound diagnosis model determines the target filter upper limit and the target filter lower limit corresponding to the filter by using the offset coefficient, and filters the first spectrum signal according to the target filter upper limit and the target filter lower limit to obtain a third spectrum signal corresponding to the first spectrum signal, including: obtaining an initial filter upper limit and an initial filter lower limit corresponding to the filter; multiplying the offset coefficient and the initial filter upper limit to obtain the target filter upper limit and multiplying the offset coefficient and the initial filter lower limit to obtain the target filter lower limit according to the signal filtering layer of the first sound diagnosis model; filtering the first probability value of the first spectrum feature corresponding to the first spectrum signal by using the target filter upper limit and the target filter lower limit according to the signal filtering layer of the first sound diagnosis model to obtain the third spectrum signal corresponding to the first spectrum signal.

[0075] Exemplarily, the initial filter upper limit and the initial filter lower limit can be numerical values determined according to experience or prior knowledge, or can be dynamically adjusted according to actual conditions. These values generally reflect the spectrum range of the target spectrum signal.

[0076] Exemplarily, the target filter upper limit is obtained by multiplying the offset coefficient and the initial filter upper limit by using the signal filtering layer. Similarly, the target filter lower limit is obtained by multiplying the offset coefficient and the initial filter lower limit. Thus, the filter is determined according to the target filter upper limit and the target filter lower limit. Thus, by using the signal filtering layer, the first probability value corresponding to the first spectrum signal is filtered by using the target filter upper limit and the target filter lower limit to obtain the third spectrum signal. The offset coefficient generally reflects the offset degree of the first spectrum signal relative to the target spectrum signal, and the initial filter upper limit and the initial filter lower limit are used to determine the range of filtering, so as to retain the key features of the first spectrum signal in the filtering process.

[0077] Specifically, filtering the first probability value corresponding to the first spectrum signal according to the filter can reduce the interference of the amplitude value, so as to obtain the third spectrum signal which is adjusted and optimized, which helps to extract the key features of the first spectrum signal and reduce the influence of noise. Thus, good support is provided for obtaining accurate first health status subsequently.

[0078] Step S104: analyzing the second frequency spectrum signal according to a preset second sound diagnosis model to obtain a second health state corresponding to the target object.

[0079] Exemplarily, the second sound diagnosis model is a deep learning model or a machine learning model for performing anomaly classification on the second frequency spectrum signal corresponding to the second frequency spectrum range. The second sound diagnosis model analyzes the input second frequency spectrum signal, and calculates an index or a feature related to the health state by using an algorithm and parameters inside the model.

[0080] Exemplarily, the second health state corresponding to the target object is obtained according to a result output by the second sound diagnosis model. The second health state can be a classification label representing different health states, or a numerical value representing a health degree or a risk degree.

[0081] In some embodiments, the second sound diagnosis model comprises a neighboring difference calculation layer, a weight calculation layer, a deviation calculation layer, and a result judgment layer. The second health state corresponding to the target object is obtained by: obtaining, according to the neighboring difference calculation layer of the second sound diagnosis model, a second frequency spectrum feature corresponding to a current period and a third frequency spectrum feature corresponding to a neighboring period of the current period, and determining a fuzzy difference value corresponding to the current period according to the second frequency spectrum feature and the third frequency spectrum feature; obtaining, according to the weight calculation layer of the second sound diagnosis model, a target weight corresponding to the current period by using the second frequency spectrum feature; performing, according to the deviation calculation layer of the second sound diagnosis model, deviation degree calculation on the fuzzy difference value and the target weight to obtain a deviation value corresponding to the current period; and performing, according to the result judgment layer of the second sound diagnosis model, comparison on the deviation value to obtain the second health state corresponding to the target object.

[0082] Exemplarily, the second frequency spectrum feature corresponding to the current period and the third frequency spectrum feature corresponding to the neighboring period of the current period are calculated by using the neighboring difference calculation layer of the second sound diagnosis model, and then the absolute value of the difference between the first frequency spectrum feature and the third frequency spectrum feature is calculated, and then the ratio between the absolute value of the difference and the second frequency spectrum feature is determined as the fuzzy difference value corresponding to the current period.

[0083] Exemplarily, the target weight corresponding to the second frequency spectrum feature of the current period is determined by using the weight calculation layer of the second sound diagnosis model. Then, the deviation degree calculation is performed on the fuzzy difference value and the target weight by using the deviation calculation layer of the second sound diagnosis model to obtain the deviation value corresponding to the current period.

[0084] For example, the second frequency spectrum signal contains 5 cycles, which are the first cycle, the second cycle, the third cycle, the fourth cycle, and the fifth cycle. When the first cycle is the current cycle, the first deviation value between the first cycle and the second cycle is calculated. When the second cycle is the current cycle, the second deviation value between the second cycle and the third cycle is calculated. Similarly, the third deviation value and the fourth deviation value are obtained.

[0085] Exemplarily, the result determination layer using the second sound diagnosis model determines the maximum value of the deviation values or the sum of all deviation values, and compares the maximum value of the deviation values or the sum of all deviation values with a preset threshold. When the maximum value of the deviation values or the sum of all deviation values is less than or equal to the preset threshold, it is determined that the second health status of the target object is normal. When the maximum value of the deviation values or the sum of all deviation values is greater than the preset threshold, it is determined that the second health status of the target object is abnormal.

[0086] For example, the second frequency spectrum signal contains 5 cycles, which are the first cycle, the second cycle, the third cycle, the fourth cycle, and the fifth cycle. Then the maximum value among the first deviation value, the second deviation value, the third deviation value, and the fourth deviation value is obtained, and then the maximum value is compared with a preset threshold to obtain the second health status of the target object. Or the first deviation value, the second deviation value, the third deviation value, and the fourth deviation value are summed, and then the sum is compared with a preset threshold to obtain the second health status of the target object.

[0087] Specifically, the adjacent difference value calculation layer using the second sound diagnosis model can calculate the changes between adjacent signals, which helps to more comprehensively understand the health status of the target object. According to the second frequency spectrum features of the current cycle, combined with the weight calculation layer of the second sound diagnosis model, the target weight can be dynamically calculated. Thus, the importance of different features can be adjusted according to the feature situation of the current cycle, so as to more accurately evaluate the health status. Through the deviation calculation layer, the deviation degree of the fuzzy difference value and the target weight is calculated, which can more comprehensively consider the deviation between the current cycle features and the expected situation. This helps to timely discover abnormal situations of the health status and perform corresponding processing and intervention. Thus, through the comparison of the deviation values by the result determination layer, the health status of the target object can be objectively evaluated. Such a method can reduce the influence of subjective factors and improve the objectivity and accuracy of health status evaluation.

[0088] In some embodiments, the deviation calculation layer of the second sound diagnosis model performs deviation degree calculation on the fuzzy difference value and the target weight to obtain a deviation value corresponding to the current period, including: performing cumulative summation on the fuzzy difference value according to the deviation calculation layer of the second sound diagnosis model to obtain a summation result; performing average calculation on the summation result according to the deviation calculation layer of the second sound diagnosis model to obtain a mean value result; and performing calculation on the mean value result and the target weight according to the deviation calculation layer of the second sound diagnosis model to obtain the deviation value corresponding to the current period; wherein the deviation value is obtained according to the following formula:

[0089]

[0090] D r denotes the deviation value corresponding to the current period, w r denotes the target weight, n denotes a total number of features corresponding to the second frequency spectrum feature, V c (r) denotes the fuzzy difference value between the cth feature in the second frequency spectrum feature and the cth feature in the third frequency spectrum feature.

[0091] Exemplarily, the deviation calculation layer of the second sound diagnosis model is used to perform cumulative summation on the fuzzy difference value to obtain a summation result. In each period, the fuzzy difference value is added to the summation result of the previous period. Then, the deviation calculation layer is further used to perform average calculation on the summation result to obtain a mean value result. This can be achieved by dividing the summation result by the number of periods.

[0092] Exemplarily, the deviation calculation layer is used to calculate the deviation value according to the following formula by combining the mean value result and the target weight:

[0093]

[0094] D r denotes the deviation value corresponding to the current period, w r denotes the target weight, n denotes a total number of features corresponding to the second frequency spectrum feature, V c (r) denotes the fuzzy difference value between the cth feature in the second frequency spectrum feature and the cth feature in the third frequency spectrum feature.

[0095] Exemplarily, the second frequency spectrum feature is discretized and the third frequency spectrum feature is discretized, so that the discretized second frequency spectrum feature and the discretized third frequency spectrum feature are one-to-one corresponding, the one-to-one corresponding result is calculated to obtain the fuzzy difference value, and the deviation value corresponding to the current period is obtained according to the above formula.

[0096] Specifically, by accumulating and summing the fuzzy difference values, the deviation in multiple periods can be considered comprehensively, so as to obtain more stable evaluation results. This helps to reduce the evaluation error caused by periodic fluctuations or temporary abnormal situations. Taking the average can help identify the trend of the deviation, such as a gradually increasing or decreasing deviation trend. By monitoring the changes in the average result, potential changes in the health status can be discovered in a timely manner, and appropriate measures can be taken. Combined with the calculation of the average result and the target weight, the degree of attention to the deviation can be adjusted more flexibly. If certain features have a greater impact on the health status, the impact can be more accurately reflected by adjusting the target weight, thereby improving the accuracy of the evaluation. By taking the average of the sum result, the impact of periodic fluctuations on the evaluation result can be reduced. This helps to ensure that the evaluation result is more stable and reliable, and is not disturbed by periodic changes. Thus, by accumulating and summing and taking the average, the deviation in the historical period can be considered comprehensively, so as to more comprehensively evaluate the health status of the target object. This helps to improve the comprehensiveness and accuracy of the evaluation, thereby improving the stability, accuracy of the evaluation, and better reflecting the actual change trend of the second health status of the target object, thereby providing more effective support for health management and intervention.

[0097] Step S105, fusing the first health status and the second health status to obtain the target health status corresponding to the target object.

[0098] For example, a weighted average or similar method is used, and the weight of each health status in the fusion needs to be determined. This can be determined based on the importance, reliability or other factors of the spectrum signal collected in the frequency spectrum range. Then, according to the defined fusion rule and weight, the first health status and the second health status are fused. This can involve simple mathematical operations such as weighted average, or more complex logical operations. Thus, the target health status of the target object is comprehensively evaluated, and the information of the first health status and the second health status is fused together, thereby providing a more comprehensive and accurate health status judgment.

[0099] In some embodiments, the fusion of the first health status and the second health status to obtain the target health status corresponding to the target object comprises: determining a first abnormal probability corresponding to a voice abnormality of the target object according to the first health status, and determining a second abnormal probability corresponding to the voice abnormality of the target object according to the second health status; fusing the first abnormal probability and the second abnormal probability to obtain a target abnormal probability corresponding to the voice abnormality of the target object; and determining the target health status corresponding to the target object according to the target abnormal probability.

[0100] Exemplarily, a first abnormal probability corresponding to the sound anomaly of the target object is determined by using the first health state, and a second abnormal probability corresponding to the sound anomaly of the target object is determined by using the second health state. The first abnormal probability and the second abnormal probability are fused by using the evidence theory to obtain a target abnormal probability corresponding to the sound anomaly of the target object.

[0101] Exemplarily, after obtaining the fused target abnormal probability, the target abnormal probability is compared with a preset probability. When the target abnormal probability is greater than or equal to the preset probability, it is determined that the target health state corresponding to the target object is abnormal; and when the target abnormal probability is less than the preset probability, it is determined that the target health state corresponding to the target object is normal.

[0102] Specifically, by determining different probabilities of sound anomalies according to the first health state and the second health state, the influence of the health state of the target object on the sound anomaly in different frequency spectrum ranges can be comprehensively considered. Thus, by fusing the first abnormal probability and the second abnormal probability, the misdiagnosis caused by a single health state can be reduced. Considering the abnormal probabilities in multiple health states and fusing them can enhance the adaptability of the system to different environments, individual differences and other factors, and improve the robustness and stability of abnormal detection. Thus, the target health state of the target object can be accurately evaluated, which assists veterinarians or breeders to better guide the formulation and adjustment of treatment plans for pigs, thereby providing better auxiliary diagnosis effect.

[0103] Referring to Figure 2 , Figure 2 A pig health state monitoring device 200 is provided for the embodiments of the present application. The pig health state monitoring device 200 comprises a data acquisition module 201, a signal processing module 202, a first determination module 203, a second determination module 204, and a target determination module 205. The data acquisition module 201 is configured to acquire a first sound signal corresponding to a target object in a first frequency spectrum range and a second sound signal corresponding to the target object in a second frequency spectrum range. The signal processing module 202 is configured to perform sound spectrum analysis on the first sound signal to obtain a corresponding first frequency spectrum signal and perform sound spectrum analysis on the second sound signal to obtain a corresponding second frequency spectrum signal. The first determination module 203 is configured to analyze the first frequency spectrum signal according to a preset first sound diagnosis model to obtain a first health state corresponding to the target object. The second determination module 204 is configured to analyze the second frequency spectrum signal according to a preset second sound diagnosis model to obtain a second health state corresponding to the target object. The target determination module 205 is configured to fuse the first health state and the second health state to obtain a target health state corresponding to the target object.

[0104] In some embodiments, the first sound diagnosis model comprises a scaling coefficient calculation layer, an offset determination layer, a signal filtering layer, a distance calculation layer, a condition determination layer, and the first determination module 203 performs the following in the process of analyzing the first frequency spectrum signal according to the preset first sound diagnosis model to obtain the first health state corresponding to the target object:

[0105] determining a target frequency spectrum signal, the target frequency spectrum signal being used to represent a standard signal of the target object in a corresponding health state;

[0106] performing scaling coefficient calculation on the first frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model to obtain a first scaling coefficient corresponding to the first frequency spectrum signal, and performing scaling coefficient calculation on the target frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model to obtain a second scaling coefficient corresponding to the target frequency spectrum signal;

[0107] performing ratio calculation on the first scaling coefficient and the second scaling coefficient according to the offset layer of the first sound diagnosis model to obtain an offset coefficient corresponding to the first frequency spectrum signal;

[0108] determining a target upper filter limit and a target lower filter limit of a filter according to the signal filtering layer of the first sound diagnosis model using the offset coefficient, and performing filter processing on the first frequency spectrum signal according to the target upper filter limit and the target lower filter limit to obtain a third frequency spectrum signal corresponding to the first frequency spectrum signal;

[0109] calculating distance information between the third frequency spectrum signal and the target frequency spectrum signal according to the distance calculation layer of the first sound diagnosis model to obtain a distance result;

[0110] comparing the distance result according to the condition determination layer of the first sound diagnosis model to determine the first health state corresponding to the target object.

[0111] In some embodiments, the first determination module 203 performs the following in the process of performing scaling coefficient calculation on the first frequency spectrum signal according to the scaling coefficient calculation layer of the preset first sound diagnosis model to obtain a first scaling coefficient corresponding to the first frequency spectrum signal, and performing scaling coefficient calculation on the target frequency spectrum signal according to the scaling coefficient calculation layer of the preset first sound diagnosis model to obtain a second scaling coefficient corresponding to the target frequency spectrum signal:

[0112] obtaining a first amplitude value corresponding to the first frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model, and determining a first probability value of a first frequency spectrum feature corresponding to the first frequency spectrum signal according to the first amplitude value;

[0113] obtaining a first frequency value corresponding to the first spectral feature according to the scaling coefficient calculation layer of the first sound diagnosis model;

[0114] determining the first scaling coefficient corresponding to the first spectral signal according to the first probability value and the first frequency value by the scaling coefficient calculation layer of the first sound diagnosis model;

[0115] obtaining a second amplitude value corresponding to the target spectral signal according to the scaling coefficient calculation layer of the first sound diagnosis model, and determining a second probability value of a second spectral feature corresponding to the target spectral signal according to the second amplitude value;

[0116] obtaining a second frequency value corresponding to the second spectral feature according to the scaling coefficient calculation layer of the first sound diagnosis model;

[0117] determining the second scaling coefficient corresponding to the target spectral signal according to the second probability value and the second frequency value by the scaling coefficient calculation layer of the first sound diagnosis model;

[0118] wherein the first scaling coefficient is calculated according to the following formula:

[0119]

[0120]

[0121] Scale1 represents the first scaling coefficient, f 1t represents the first frequency value corresponding to the t-th feature value in the first spectral feature, p 1t represents the first probability value corresponding to the t-th feature value in the first spectral feature, A(1t) represents the first amplitude value corresponding to the t-th feature value in the first spectral feature, and T represents the total number of features corresponding to the first spectral feature in the first spectral signal.

[0122] In some embodiments, the first determination module 203 determines a target filter upper limit and a target filter lower limit corresponding to the filter according to the signal filtering layer of the first sound diagnosis model using the offset coefficient, and performs filtering processing on the first spectral signal according to the target filter upper limit and the target filter lower limit to obtain a third spectral signal corresponding to the first spectral signal, in which:

[0123] obtaining an initial filter upper limit and an initial filter lower limit corresponding to the filter;

[0124] multiplying the offset coefficient and the initial filter upper limit according to the signal filter layer of the first sound diagnosis model to obtain the target filter upper limit and multiplying the offset coefficient and the initial filter lower limit according to the signal filter layer of the first sound diagnosis model to obtain the target filter lower limit;

[0125] filtering the first probability value of the first frequency spectrum feature corresponding to the first frequency spectrum signal using the target filter upper limit and the target filter lower limit according to the signal filter layer of the first sound diagnosis model to obtain the third frequency spectrum signal corresponding to the first frequency spectrum signal.

[0126] In some embodiments, the second sound diagnosis model comprises an adjacent difference value calculation layer, a weight calculation layer, an offset calculation layer, and a result judgment layer. The second determination module 204 performs the following steps in the process of analyzing the second frequency spectrum signal according to the preset second sound diagnosis model to obtain the second health state corresponding to the target object:

[0127] obtaining a second frequency spectrum feature corresponding to a current period and a third frequency spectrum feature corresponding to a neighboring period of the current period according to the adjacent difference value calculation layer of the second sound diagnosis model, and determining a fuzzy difference value corresponding to the current period according to the second frequency spectrum feature and the third frequency spectrum feature;

[0128] obtaining a target weight corresponding to the current period using the second frequency spectrum feature according to the weight calculation layer of the second sound diagnosis model;

[0129] performing offset degree calculation on the fuzzy difference value and the target weight according to the offset calculation layer of the second sound diagnosis model to obtain an offset value corresponding to the current period;

[0130] comparing the offset value according to the result judgment layer of the second sound diagnosis model to obtain the second health state corresponding to the target object.

[0131] In some embodiments, the second determination module 204 performs the following steps in the process of performing offset degree calculation on the fuzzy difference value and the target weight according to the offset calculation layer of the second sound diagnosis model to obtain the offset value corresponding to the current period:

[0132] performing cumulative summation on the fuzzy difference value according to the offset calculation layer of the second sound diagnosis model to obtain a summation result;

[0133] obtaining a mean value result by taking an average of the summation result according to the offset calculation layer of the second sound diagnosis model;

[0134] The deviation value corresponding to the current period is obtained by calculating the mean value result and the target weight according to the deviation calculation layer of the second sound diagnosis model.

[0135] The deviation value is obtained according to the following formula:

[0136]

[0137] D r The deviation value corresponding to the current period is obtained by calculating the mean value result and the target weight according to the deviation calculation layer of the second sound diagnosis model. r The target weight is represented by n, and the total number of features corresponding to the second frequency spectrum feature is represented by V c (r) represents the fuzzy difference value between the cth feature in the second frequency spectrum feature and the cth feature in the third frequency spectrum feature.

[0138] In some embodiments, the target determination module 205 performs the following in the process of obtaining the target health state corresponding to the target object by fusing the first health state and the second health state:

[0139] determining a first abnormal probability corresponding to a sound abnormality of the target object according to the first health state and determining a second abnormal probability corresponding to a sound abnormality of the target object according to the second health state;

[0140] fusing the first abnormal probability and the second abnormal probability to obtain a target abnormal probability corresponding to a sound abnormality of the target object;

[0141] determining the target health state corresponding to the target object according to the target abnormal probability.

[0142] In some embodiments, the pig health state monitoring device 200 can be applied to a terminal device.

[0143] It should be noted that, for the convenience and brevity of description, the specific working process of the pig health state monitoring device 200 described above can refer to the corresponding process in the foregoing pig health state monitoring method embodiments, which will not be described here.

[0144] Please refer to Figure 3 , Figure 3 The structure of a terminal device provided in an embodiment of the present application is shown in the schematic block diagram.

[0145] As Figure 3 shown, the terminal device 300 includes a processor 301 and a memory 302, and the processor 301 and the memory 302 are connected through a bus 303, such as an I2C (Inter-integrated Circuit) bus.

[0146] Specifically, the processor 301 is configured to provide calculation and control capabilities to support the operation of the entire terminal device. The processor 301 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0147] Specifically, the memory 302 can be a Flash chip, a read-only memory (ROM) disk, an optical disk, a U disk or a mobile hard disk, etc.

[0148] Those skilled in the art can understand that, Figure 3 The structure shown in the figure is only a block diagram of part of the structure related to the embodiment of the present application, and does not constitute a limitation on the terminal device to which the embodiment of the present application is applied. Specifically, the server can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0149] The processor is configured to run a computer program stored in the memory, and when the computer program is executed, any one of the pig health state monitoring methods provided by the embodiments of the present application is implemented.

[0150] In an embodiment, the processor is configured to run a computer program stored in the memory, and when the computer program is executed, the following steps are implemented:

[0151] Collecting a first sound signal corresponding to the target object in a first frequency spectrum range and collecting a second sound signal corresponding to the target object in a second frequency spectrum range;

[0152] Performing sound spectrum analysis on the first sound signal to obtain a corresponding first frequency spectrum signal and performing sound spectrum analysis on the second sound signal to obtain a corresponding second frequency spectrum signal;

[0153] Analyzing the first frequency spectrum signal according to a preset first sound diagnosis model to obtain a first health state corresponding to the target object;

[0154] analyzing the second frequency spectrum signal according to a preset second sound diagnosis model to obtain a second health state corresponding to the target object;

[0155] fusing the first health state and the second health state to obtain a target health state corresponding to the target object.

[0156] In some embodiments, the first sound diagnosis model comprises a scaling coefficient calculation layer, an offset determination layer, a signal filtering layer, a distance calculation layer, and a condition determination layer. In the process of analyzing the first frequency spectrum signal according to the preset first sound diagnosis model to obtain the first health state corresponding to the target object, the processor 301 performs:

[0157] determining a target frequency spectrum signal, the target frequency spectrum signal being used to represent a standard signal of the target object in a corresponding health state;

[0158] performing scaling coefficient calculation on the first frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model to obtain a first scaling coefficient corresponding to the first frequency spectrum signal, and performing scaling coefficient calculation on the target frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model to obtain a second scaling coefficient corresponding to the target frequency spectrum signal;

[0159] performing ratio calculation on the first scaling coefficient and the second scaling coefficient according to the offset layer of the first sound diagnosis model to obtain an offset coefficient corresponding to the first frequency spectrum signal;

[0160] determining a target filter upper limit and a target filter lower limit corresponding to a filter according to the signal filtering layer of the first sound diagnosis model using the offset coefficient, and performing filtering processing on the first frequency spectrum signal according to the target filter upper limit and the target filter lower limit to obtain a third frequency spectrum signal corresponding to the first frequency spectrum signal;

[0161] calculating distance information between the third frequency spectrum signal and the target frequency spectrum signal according to the distance calculation layer of the first sound diagnosis model to obtain a distance result;

[0162] comparing the distance result according to the condition determination layer of the first sound diagnosis model to determine the first health state corresponding to the target object.

[0163] In some embodiments, the processor 301 performs the following during the process of scaling the first frequency spectrum signal according to the scaling coefficient calculation layer of the preset first sound diagnosis model to obtain the first scaling coefficient corresponding to the first frequency spectrum signal and scaling the target frequency spectrum signal according to the scaling coefficient calculation layer of the preset first sound diagnosis model to obtain the second scaling coefficient corresponding to the target frequency spectrum signal:

[0164] obtaining a first amplitude value corresponding to the first frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model, and determining a first probability value of a first frequency spectrum feature corresponding to the first frequency spectrum signal according to the first amplitude value;

[0165] obtaining a first frequency value corresponding to the first frequency spectrum feature according to the scaling coefficient calculation layer of the first sound diagnosis model;

[0166] determining the first scaling coefficient corresponding to the first frequency spectrum signal according to the first probability value and the first frequency value by the scaling coefficient calculation layer of the first sound diagnosis model;

[0167] obtaining a second amplitude value corresponding to the target frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model, and determining a second probability value of a second frequency spectrum feature corresponding to the target frequency spectrum signal according to the second amplitude value;

[0168] obtaining a second frequency value corresponding to the second frequency spectrum feature according to the scaling coefficient calculation layer of the first sound diagnosis model;

[0169] determining the second scaling coefficient corresponding to the target frequency spectrum signal according to the second probability value and the second frequency value by the scaling coefficient calculation layer of the first sound diagnosis model;

[0170] wherein the first scaling coefficient is calculated according to the following formula:

[0171]

[0172]

[0173] Scale1 represents the first scaling coefficient, f 1t represents the first frequency value corresponding to the t-th eigenvalue in the first frequency spectrum feature, p 1t represents the first probability value corresponding to the t-th eigenvalue in the first frequency spectrum feature, A(1t) represents the first amplitude value corresponding to the t-th eigenvalue in the first frequency spectrum feature, and T represents the total number of features corresponding to the first frequency spectrum feature in the first frequency spectrum signal.

[0174] In some embodiments, the processor 301 determines a target filter upper limit and a target filter lower limit corresponding to the filter according to the offset coefficient in the signal filtering layer of the first sound diagnosis model, and performs filtering processing on the first spectral signal according to the target filter upper limit and the target filter lower limit, to obtain a third spectral signal corresponding to the first spectral signal.

[0175] obtains an initial filter upper limit and an initial filter lower limit corresponding to the filter;

[0176] multiplies the offset coefficient and the initial filter upper limit to obtain the target filter upper limit and the initial filter lower limit according to the signal filtering layer of the first sound diagnosis model, and multiplies the offset coefficient and the initial filter lower limit to obtain the target filter lower limit according to the signal filtering layer of the first sound diagnosis model;

[0177] filters the first probability value of the first spectral feature corresponding to the first spectral signal according to the target filter upper limit and the target filter lower limit according to the signal filtering layer of the first sound diagnosis model, to obtain the third spectral signal corresponding to the first spectral signal.

[0178] In some embodiments, the second sound diagnosis model includes an adjacent difference value calculation layer, a weight calculation layer, a deviation calculation layer, and a result judgment layer. In the process of analyzing the second spectral signal according to the preset second sound diagnosis model to obtain a second health state corresponding to the target object, the processor 301 performs:

[0179] obtains a second spectral feature corresponding to a current period and a third spectral feature corresponding to an adjacent period of the current period according to the adjacent difference value calculation layer of the second sound diagnosis model, and determines a fuzzy difference value corresponding to the current period according to the second spectral feature and the third spectral feature;

[0180] obtains a target weight corresponding to the current period according to the second spectral feature according to the weight calculation layer of the second sound diagnosis model;

[0181] performs deviation degree calculation on the fuzzy difference value and the target weight according to the deviation calculation layer of the second sound diagnosis model, to obtain a deviation value corresponding to the current period;

[0182] compares the deviation value according to the result judgment layer of the second sound diagnosis model, to obtain the second health state corresponding to the target object.

[0183] In some embodiments, the processor 301 performs the following in the process of calculating the deviation value corresponding to the current period according to the deviation calculation layer of the second sound diagnosis model based on the fuzzy difference value and the target weight:

[0184] accumulating and summing the fuzzy difference value according to the deviation calculation layer of the second sound diagnosis model to obtain a sum result;

[0185] calculating the average value of the sum result according to the deviation calculation layer of the second sound diagnosis model to obtain an average result;

[0186] calculating the average result and the target weight according to the deviation calculation layer of the second sound diagnosis model to obtain the deviation value corresponding to the current period;

[0187] wherein the deviation value is obtained according to the following formula:

[0188]

[0189] D r denotes the deviation value corresponding to the current period, w r denotes the target weight, n denotes the total number of features corresponding to the second frequency spectrum feature, V c (r) denotes the fuzzy difference value between the cth feature in the second frequency spectrum feature and the cth feature in the third frequency spectrum feature.

[0190] In some embodiments, the processor 301 performs the following in the process of obtaining the target health state corresponding to the target object by fusing the first health state and the second health state:

[0191] determining a first abnormal probability corresponding to the sound abnormality of the target object according to the first health state and determining a second abnormal probability corresponding to the sound abnormality of the target object according to the second health state;

[0192] fusing the first abnormal probability and the second abnormal probability to obtain a target abnormal probability corresponding to the sound abnormality of the target object;

[0193] determining the target health state corresponding to the target object according to the target abnormal probability.

[0194] It should be noted that, for the convenience and brevity of description, the specific working process of the terminal device described above can refer to the corresponding process in the foregoing method embodiment of pig health state monitoring, which will not be described here.

[0195] The embodiment of the present application further provides a storage medium for computer readable storage, the storage medium storing one or more programs, which can be executed by one or more processors to implement the steps of any pig health state monitoring method provided in the specification of the embodiment of the present application.

[0196] The storage medium can be an internal storage unit of the terminal device, for example, a hard disk or a memory of the terminal device. The storage medium can also be an external storage device of the terminal device, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc.

[0197] Those skilled in the art can understand that all or some of the steps in the above disclosed method, the functions of the modules / units in the system and the device can be implemented as software, firmware, hardware or appropriate combination thereof. In the hardware embodiment, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component can have multiple functions, or one function or step can be performed by several physical components in cooperation. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software can be distributed on a computer readable medium, which can include computer storage media (or non-transitory media) and communication media (or transitory media). As known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. In addition, it is known to those skilled in the art that communication media typically includes computer readable instructions, data structures, program modules or other data in modulated data signals such as carrier waves or other transport mechanisms, and can include any information delivery medium.

[0198] It should be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items, and that the term "comprises" or "comprising" or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0199] The above-mentioned embodiment numbers of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments. The above description is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the present application, and these modifications or replacements should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method of monitoring the health status of a pig, characterized in that, The method comprises: collecting a first sound signal corresponding to a target object in a first frequency spectrum range and collecting a second sound signal corresponding to the target object in a second frequency spectrum range; performing acoustic spectrum analysis on the first sound signal to obtain a corresponding first frequency spectrum signal and performing acoustic spectrum analysis on the second sound signal to obtain a corresponding second frequency spectrum signal; analyzing the first frequency spectrum signal according to a preset first sound diagnosis model to obtain a first health state corresponding to the target object; analyzing the second frequency spectrum signal according to a preset second sound diagnosis model to obtain a second health state corresponding to the target object; fusing the first health state and the second health state to obtain a target health state corresponding to the target object; wherein the first sound diagnosis model comprises a scaling coefficient calculation layer, an offset determination layer, a signal filtering layer, a distance calculation layer, and a condition determination layer, and the analysis of the first frequency spectrum signal according to the preset first sound diagnosis model to obtain the first health state corresponding to the target object comprises: determining a target frequency spectrum signal, which is used to represent a standard signal of the target object in a corresponding health state; performing scaling coefficient calculation on the first frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model to obtain a first scaling coefficient corresponding to the first frequency spectrum signal, and performing scaling coefficient calculation on the target frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model to obtain a second scaling coefficient corresponding to the target frequency spectrum signal; performing ratio calculation on the first scaling coefficient and the second scaling coefficient according to the offset determination layer of the first sound diagnosis model to obtain an offset coefficient corresponding to the first frequency spectrum signal; determining a target filter upper limit and a target filter lower limit corresponding to a filter by using the offset coefficient according to the signal filtering layer of the first sound diagnosis model, and performing filter processing on the first frequency spectrum signal according to the target filter upper limit and the target filter lower limit to obtain a third frequency spectrum signal corresponding to the first frequency spectrum signal; calculating distance information between the third frequency spectrum signal and the target frequency spectrum signal according to the distance calculation layer of the first sound diagnosis model to obtain a distance result; comparing the distance result according to the condition determination layer of the first sound diagnosis model to determine the first health state corresponding to the target object; wherein the scaling coefficient calculation on the first frequency spectrum signal according to the preset scaling coefficient calculation layer of the first sound diagnosis model to obtain the first scaling coefficient corresponding to the first frequency spectrum signal and the scaling coefficient calculation on the target frequency spectrum signal according to the preset scaling coefficient calculation layer of the first sound diagnosis model to obtain the second scaling coefficient corresponding to the target frequency spectrum signal comprises: obtaining a first amplitude value corresponding to the first frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model, and determining a first probability value of a first frequency spectrum feature corresponding to the first frequency spectrum signal according to the first amplitude value; The scaling coefficient calculation layer according to the first sound diagnosis model obtains a first frequency value corresponding to the first spectral feature; The scaling coefficient calculation layer according to the first sound diagnosis model determines the first scaling coefficient corresponding to the first spectral signal by using the first probability value and the first frequency value; The scaling coefficient calculation layer according to the first sound diagnosis model obtains a second amplitude value corresponding to the target spectral signal, and determines a second probability value of a second spectral feature corresponding to the target spectral signal according to the second amplitude value; The scaling coefficient calculation layer according to the first sound diagnosis model obtains a second frequency value corresponding to the second spectral feature; The scaling coefficient calculation layer according to the first sound diagnosis model determines the second scaling coefficient corresponding to the target spectral signal by using the second probability value and the second frequency value; The first scaling coefficient is calculated according to the following formula: ； ； denotes the first scaling coefficient, denotes the first frequency value corresponding to the t-th eigenvalue in the first spectral feature, denotes the first probability value corresponding to the t-th eigenvalue in the first spectral feature, denotes the first amplitude value corresponding to the t-th eigenvalue in the first spectral feature, and T denotes the total number of eigenvalues corresponding to the first spectral feature in the first spectral signal.

2. The method of claim 1, wherein, The signal filtering layer according to the first sound diagnosis model determines a target filter upper limit and a target filter lower limit corresponding to a filter by using the offset coefficient, and performs filtering processing on the first spectral signal according to the target filter upper limit and the target filter lower limit to obtain a third spectral signal corresponding to the first spectral signal, including: obtaining an initial filter upper limit and an initial filter lower limit corresponding to the filter; The signal filtering layer according to the first sound diagnosis model multiplies the offset coefficient and the initial filter upper limit to obtain the target filter upper limit, and multiplies the offset coefficient and the initial filter lower limit to obtain the target filter lower limit; The signal filtering layer according to the first sound diagnosis model filters the first probability value of the first spectral feature corresponding to the first spectral signal by using the target filter upper limit and the target filter lower limit to obtain the third spectral signal corresponding to the first spectral signal.

3. The method of claim 1, wherein, The second sound diagnosis model includes an adjacent difference value calculation layer, a weight calculation layer, a deviation calculation layer, and a result judgment layer. The second health state corresponding to the target object is obtained by analyzing the second spectral signal according to the preset second sound diagnosis model, including: The adjacent difference value calculation layer according to the second sound diagnosis model obtains a second spectral feature corresponding to a current period and a third spectral feature corresponding to a neighboring period of the current period, and determines a fuzzy difference value corresponding to the current period according to the second spectral feature and the third spectral feature; The weight calculation layer according to the second sound diagnosis model obtains a target weight corresponding to the current period by using the second spectral feature; The deviation calculation layer according to the second sound diagnosis model performs deviation degree calculation on the fuzzy difference value and the target weight to obtain a deviation value corresponding to the current period; The result judgment layer according to the second sound diagnosis model compares the deviation value to obtain the second health state corresponding to the target object.

4. The method of claim 3, wherein, The deviation calculation layer according to the second sound diagnosis model performs deviation degree calculation on the fuzzy difference value and the target weight, and obtains a deviation value corresponding to the current period, including: The deviation calculation layer according to the second sound diagnosis model performs cumulative summation on the fuzzy difference value, and obtains a summation result; The deviation calculation layer according to the second sound diagnosis model performs average value calculation on the summation result, and obtains an average value result; The deviation calculation layer according to the second sound diagnosis model performs calculation on the average value result and the target weight, and obtains the deviation value corresponding to the current period; Wherein, the deviation value is obtained according to the following formula: ； represents the deviation value corresponding to the current period, represents the target weight, n represents the total number of features corresponding to the second spectral feature, represents the fuzzy difference value between the cth feature in the second spectral feature and the cth feature in the third spectral feature.

5. The method as claimed in claim 1, wherein, The fusion of the first health state and the second health state obtains a target health state corresponding to the target object, including: According to the first health state, a first abnormal probability corresponding to a sound abnormality of the target object is determined, and according to the second health state, a second abnormal probability corresponding to a sound abnormality of the target object is determined; The first abnormal probability and the second abnormal probability are fused to obtain a target abnormal probability corresponding to a sound abnormality of the target object; According to the target abnormal probability, the target health state corresponding to the target object is determined.

6. A device for monitoring the health status of a pig, characterized in that Including: The data acquisition module is used for collecting a first sound signal corresponding to a target object in a first frequency spectrum range and collecting a second sound signal corresponding to the target object in a second frequency spectrum range; The signal processing module is used for performing sound spectrum analysis on the first sound signal to obtain a corresponding first frequency spectrum signal, and performing sound spectrum analysis on the second sound signal to obtain a corresponding second frequency spectrum signal; The first determining module is configured to analyze the first frequency spectrum signal according to a preset first sound diagnosis model to obtain a first health state corresponding to the target object; wherein the first sound diagnosis model comprises a scaling coefficient calculation layer, an offset determination layer, a signal filtering layer, a distance calculation layer, and a condition determination layer; the analysis of the first frequency spectrum signal according to the preset first sound diagnosis model to obtain the first health state corresponding to the target object comprises: determining a target frequency spectrum signal, which is used to represent a standard signal of the target object in a corresponding health state; performing scaling coefficient calculation on the first frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model to obtain a first scaling coefficient corresponding to the first frequency spectrum signal, and performing scaling coefficient calculation on the target frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model to obtain a second scaling coefficient corresponding to the target frequency spectrum signal; performing ratio calculation on the first scaling coefficient and the second scaling coefficient according to the offset determination layer of the first sound diagnosis model to obtain an offset coefficient corresponding to the first frequency spectrum signal; determining a target filter upper limit and a target filter lower limit of a filter according to the offset coefficient by using the signal filtering layer of the first sound diagnosis model, and performing filter processing on the first frequency spectrum signal according to the target filter upper limit and the target filter lower limit to obtain a third frequency spectrum signal corresponding to the first frequency spectrum signal; calculating distance information between the third frequency spectrum signal and the target frequency spectrum signal according to the distance calculation layer of the first sound diagnosis model to obtain a distance result; and comparing the distance result according to the condition determination layer of the first sound diagnosis model to determine the first health state corresponding to the target object; wherein the scaling coefficient calculation on the first frequency spectrum signal according to the scaling coefficient calculation layer of the preset first sound diagnosis model to obtain the first scaling coefficient corresponding to the first frequency spectrum signal, and the scaling coefficient calculation on the target frequency spectrum signal according to the scaling coefficient calculation layer of the preset first sound diagnosis model to obtain the second scaling coefficient corresponding to the target frequency spectrum signal, comprises: obtaining a first amplitude value corresponding to the first frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model, and determining a first probability value of a first frequency spectrum feature corresponding to the first frequency spectrum signal according to the first amplitude value; obtaining a first frequency value corresponding to the first frequency spectrum feature according to the scaling coefficient calculation layer of the first sound diagnosis model; determining the first scaling coefficient corresponding to the first frequency spectrum signal according to the first probability value and the first frequency value by using the scaling coefficient calculation layer of the first sound diagnosis model; obtaining a second amplitude value corresponding to the target frequency spectrum signal according to the scaling coefficient calculation layer of the first sound diagnosis model, and determining a second probability value of a second frequency spectrum feature corresponding to the target frequency spectrum signal according to the second amplitude value.The second frequency value corresponding to the second spectral feature is obtained according to the scaling coefficient calculation layer of the first sound diagnosis model; the second scaling coefficient corresponding to the target spectral signal is determined by using the second probability value and the second frequency value according to the scaling coefficient calculation layer of the first sound diagnosis model; wherein the first scaling coefficient is calculated according to the following formula: ； ； denotes the first scaling coefficient, denotes the first frequency value corresponding to the t-th eigenvalue in the first spectral feature, denotes the first probability value corresponding to the t-th eigenvalue in the first spectral feature, denotes the first amplitude value corresponding to the t-th eigenvalue in the first spectral feature, and T denotes the total number of eigenvalues corresponding to the first spectral feature in the first spectral signal. The second determination module is used for analyzing the second frequency spectrum signal according to a preset second sound diagnosis model to obtain a second health state corresponding to the target object; The target determination module is used for fusing the first health state and the second health state to obtain a target health state corresponding to the target object.

7. A terminal device, characterized by comprising: The terminal device includes a processor, a memory; The memory is used for storing a computer program; The processor is used for executing the computer program and realizing the pig health state monitoring method in any one of claims 1 to 5 when executing the computer program.

8. A computer storage medium for computer storage, characterized in that The computer storage medium stores one or more programs, which can be executed by one or more processors to realize the steps of the pig health state monitoring method in any one of claims 1 to 5. The computer storage medium stores one or more programs, which can be executed by one or more processors to realize the steps of the pig health state monitoring method in any one of claims 1 to 5.

Images