A pet intelligent drinking water health monitoring pad and a monitoring method

Abstract

The application discloses a kind of pet intelligent drinking water health monitoring pad and monitoring method, belong to pet health management field, for solving the problem of data distortion caused by disturbance and drift aliasing in existing weighing type pet drinking water monitoring, linear formula reference deviates from individual metabolic demand, single day threshold early warning false alarm rate is high and lacks personalized intervention. Including: abnormal disturbance, trend change and steady state baseline interval are distinguished by local analysis window, combined with environmental drift compensation and water bowl state recognition to obtain single day actual drinking water total net weight data;Through metabolic weight formula, life stage coefficient, staple type coefficient and snack frequency coefficient are multiplied to calculate daily recommended drinking water reference value;Through four-stage asymmetric interval combined with continuous abnormal count to trigger early warning, output personalized recommended strategy in historical efficiency according to intervention strategy case library. In this way, the daily data obtained by the user is clean and reliable, and the early warning only responds when the abnormality is continuous, and the specific action is also attached.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of pet health management, and particularly relates to a pet intelligent drinking water health monitoring pad and a monitoring method. BACKGROUND

[0002] With the continuous growth of the number of pet-keeping families, the daily drinking water management of pets gradually transits from manual observation to automatic monitoring. At present, pet water bowls or drinking water tables with weighing functions have appeared on the market. The basic idea is to install a pressure sensor at the bottom of the water bowl, calculate the drinking water volume through weight change, and then upload it to a mobile phone application for the owner to check.

[0003] However, such a weighing type monitoring scheme has the following disadvantages in actual use:

[0004] The original signal collected by the weighing sensor is not only "pet drinking water" - the sensor zero point slowly drifts caused by the stepping pad, the stirring water bowl, and the environment temperature and humidity, which are superimposed on the same weight curve; most of the existing devices only use the simple threshold logic of "weight drop more than a fixed gram to determine as a drinking water" to make a judgment, and cannot separate the real drinking water event from the disturbance and drift, which results in the virtual high or missed recording of the daily cumulative drinking water volume, and the data credibility is not high;

[0005] The drinking water required amount is greatly different due to individual differences. The metabolic rate of a 0.5 kg kitten and a 40 kg adult dog is not in a linear multiple relationship, and the existing products generally use a linear formula of "per kilogram of body weight x fixed milliliter", without considering the correction effect of life stage, sterilization state, staple food water content, and snack frequency, which leads to the underestimation of small pets and the overestimation of large pets, and the subsequent early warning loses reasonable reference;

[0006] Most of the existing early warnings use single-day threshold comparison, and immediately push a reminder when the daily drinking water volume is low. However, the daily drinking water volume of pets naturally fluctuates, and occasional low does not constitute a health risk, and frequent false reports make users gradually ignore the reminders. After the early warning is triggered, only a general prompt is given, without giving a personalized intervention suggestion verified by history in combination with the breed, age, and deviation degree of the pet, and the user still does not know what to do after receiving the early warning.

[0007] Therefore, we propose a pet intelligent drinking water health monitoring pad and a monitoring method. SUMMARY

[0008] The purpose of the present application is to provide a pet intelligent drinking water health monitoring pad and a monitoring method to solve the problems raised in the background.

[0009] To achieve the above purpose, the present application provides the following technical scheme: a pet intelligent drinking water health monitoring pad:

[0010] The monitoring pad body is integrated with a high-precision acquisition module, a reference calculation module and an age division early warning module;

[0011] The high-precision acquisition module comprises a weighing sensor, an analog-to-digital converter, a microcontroller and a wireless communication module, the signal output end of the weighing sensor is electrically connected with the input end of the analog-to-digital converter, the data output end of the analog-to-digital converter is electrically connected with the data input end of the microcontroller, and the microcontroller communicates with the user terminal through the wireless communication module; the microcontroller divides the collected original weight data sequence into continuous local analysis windows in time sequence, distinguishes abnormal disturbance intervals, trend change intervals and steady baseline intervals based on each local analysis window, calculates the actual net weight data of a single pet drinking process in combination with environmental drift compensation, and obtains the single-day actual total net weight data of drinking through a single-day drinking accumulator;

[0012] The reference calculation module calculates the daily recommended drinking reference value according to the individual characteristics and feeding information of the target pet;

[0013] The age division early warning module generates an early warning signal based on the single-day actual total net weight data of drinking and the daily recommended drinking reference value, and pushes the early warning signal to the user terminal through the wireless communication module.

[0014] Preferably, the microcontroller calculates the local variance of the original weight data as the weight fluctuation degree in each local analysis window, and calculates the mean value of the difference values of adjacent original weight data in the local analysis window as the average weight change rate;

[0015] When the weight fluctuation degree is greater than a preset disturbance variance threshold, the local analysis window is marked as an abnormal disturbance interval;

[0016] When the weight fluctuation degree is less than or equal to the preset disturbance variance threshold, and the average weight change rate is greater than a preset steady-state rate threshold, the local analysis window is marked as a trend change interval;

[0017] When the weight fluctuation degree is less than or equal to the preset disturbance variance threshold, and the average weight change rate is less than or equal to the preset steady-state rate threshold, the local analysis window is marked as a steady baseline interval.

[0018] Preferably, the microcontroller splices the steady baseline interval, the trend change interval and the abnormal disturbance interval after smoothing and denoising into a filtered weight data sequence in time sequence, identifies the starting point and the ending point of the weight value showing a continuous downward trend in the filtered weight data sequence; when the absolute value of the weight difference between the starting point and the ending point is greater than a preset minimum drinking threshold and the time interval between the starting point and the ending point is greater than a preset minimum drinking time interval, a single pet drinking process is confirmed.

[0019] The microcontroller selects no less than preset minimum sampling points of original weight data points in a steady-state baseline interval, performs one-dimensional linear regression fitting with sampling time as the horizontal coordinate and original weight data as the vertical coordinate, and takes the slope of the fitted straight line as the environmental drift rate of the steady-state baseline interval;

[0020] The microcontroller extracts the weight average of the steady-state baseline interval before the starting point of the single pet drinking water process as the pre-drinking water baseline anchor point, extracts the weight average of the steady-state baseline interval after the end of the single pet drinking water process as the post-drinking water baseline anchor point, and when the single pet drinking water process ends immediately followed by an abnormal disturbance interval, the weight average of the first steady-state baseline interval after the end of the abnormal disturbance interval is extracted as the post-drinking water baseline anchor point; the difference between the pre-drinking water baseline anchor point and the post-drinking water baseline anchor point is taken as the initial net weight;

[0021] The microcontroller multiplies the environmental drift rate corresponding to the single pet drinking water process by the time interval of the single pet drinking water process to obtain the drift loss, and subtracts the drift loss from the initial net weight to obtain the actual net weight data.

[0022] Preferably, when the current weight average in the original weight data sequence is less than the preset empty bowl weight threshold, the microcontroller determines that the water bowl is in a removed state, suspends the net weight calculation and records the removal timestamp; when the current weight average is restored to be greater than the preset empty bowl weight threshold, it is determined that the water bowl is in a put-back state, and the suspension state is released;

[0023] The microcontroller compares the time interval between the removal timestamp and the put-back timestamp, and determines it as a water change event and clears the single-day drinking water accumulator when the time interval is greater than or equal to the preset water change time interval threshold, and determines it as an in-place adjustment event and retains the single-day drinking water accumulator when the time interval is less than the preset water change time interval threshold;

[0024] The microcontroller clears the single-day drinking water accumulator when a cross-day time signal is detected.

[0025] Preferably, the reference calculation module obtains the pet type, current body weight, current age, sterilization status, daily main food type and snack frequency of the target pet, matches the target life stage based on the pet type and current age, and further divides the target life stage into two physiological states of sterilized and unsterilized in the adult stage.

[0026] The reference calculation module calculates the resting energy requirement by using a metabolic body weight formula, and the metabolic body weight formula represents the resting energy requirement as a power function of the current body weight; the maintenance energy requirement is obtained by multiplying the resting energy requirement by a life stage coefficient corresponding to the target life stage, and the life stage coefficient is valued according to the target life stage and the sterilization state; the daily recommended water reference value is obtained by multiplying the maintenance energy requirement by a staple type coefficient corresponding to the daily staple type and a snack frequency coefficient corresponding to the snack frequency.

[0027] Preferably, the reference calculation module performs a safety clamping after obtaining the daily recommended water reference value: when the daily recommended water reference value is lower than a preset minimum water threshold, the daily recommended water reference value is taken as the preset minimum water threshold; when the daily recommended water reference value is higher than a preset upper limit of body weight ratio threshold, the daily recommended water reference value is taken as the preset upper limit of body weight ratio threshold.

[0028] Preferably, the age-based early warning module calculates the ratio of the single-day actual water total net weight data to the daily recommended water reference value as the water completion rate.

[0029] The age-based early warning module presets a four-level asymmetric interval system: when the water completion rate is less than a preset low threshold, it is marked as a low attention state, the continuous low count is incremented by one and the continuous normal count is cleared; when the water completion rate is greater than or equal to the preset low threshold and less than or equal to a preset upper limit of normal threshold, it is marked as a normal state, and the continuous low count and the continuous high count are both cleared; when the water completion rate is greater than the preset upper limit of normal threshold and less than or equal to a preset excessive early warning threshold, it is marked as a high attention state, and the continuous high count is incremented by one; when the water completion rate is greater than the preset excessive early warning threshold, it is marked as an excessive attention state, and the continuous high count is incremented by one and the continuous normal count is cleared.

[0030] The age-based early warning module triggers a low early warning signal when the continuous low count reaches a preset continuous day threshold, and triggers an excessive early warning signal when the continuous high count reaches the preset continuous day threshold.

[0031] Preferably, the age-based early warning module calculates the deviation degree of the water completion rate after the early warning signal is triggered, takes the pet type, life stage and early warning signal type of the target pet as the core matching label, and screens the matching sample in the preset intervention strategy case library according to the preset deviation degree classification interval.

[0032] The age-based early warning module sorts the intervention schemes according to the historical efficiency from high to low, selects the intervention scheme at the top of the sorting as the recommended strategy, determines the recommended strategy according to the preset intervention priority rule when the historical efficiencies of multiple intervention schemes are the highest, and pushes the recommended strategy to the user terminal through the wireless communication module.

[0033] Preferably, the user terminal is provided with an intervention effect feedback entry for receiving user-submitted valid or invalid feedback marks;

[0034] The age division early warning module updates the historical effective rate as the ratio of the number of valid feedback samples to the total number of feedback samples according to a preset iteration period for each intervention scheme.

[0035] Compared with the prior art, the present application has the following advantages:

[0036] (1) The pet intelligent drinking water health monitoring pad and monitoring method, by the microcontroller, the original weight data sequence is cut into local analysis window and is judged as three types of intervals of abnormal disturbance, trend change or steady baseline one by one, the environmental drift rate is extracted in the steady baseline interval for compensation of net weight calculation, and the time interval of moving away and returning of the water bowl is used to distinguish water changing and in-place adjustment event management of the single-day drinking water accumulator. The effect achieved is that: sudden interference is isolated, slow drift is quantified and deducted, water changing and bowl moving are correctly distinguished, and the output of single-day actual drinking water total net weight data only reflects the amount of water actually drunk by the pet. The advantage of this is that even if the environment evaporates obviously or the pet has the habit of moving the bowl, the daily data is still clean and reliable.

[0037] (2) The pet intelligent drinking water health monitoring pad and monitoring method, by the metabolic weight formula, the resting energy requirement is calculated in the form of the power function of the current body weight, and then the life stage coefficient, the main food type coefficient and the snack frequency coefficient according to the life stage and the sterilization state are multiplied to obtain the daily recommended drinking water reference value, and the result is executed with upper and lower limit safety clamping. The effect achieved is that: the reference value follows the real metabolic level, the power function form corrects the systematic deviation caused by the size difference, the multi-dimensional coefficient covers the difference of diet structure and physiological state, and the clamping ensures the reasonable output under extreme body weight. The advantage of this is that the reference value of each pet is suitable for its own needs, and the subsequent early warning has a reasonable reference, and normal drinking water will not be misjudged as abnormal.

[0038] (3) The pet intelligent drinking water health monitoring pad and monitoring method, by four-level asymmetric interval division, the drinking water completion rate state is divided, only when the continuous abnormal count reaches the preset day threshold, the early warning is triggered, after the triggering, the pet type, life stage and deviation degree are taken as labels to sort and output the recommended strategy in the intervention strategy case library according to the historical effective rate, and the effective rate is continuously iterated by user feedback. The effect achieved is that: single-day occasional fluctuation does not alarm, only continuous abnormality triggers, the false alarm rate is greatly reduced; the alarm pushes the personalized intervention scheme verified by history, and the recommendation accuracy is gradually improved with feedback. The advantage of this is that the user will not be confused by false alarms and ignore the real problem, and after receiving the early warning, there is also a specific executable suggestion, forming a closed loop of discovery, intervention and verification. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 The flow chart of the present application. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part 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 work fall within the scope of protection of the present application.

[0041] Embodiment one;

[0042] Please refer to Figure 1 The present application provides a pet intelligent drinking water health monitoring pad, comprising: a high-precision acquisition module, a reference calculation module, and an age-based early warning module are integrated in the monitoring pad, and the specific implementation contents of each module are as follows:

[0043] The high-precision acquisition module includes a weighing sensor, an analog-to-digital converter, a microcontroller, and a wireless communication module. Abnormal disturbances, trend changes, and steady baseline intervals are distinguished through local analysis windows. The single drinking water net weight is calculated using an environmental drift compensation mechanism. The water bowl state recognition and water change event judgment are combined to manage the single-day drinking water accumulator. Finally, the single-day actual drinking water total net weight data is output, and is uploaded to the user terminal through the wireless communication module. In this embodiment, the wireless communication module uses a Wi-Fi communication unit to realize real-time remote transmission of drinking water data, early warning signals, and intervention strategies. The specific process is as follows:

[0044] A weighing sensor is installed at the bottom of the monitoring pad, and the signal output end of the weighing sensor is electrically connected with the input end of the analog-to-digital converter. The data output end of the analog-to-digital converter is electrically connected with the data input end of the microcontroller. The microcontroller interacts with the external user terminal through the wireless communication module.

[0045] The specific execution steps include:

[0046] When the monitoring pad is powered on for the first time or receives the skinning instruction sent by the user terminal, the microcontroller executes the skinning program, reads the initial weight value output by the weighing sensor, and records the initial weight value as the current total weight reference.

[0047] When the monitoring pad is in a continuous monitoring state, the analog-to-digital converter acquires the weight change signal of the weighing sensor according to the preset sampling frequency to obtain the original weight data sequence.

[0048] For the original weight data sequence, the microcontroller divides it into continuous local analysis windows in chronological order.

[0049] Within each local analysis window, the local variance of the original weight data is calculated as the weight fluctuation, and the mean of the differences between adjacent original weight data within the local analysis window is calculated as the average weight change rate.

[0050] When the weight fluctuation exceeds the preset disturbance variance threshold, it is marked as an abnormal disturbance interval;

[0051] When the weight fluctuation is less than or equal to the preset disturbance variance threshold and the average weight change rate is greater than the preset steady-state rate threshold, it is marked as a trend change interval.

[0052] When the weight fluctuation is less than or equal to the preset disturbance variance threshold and the average weight change rate is less than or equal to the preset steady-state rate threshold, it is marked as the steady-state baseline interval.

[0053] Within the steady-state baseline range, the environmental drift rate is extracted based on the linear variation characteristics of the original weight data over time, specifically:

[0054] Within the steady-state baseline interval, select no fewer than the preset minimum number of original weight data points. Perform univariate linear regression fitting with sampling time as the x-axis and original weight data as the y-axis to obtain a fitted straight line of weight change over time. The slope of this fitted straight line is the environmental drift rate of the corresponding steady-state baseline interval, expressed in units of mass / time.

[0055] If a single drinking process spans multiple consecutive steady-state baseline intervals, the environmental drift rate corresponding to the nearest steady-state baseline interval before the start of the drinking process is taken, or the weighted average of the environmental drift rates of each coverage interval calculated according to the time proportion is taken as the environmental drift rate of this drinking process.

[0056] Within the abnormal disturbance range, digital filtering preprocessing is used to smooth and denoise the original weight data;

[0057] The steady-state baseline interval, the trend change interval, and the abnormal disturbance interval after smoothing and denoising are spliced ​​together in chronological order to obtain the filtered weight data sequence.

[0058] In the filtered weight data sequence, identify the starting and ending points where the weight value shows a continuous downward trend;

[0059] When the absolute value of the weight difference between the starting point and the ending point is greater than the preset minimum drinking threshold, and the time interval between the starting point and the ending point is greater than the preset minimum drinking time interval, it is confirmed as a single pet drinking process.

[0060] For each confirmed single pet drinking session, calculate the actual net weight data:

[0061] The average weight of the steady-state baseline interval before the start of a single pet drinking process is extracted and used as the baseline anchor point before drinking.

[0062] The average weight of the steady-state baseline interval after a single pet drinking process is extracted and used as the baseline anchor point after drinking.

[0063] If an abnormal disturbance interval immediately follows the end of a single pet drinking process, the average weight of the first steady-state baseline interval after the end of the abnormal disturbance interval is extracted as the baseline anchor point after drinking.

[0064] Calculate the difference between the baseline anchor point before and after drinking water to obtain the initial net weight;

[0065] Extract the environmental drift rate from the start to the end of a single pet drinking process, and multiply the environmental drift rate by the time interval of a single pet drinking process to calculate the drift loss.

[0066] Subtract the drift loss from the initial net weight to obtain the actual net weight data for a single pet drinking session, and then add the actual net weight data to the daily water accumulator.

[0067] During the calculation of the actual net weight data for a single pet drinking session, the following steps are performed: water bowl status identification and current total weight baseline update.

[0068] When the average weight in the original weight data sequence is less than the preset empty bowl weight threshold, it is determined that the water bowl is being removed, the net weight is calculated and the removal timestamp is recorded.

[0069] When the current average weight recovers to a value greater than the preset empty bowl weight threshold, the water bowl is determined to be in the put-back state, and the hanging state is released.

[0070] If the time interval between the removal timestamp and the return timestamp is greater than or equal to the preset water change interval threshold, it is determined as a water change event, and the daily drinking water accumulator is cleared.

[0071] If the time interval is less than the preset water change time interval threshold, it is determined to be an in-situ adjustment event, and the daily drinking water accumulator is retained; the average weight of the steady-state baseline interval after the water bowl is put back is updated to the new current total weight benchmark.

[0072] When a cross-day time signal is detected, the daily water accumulator is reset to zero.

[0073] The value in the daily water accumulator is used as the total net weight of actual drinking water for the day and uploaded to the user terminal via the Wi-Fi communication unit.

[0074] It should be noted that the integrated hardware architecture enables non-invasive monitoring without modifying existing pet water bowls, and the installation is convenient and compatible with most pet water bowl sizes on the market.

[0075] By accurately distinguishing abnormal disturbances, trend changes, and steady-state baseline ranges through local analysis windows, external interference signals such as pet trampling, bowl bumping, and object placement can be effectively filtered out, greatly improving the anti-interference capability of the data acquisition process and preventing invalid data from being mixed into the monitoring sequence.

[0076] Introducing an environmental drift compensation mechanism can eliminate the long-term impact of environmental factors such as temperature, humidity, and air pressure on weighing accuracy, ensuring the accuracy of single drinking water net weight calculation results and solving the problem of data deviation caused by environmental interference in traditional weighing and monitoring equipment.

[0077] By combining water bowl status recognition and water change event judgment logic, it can accurately distinguish between three application scenarios: water bowl removal, water change, and in-situ adjustment, avoiding problems such as data mis-accumulation or incorrect zeroing, and ensuring the completeness and authenticity of daily drinking water data statistics.

[0078] With the daily drinking water accumulator management mechanism that automatically resets across days, continuous automated monitoring is achieved without human intervention. The output data on the actual total net weight of drinking water for each day is accurate and reliable, providing precise basic data support for the subsequent benchmark calculation module's water demand assessment and the age-based early warning module's health risk determination.

[0079] The baseline calculation module calculates the corresponding daily recommended water intake baseline value based on the individual characteristics and feeding information of the target pet, using a metabolic weight formula combined with multi-dimensional correction coefficients. The specific process is as follows:

[0080] Obtain the target pet's pet type, current weight, current age, spayed / neutered status, daily staple food type, and snack frequency.

[0081] Among them, the types of daily staple foods include: dry food, wet food, mixed food, freeze-dried food, and homemade fresh food;

[0082] Snack frequency includes: rarely eating, occasionally feeding, and frequently feeding;

[0083] We will construct a pre-defined life stage classification rule for cats and dogs, within which different species will have different age boundary standards in months.

[0084] Based on the target pet's pet type and current age, match and determine its target life stage;

[0085] In the adult stage, a further distinction is made between sterilized and non-sterile physiological states to match the differences in drinking water requirements under different metabolic levels.

[0086] The preferred life stage division rules and corresponding coefficient values ​​can be found in the table below:

[0087] .

[0088] Resting energy requirements are calculated using the internationally recognized metabolic body mass formula:

[0089] Resting energy requirement (RER) = 70 × current body weight (kg)^0.75;

[0090] Among them, 0.75 is the metabolic body mass index, which, compared with the linear body mass formula, can more accurately reflect the differences in metabolic rate among pets of different sizes, avoiding the problem of underestimating the water requirements of small pets and overestimating the water requirements of large pets.

[0091] Based on resting energy requirements and combined with preset life stage coefficients, maintenance energy requirements are calculated as follows:

[0092] Maintenance Energy Demand (MER) = RER × Life Stage Coefficient;

[0093] The life stage coefficient is within a preset range, and its value follows the following rules:

[0094] The coefficient for childhood is greater than that for adulthood, and the coefficient for adulthood is greater than that for old age.

[0095] Within the same growth period, the coefficient of unsterilized individuals is greater than that of sterilized individuals;

[0096] The coefficient for early childhood is greater than that for late childhood; at the same life stage, the coefficient for dogs is greater than that for cats. Optimal coefficient values ​​can be found in the table above.

[0097] Based on the target pet's daily staple food type, query the preset staple food type coefficient table to obtain the corresponding coefficient.

[0098] The staple food type coefficient reflects the substitution effect of different staple foods' water content on drinking water requirements. It falls within a preset value range, with the following rule: the higher the water content of the staple food, the lower the corresponding coefficient. Optimal coefficient values ​​can be found in the table below.

[0099] .

[0100] Based on the target pet's snacking frequency, the corresponding coefficient is obtained by consulting a preset snacking frequency coefficient table. The snacking frequency coefficient reflects the marginal substitution effect of snack intake on daily water needs, and falls within a preset value range. The rule is: the higher the snacking frequency, the lower the corresponding coefficient. Optimal coefficient values ​​can be found in the table below:

[0101] .

[0102] The recommended daily water intake for the target pet is calculated by multiplying multiple coefficients:

[0103] Recommended daily water intake baseline = MER × staple food type coefficient × snack frequency coefficient;

[0104] After the calculation is completed, the preset safety clamping rules are applied to define reasonable value boundaries:

[0105] Lower limit clamping: If the calculation result is lower than the preset minimum drinking water threshold, then the minimum threshold is taken to ensure that the minimum drinking water recommendation is not lower than the bottom line for maintaining basic physiological functions;

[0106] Upper limit clamp: If the calculated result is higher than the preset upper limit threshold for body weight ratio, then the upper limit value is taken. This threshold is based on the clinical safety threshold for the maximum daily water intake of dogs and cats to avoid outputting unreasonable high values ​​due to extreme parameter combinations.

[0107] In this embodiment, the minimum water intake threshold is set to 30 grams / day, and the upper limit threshold for body weight ratio is set to 100 grams / kilogram / day.

[0108] For example, for a very small kitten weighing 0.5 kg, if the calculated baseline value is 22 grams / day, then clamp it to 30 grams / day; for a very large dog weighing 40 kg, if the calculated baseline value is 4500 grams / day, then clamp it to 4000 grams / day.

[0109] The clinical significance of lower limit clamping is to prevent the calculated value of very small dogs and cats from being too low due to their small weight, thus losing its clinical reference value, and to ensure that the minimum recommended water intake is not lower than the bottom line required to maintain basic physiological functions.

[0110] The upper limit clamp references the safe threshold of the maximum daily water intake for dogs and cats to avoid misleading users by outputting unreasonable high values ​​due to extreme combinations of parameters.

[0111] Example of optimized calculation: In one specific implementation, referring to the optimized coefficient values ​​in the table above, four typical feeding scenarios are selected for complete calculation explanation:

[0112] Example 1: Adult, spayed / neutered cat, primarily eats dry food, rarely eats treats.

[0113] The target pet weighs 4.2 kg, has a life stage coefficient of 1.2, a staple food type coefficient of 1.0, and a treat frequency coefficient of 1.0.

[0114] RER=70×4.2^0.75=70×2.99≈209.3

[0115] MER = 209.3 × 1.2 ≈ 251.2

[0116] Recommended daily water intake = 251.2 × 1.0 × 1.0 ≈ 251g / day

[0117] Example 2: Late-stage puppy, mainly fed dry food, with occasional treats;

[0118] The target pet has a weight of 5.0 kg, a life stage coefficient of 2.0, a staple food type coefficient of 1.0, and a treat frequency coefficient of 0.9.

[0119] RER=70×5.0^0.75=70×3.34≈233.8

[0120] MER = 233.8 × 2.0 ≈ 467.6

[0121] Recommended daily water intake = 467.6 × 1.0 × 0.9 ≈ 421g / day

[0122] Example 3: Senior cat, mainly wet food, frequently given treats.

[0123] The target pet weighs 4.8 kg, has a life stage coefficient of 1.1, a staple food type coefficient of 0.67, and a treat frequency coefficient of 0.8.

[0124] RER=70×4.8^0.75=70×3.37≈235.9

[0125] MER = 235.9 × 1.1 ≈ 259.5

[0126] Recommended daily water intake = 259.5 × 0.67 × 0.8 ≈ 139g / day

[0127] Example 4: Adult unneutered dog, fed a mixed diet, and frequently given treats.

[0128] The target pet has a weight of 15.0 kg, a life stage coefficient of 1.8, a staple food type coefficient of 0.83, and a treat frequency coefficient of 0.8.

[0129] RER=70×15.0^0.75=70×7.62≈533.4

[0130] MER = 533.4 × 1.8 ≈ 960.1

[0131] Recommended daily water intake = 960.1 × 0.83 × 0.8 ≈ 637g / day.

[0132] It should be noted that using the metabolic weight formula instead of the traditional linear weight coefficient method can correct for differences in metabolic rate among pets of different sizes, avoiding the systematic bias of underestimating the water requirements of small pets and overestimating the water requirements of large pets. Combined with multi-dimensional coefficient correction based on life stage, sterilization status, staple food type, and treat frequency, it can cover diverse feeding scenarios, making the calculated drinking baseline value more consistent with the actual metabolic needs of pets. The safety clamping rule can ensure the rationality of the output value under extreme weight scenarios, ensuring that the drinking recommendations have clinical reference value. The entire calculation can be completed locally on the device without relying on external data support.

[0133] Age-based early warning module: Based on measured water intake and personalized baseline values, it calculates the water intake completion rate, generates early warning signals using an asymmetric interval system combined with a continuous anomaly detection mechanism, and generates personalized recommendation strategies through an intervention case library. The specific process is as follows:

[0134] Obtain the target pet's actual total net weight of water consumed in a single day and the daily recommended water intake baseline, and calculate the water consumption completion rate:

[0135] Drinking water completion rate = (Total net weight of actual water consumed on a single day / Daily recommended water intake) × 100%;

[0136] A four-level asymmetric interval system is pre-set to determine the daily drinking status based on the drinking completion rate:

[0137] If the drinking water completion rate is less than the preset low threshold: mark it as a low concern state, increment the count for consecutive low rates by 1, and reset the count for consecutive normal rates to zero.

[0138] If the drinking water completion rate is greater than or equal to the preset low threshold and less than or equal to the preset normal upper threshold, it is marked as a normal state, and both the consecutive low count and the consecutive high count are cleared to zero.

[0139] If the water consumption completion rate is greater than the preset normal upper limit threshold and less than or equal to the preset excessive warning threshold: it is marked as a high observation state. The count of consecutive high rates will be accumulated, and no warning will be triggered for the time being. Instead, the user will be prompted by the interface color gradient.

[0140] If the water consumption completion rate exceeds the preset excessive warning threshold: mark it as an excessive concern state, increment the count by 1 for consecutive high rates, and reset the count to zero for consecutive normal rates.

[0141] The preferred values ​​for the interval thresholds can be found as follows:

[0142] .

[0143] A preset continuous anomaly detection window is established, and an early warning signal is only generated if the abnormal state persists.

[0144] Low Water Alert: When the consecutive low water count reaches the preset consecutive days threshold, a low water alert signal is triggered, prompting users to pay attention to their pets' insufficient water intake and simultaneously pushing water replenishment intervention suggestions.

[0145] Overdose warning: When the continuous high count reaches the preset consecutive days threshold, an overdose warning signal is triggered, indicating that the user's pet has persistent symptoms of excessive drinking and there is a pathological risk, and it is recommended to seek medical attention.

[0146] An abnormal status on a single day will not trigger an alert. In this embodiment, the preset threshold for consecutive days is 2 days.

[0147] When the target pet has no valid drinking water record on a certain monitoring day, all continuous abnormality counters are reset to zero.

[0148] When the warning signal is triggered, the deviation of the current drinking water completion rate is first calculated: with the 100% completion rate corresponding to the daily recommended drinking water baseline value as a reference, the absolute value of the difference between the daily drinking water completion rate and 100% is calculated as the deviation of the drinking water completion rate.

[0149] In scenarios with low warning levels, the difference is taken as a value lower than the preset low threshold; in scenarios with excessive warning levels, the difference is taken as a value higher than the preset excessive warning threshold.

[0150] Using the target pet's pet type, life stage, and warning signal type as core matching tags, and combined with preset deviation degree grading intervals, matching samples are screened in the basic version of the intervention strategy case library. The basic version of the intervention strategy case library is pre-grouped and stored according to four dimensions: warning type, pet type, life stage, and deviation degree interval.

[0151] During the screening process, sample groups with consistent core labels are first matched, then the preset grade interval to which the current deviation level belongs is determined, and all historical sample records corresponding to that interval are extracted.

[0152] For each intervention program corresponding to the selected historical samples, a ranking and screening algorithm is used to statistically analyze and compare the historical effectiveness. The intervention programs are ranked from high to low according to their historical effectiveness, and the intervention program ranked first is selected as the preferred recommendation strategy.

[0153] If multiple intervention programs have the same historical effectiveness rate, the final recommended program will be determined according to the preset intervention priority rules: hydration interventions take precedence over dietary adjustment interventions, and dietary adjustment interventions take precedence over medical advice interventions.

[0154] The final recommended strategy is pushed to the user terminal via the Wi-Fi communication unit.

[0155] The basic self-learning mechanism is an iterative mechanism for the basic intervention strategy case library. The specific implementation process is as follows:

[0156] The user terminal has an entry point for feedback on the intervention effect. After adopting the recommended strategy, users can submit feedback markings of "effective" or "ineffective".

[0157] The system updates the case library data in batches at fixed times every day according to a preset iteration cycle. For each intervention plan under each feature group, it counts the number of effective feedback samples and the total number of feedback samples. According to the formula that the historical effectiveness rate is equal to the number of effective feedback samples divided by the total number of feedback samples, the system recalculates and updates the historical effectiveness rate of the corresponding intervention plan.

[0158] As user feedback data continues to accumulate, the efficiency of each solution in the case library is continuously iterated and optimized, and the accuracy of subsequent recommendations will gradually improve.

[0159] It should be noted that the combination of asymmetric intervals and continuous anomaly detection can effectively filter out false alarms caused by occasional daily fluctuations, balancing pathological risk sensitivity with user experience; the data missing reset mechanism can avoid false trend judgments caused by data interruption, improving the reliability of early warnings; the intervention recommendation logic based on historical effectiveness is simple and easy to implement, can run stably on local devices, and meets the functional requirements of basic products.

[0160] The aforementioned basic intervention strategy case library and multidimensional feature intervention case library are collectively referred to as the intervention strategy case library.

[0161] Example 2: Cloud-based Optimized Advanced Architecture

[0162] The high-precision data acquisition module in this embodiment is exactly the same as that in Embodiment 1. The difference is that the benchmark calculation module and the age-based early warning module adopt a cloud-based advanced architecture. Based on the locally acquired data, cloud computing power is used to achieve more refined clustering analysis and intelligent matching, further improving the personalization of drinking water benchmarks and the accuracy of intervention plan adaptation.

[0163] Cloud-based advanced benchmark calculation module:

[0164] Obtain the target pet's current weight, current age, breed type, and daily staple food type.

[0165] Construct a target pet life stage feature library, which includes the target pet’s puppy, adult, and senior stages, and pre-sets a standard age range for each life stage.

[0166] The target pet's current age is entered into the target pet life stage feature database for matching, the standard age range in which the current age is located is determined, and the target life stage corresponding to the target pet is output.

[0167] Obtain sample datasets from a multi-source pet health database in the cloud via a network interface;

[0168] Select sample data that are consistent with the basic feature labels of the target pet from the sample dataset to construct a homogeneous sample set; the basic feature labels include, but are not limited to: breed type label, physiological development label, and sterilization status label.

[0169] For each life stage, a preset number of sample data belonging to that life stage are extracted from a homogeneous sample set. The sample data includes: sample weight, sample staple food type, multiple environmental temperature and humidity monitoring values ​​during the sample monitoring period, and multiple actual water consumption monitoring values ​​during the sample monitoring period.

[0170] For the sample data within each life stage, multiple staple food sample groups are formed based on the staple food type, corresponding to the number of staple food types. Staple food types include: dry food, wet food, and mixed food.

[0171] For each staple food sample group, the ratio of the actual average daily water intake to the body weight of each sample in the group was calculated to obtain the water requirement per unit body weight of each sample.

[0172] The actual daily average water consumption was obtained by summing up multiple actual water consumption monitoring values ​​within the sample monitoring period to obtain the total water consumption, and then dividing the total water consumption by the number of days in the sample monitoring period to obtain the actual daily average water consumption.

[0173] The sample weight is the measured weight value directly extracted from the sample data.

[0174] For each sample within each staple food sample group, the average value of multiple environmental temperature and humidity monitoring values ​​during the sample monitoring period is calculated to obtain the average environmental temperature and humidity of the sample.

[0175] The water requirement per unit body weight of each sample is concatenated with the corresponding average ambient temperature and humidity to construct a multidimensional feature vector for each sample.

[0176] A density-based clustering algorithm was used to perform cluster analysis on the multidimensional feature vectors in each staple food sample group. Samples corresponding to multidimensional feature vectors with local density below a preset density threshold were identified as outliers and removed, thus obtaining the core sample clusters under each staple food type.

[0177] For each core sample cluster, the maximum and minimum values ​​of water demand per unit body weight are extracted and used as the upper and lower limits of the basic water demand coefficient for the corresponding staple food type.

[0178] The life stages, staple food types, and corresponding lower and upper limits of the basic water requirement coefficient are structured and organized to construct a mapping table of life stages, staple food types, and basic water requirement coefficients.

[0179] Based on the target pet’s target life stage and daily staple food type, query the life stage-staple food type-basic water requirement coefficient mapping table to obtain the corresponding target basic water requirement coefficient range.

[0180] Furthermore, a correlation analysis was conducted based on the physiological water metabolism characteristics of different staple food types to determine staple food correction factors. The specific determination rules are as follows:

[0181] When the daily staple food is dry food, because dry food itself has low water content, the pet's metabolic water requirement tends to be at the upper limit of the distribution. The staple food correction factor takes the upper limit of the target basic water requirement coefficient range.

[0182] When the daily staple food type is wet grain, due to the high water content of wet grain, the additional water demand tends to be at the lower limit of the distribution. The staple food correction factor is taken as the lower limit of the target basic water demand coefficient range.

[0183] When the daily staple food type is mixed grain, the staple food correction factor is taken as the median value of the target basic water requirement coefficient range.

[0184] Based on the target pet's current weight and staple food correction factor, calculate its corresponding daily recommended water intake baseline value. The calculation formula is as follows:

[0185] ;

[0186] in, This indicates the recommended daily water intake (in mL / day). This indicates your current weight (in kg). This indicates the staple food correction factor (unit: mL / kg / day).

[0187] It should be noted that this module divides the life stages based on the target pet's age in months, and constructs a homogeneous sample set by combining multi-dimensional characteristics such as breed, physiological development, and sterilization status. By using a density-based clustering algorithm to remove abnormal outliers, it can improve the problem of benchmark value deviation caused by insufficient consideration of individual differences in traditional general drinking water standards, making the statistical results of the basic water demand coefficient more scientific and targeted.

[0188] By introducing a staple food type differentiation correction mechanism, and combining the physiological water metabolism characteristics of dry food, wet food, and mixed food, the upper limit, lower limit, and median value of the basic water requirement coefficient range are respectively taken as staple food correction factors to match the actual metabolic water requirement of pets under different dietary structures, which can reduce the problem that a single benchmark value is difficult to adapt to diverse dietary scenarios.

[0189] The constructed life stage-staple food type-basic water requirement coefficient mapping table integrates sample data from a multi-source pet health database in the cloud, covering pet groups of different breeds and physiological states. It has good universality and scalability, and can quickly query and match the basic water requirement coefficient range of the target pet.

[0190] The final calculated daily recommended water intake baseline value is tailored to the individual characteristics and living habits of the target pet, providing a personalized reference standard for the subsequent age-based early warning module to calculate water intake deviation and determine health risks, which helps to improve the accuracy and reliability of subsequent health warnings.

[0191] Cloud-based advanced age-based early warning and intervention module:

[0192] Obtain the target pet's actual total net weight of water consumed in a single day, the daily recommended water intake baseline, the target life stage, and the target pet's multidimensional characteristic tags (including but not limited to: breed type, current weight, spayed / neutered status, and daily staple food type).

[0193] Based on the target pet's actual total net weight of water consumed in a single day and the daily recommended water intake baseline, the water deviation degree, which characterizes the degree of deviation from the actual water consumption status, is calculated. The calculation formula is as follows:

[0194] ;

[0195] Where D represents the target pet's deviation from drinking water. The data represents the total net weight of the target pet's actual daily water consumption. The recommended daily water intake for the target pet;

[0196] When D is positive, it indicates insufficient water intake; when D is negative, it indicates excessive water intake.

[0197] Preset exclusive warning threshold sets for each life stage of the target pet (juvenile stage, adult stage, and old age stage);

[0198] The dedicated early warning threshold set includes: initial early warning threshold, severe early warning threshold, excessive thirst early warning threshold, and preset assessment days (wherein, the absolute values ​​of each threshold corresponding to the elderly stage are smaller than those of the childhood and adult stages, in order to achieve highly sensitive monitoring).

[0199] Extract the specific warning threshold set corresponding to the current target life stage of the target pet, and compare the water deviation degree D and its time series characteristics within the preset evaluation days with the corresponding specific warning threshold set:

[0200] If the daily water consumption deviation D is greater than or equal to the initial warning threshold, an initial warning signal will be generated.

[0201] If the deviation degree D of drinking water is greater than or equal to the severe warning threshold for a consecutive preset number of assessment days, a severe warning signal will be generated.

[0202] Among them, the severe warning signal replaces the first warning signal of the same period after it is generated.

[0203] If the deviation of drinking water D is less than zero, and its absolute value is greater than or equal to the preset excessive drinking warning threshold, then an excessive drinking warning signal is generated.

[0204] For each type of warning signal, historical sample data of various intervention programs implemented in the past are extracted from the historical pet health intervention database, including:

[0205] Pre-intervention characteristics: life stage, breed, weight, and deviation range of water intake;

[0206] Post-intervention effectiveness indicators include: rate of recovery of water intake, weight fluctuation rate, and incidence of adverse reactions.

[0207] The above features and indicators are integrated into a structured dataset to form a multidimensional feature matrix for each intervention scheme.

[0208] For each intervention plan, the original observation data of all historical samples under that plan are processed by range transformation to map them to the interval [0, 1]. The specific calculation process is as follows:

[0209] ;

[0210] in, Let be the standardized index value of the i-th historical sample under the j-th intervention effect index. For the i-th historical sample, the original observation value under the j-th intervention effect index is given. Let i be the minimum and maximum values ​​of the j-th indicator in all historical samples, i be the index of the historical sample (i = 1, 2, ..., n), j be the index of the post-intervention effect indicator (j = 1, 2, ..., m), n be the total number of historical samples, and m be the total number of post-intervention effect indicators.

[0211] For each intervention plan, based on the distribution characteristics of its historical sample data, the objective importance of each indicator is determined using the entropy weight method, as follows:

[0212] For the historical sample set corresponding to this intervention plan, calculate the feature weight of the i-th historical sample under the effect index after the j-th intervention. The calculation formula is as follows:

[0213] ;

[0214] Based on the aforementioned feature weights, the information entropy of the j-th post-intervention effect indicator is calculated. The calculation formula is:

[0215] ;

[0216] Where k represents a constant, and ;

[0217] when When, define In fact, it adopts extreme processing, which means that when a sample makes no contribution to a certain indicator, the information contribution of the sample to the indicator is zero.

[0218] Based on the information entropy results, calculate the objective weight of the j-th intervention effect indicator. The calculation formula is:

[0219] ;

[0220] For each historical sample, the standardized values ​​of each historical sample on each indicator under this scheme are weighted and summed with their corresponding objective weights to obtain the comprehensive intervention effectiveness index of that sample under this scheme. The calculation formula is as follows:

[0221] ;

[0222] in, This represents the overall intervention efficacy index for the i-th historical sample under this intervention program.

[0223] For each historical sample, calculate its comprehensive intervention efficacy index under all intervention programs, and select the intervention program corresponding to the largest comprehensive intervention efficacy index as the best guidance strategy for that historical sample;

[0224] Based on the three-dimensional correlation between early warning signal type, pre-intervention feature vector and corresponding optimal guidance strategy, a multi-dimensional feature intervention case library is constructed by structurally integrating the full set of historical sample data.

[0225] The pre-intervention feature vector is constructed as follows: one-hot encoding is used for life stage and breed category features, and range normalization is used to the range [0, 1] for weight and water deviation range numerical features. Then, each dimension is concatenated into a numerical vector for subsequent similarity calculation.

[0226] In the real-time decision-making stage, the current warning signals and pre-intervention features of the target pet are first captured and converted into feature vectors and then entered into the case library. Subsequently, relying on the nearest neighbor search algorithm, the feature similarity score between the target pet and each historical sample in the library is calculated, and the top q historical samples with the highest similarity scores are selected, where q is the preset number of nearest neighbor samples.

[0227] Finally, from these q highly similar samples, the intervention plan with the best comprehensive intervention effectiveness index is extracted as the basis for decision-making, and a personalized recommendation strategy for the current target pet is generated in real time and pushed to the user terminal.

[0228] The advanced self-learning mechanism iterates at fixed intervals: After each batch of user feedback is completed, the user feedback and related re-examination data are converted into the corresponding sample's post-intervention effect indicator update value according to the preset feedback-indicator mapping rules (for example, after a user submits valid feedback, the system prompts the user to enter the re-examination water intake and re-examination weight, and updates the water intake recovery rate and weight fluctuation rate accordingly). The comprehensive intervention efficacy index of each plan is recalculated, and the mapping relationship of the best guidance strategy in the case library is updated to achieve continuous self-optimization of the case library.

[0229] It should be noted that setting different exclusive warning threshold sets for different life stages of pets, and appropriately lowering the absolute value of the threshold for the elderly stage, can better adapt to the physiological characteristics of pets of different ages and reduce misjudgments or omissions caused by a single warning standard.

[0230] When constructing a multidimensional intervention case library, we combine multiple intervention effect indicators such as the rate of water recovery, weight fluctuation rate, and incidence of adverse reactions. We determine the objective weight of each indicator through data processing methods and calculate the comprehensive intervention efficacy index. This allows us to screen intervention programs and make the evaluation of the programs more in line with the actual application effect.

[0231] In the strategy generation stage, by matching historical samples with similar features, intervention plans with high comprehensive intervention efficacy index are extracted to generate personalized intervention recommendations that are suitable for the current pet situation, which are more reliable than general suggestions.

[0232] This module, together with the preceding high-precision data acquisition module and benchmark calculation module, forms a complete monitoring-early warning-intervention chain. It can promptly provide users with feedback on any abnormalities in their pet's drinking water and offer corresponding suggestions, helping users to pay attention to their pet's daily drinking health and providing some data support for the long-term health management of their pets.

[0233] This invention employs a dual-layer architecture, combining local and cloud-based implementation. The high-precision data acquisition module is shared across both architectures, ensuring consistent accuracy in drinking water data collection and reducing hardware development costs for multiple product versions. The local architecture performs all calculations on the device itself, allowing for independent operation without internet connectivity. It is convenient to deploy, stable in operation, and meets the monitoring and early warning needs of basic products. The cloud architecture, building upon local data acquisition, leverages cloud computing power and multi-source data to further optimize baseline accuracy and intervention recommendation effects. It supports advanced health management functions and is compatible with high-end products. The overall solution can cover product forms with different computing power configurations and functional positioning, demonstrating strong applicability and scalability.

[0234] A method for monitoring the health of pets using a smart drinking water monitoring pad includes:

[0235] Step 1: Collect raw weight data sequence through weighing sensor, divide the raw weight data sequence into continuous local analysis windows in time order, distinguish abnormal disturbance intervals, trend change intervals and steady-state baseline intervals based on each local analysis window, calculate the actual net weight data of a single pet drinking process by combining environmental drift compensation, and accumulate the data through a daily drinking accumulator to obtain the total actual net weight data of the day.

[0236] Step 2: Calculate the recommended daily water intake based on the target pet's individual characteristics and feeding information;

[0237] Step 3: Generate an early warning signal based on the actual total net weight of drinking water on a single day and the daily recommended drinking water baseline value, and push the early warning signal to the user terminal.

[0238] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart pet drinking health monitoring pad, characterized in that, The monitoring pad integrates a high-precision data acquisition module, a baseline calculation module, and an age-based early warning module. The high-precision data acquisition module includes: a weighing sensor, an analog-to-digital converter (ADC), a microcontroller, and a wireless communication module. The signal output terminal of the weighing sensor is electrically connected to the input terminal of the ADC, and the data output terminal of the ADC is electrically connected to the data input terminal of the microcontroller. The microcontroller communicates with the user terminal through the wireless communication module. The microcontroller divides the acquired raw weight data sequence into continuous local analysis windows according to time sequence. Based on each local analysis window, it distinguishes between abnormal disturbance intervals, trend change intervals, and steady-state baseline intervals. Combined with environmental drift compensation, it calculates the actual net weight data of a single pet drinking process and accumulates it through a daily drinking accumulator to obtain the total actual net weight data of the day. The baseline calculation module calculates the recommended daily water intake baseline value based on the individual characteristics and feeding information of the target pet; The age-based early warning module generates an early warning signal based on the actual total net weight of water consumed on a single day and the daily recommended water intake baseline, and pushes it to the user terminal via the wireless communication module.

2. The pet smart drinking health monitoring pad according to claim 1, characterized in that: The microcontroller calculates the local variance of the original weight data as the weight fluctuation within each local analysis window, and calculates the mean of the differences between adjacent original weight data within the local analysis window as the average weight change rate. When the weight fluctuation exceeds the preset disturbance variance threshold, the local analysis window is marked as an abnormal disturbance interval; When the weight fluctuation is less than or equal to the preset disturbance variance threshold and the average weight change rate is greater than the preset steady-state rate threshold, the local analysis window is marked as the trend change interval. When the weight fluctuation is less than or equal to the preset disturbance variance threshold and the average weight change rate is less than or equal to the preset steady-state rate threshold, the local analysis window is marked as the steady-state baseline interval.

3. The pet smart drinking health monitoring pad according to claim 2, characterized in that: The microcontroller concatenates the steady-state baseline interval, trend change interval, and abnormal disturbance interval after smoothing and denoising into a filtered weight data sequence in chronological order. In the filtered weight data sequence, it identifies the starting and ending points where the weight value shows a continuous downward trend. When the absolute value of the weight difference between the starting and ending points is greater than the preset minimum drinking threshold and the time interval between the starting and ending points is greater than the preset minimum drinking time interval, it is confirmed as a single pet drinking process. The microcontroller selects no fewer than the preset minimum number of original weight data points within the steady-state baseline interval, performs univariate linear regression fitting with sampling time as the abscissa and original weight data as the ordinate, and uses the slope of the fitted line as the environmental drift rate within the steady-state baseline interval. The microcontroller extracts the average weight of the steady-state baseline interval before the start of a single pet drinking process as the pre-drinking baseline anchor point, and extracts the average weight of the steady-state baseline interval after the end of a single pet drinking process as the post-drinking baseline anchor point. When an abnormal disturbance interval immediately follows after the end of a single pet drinking process, the average weight of the first steady-state baseline interval after the end of the abnormal disturbance interval is extracted as the post-drinking baseline anchor point. The difference between the baseline anchor point before drinking water and the baseline anchor point after drinking water is used as the initial net weight. The microcontroller multiplies the environmental drift rate corresponding to a single pet drinking process by the time interval of the single pet drinking process to obtain the drift loss amount, and subtracts the drift loss amount from the initial net weight to obtain the actual net weight data.

4. The pet intelligent drinking health monitoring pad according to claim 1, characterized in that: When the average current weight in the original weight data sequence is less than the preset empty bowl weight threshold, the microcontroller determines that the water bowl is in the removal state, calculates the net weight and records the removal timestamp; when the average current weight recovers to be greater than the preset empty bowl weight threshold, it determines that the water bowl is in the return state and releases the suspension state. The microcontroller compares the time interval between the removal timestamp and the return timestamp. When the time interval is greater than or equal to the preset water exchange time interval threshold, it is determined as a water exchange event and the daily drinking water accumulator is cleared. When the time interval is less than the preset water exchange time interval threshold, it is determined as an in-situ adjustment event and the daily drinking water accumulator is retained. The microcontroller resets the daily water consumption accumulator when it detects a cross-day time signal.

5. The pet intelligent drinking health monitoring pad according to claim 1, characterized in that: The benchmark calculation module obtains the target pet's pet type, current weight, current age in months, spaying / neutering status, daily staple food type and snack frequency. Based on the pet type and current age in months, it matches the target life stage. The target life stage in adulthood is further divided into two physiological states: spayed / neutered and non-spayed / neutered. The baseline calculation module uses the metabolic weight formula to calculate the resting energy requirement, which represents that the resting energy requirement changes with the current weight in a power function manner. The maintenance energy requirement is obtained by multiplying the resting energy requirement by the life stage coefficient corresponding to the target life stage. The life stage coefficient is selected according to the target life stage and sterilization status. The maintenance energy requirement is then multiplied by the staple food type coefficient corresponding to the daily staple food type and the snack frequency coefficient corresponding to the snack frequency to obtain the daily recommended water intake baseline value.

6. The pet smart drinking health monitoring pad according to claim 5, characterized in that: After obtaining the daily recommended water intake baseline value, the baseline calculation module performs a safety clamp: when the daily recommended water intake baseline value is lower than the preset minimum water intake threshold, the daily recommended water intake baseline value is taken as the preset minimum water intake threshold; when the daily recommended water intake baseline value is higher than the preset upper limit threshold of body weight ratio, the daily recommended water intake baseline value is taken as the preset upper limit threshold of body weight ratio.

7. The pet intelligent drinking health monitoring pad according to claim 1, characterized in that: The age-based early warning module calculates the ratio of the actual total net weight of water consumed on a single day to the daily recommended water intake baseline as the water consumption completion rate; The age-based early warning module is pre-set to a four-level asymmetric interval system: when the drinking water completion rate is less than the preset low threshold, it is marked as a low concern state, the consecutive low count is incremented by one and the consecutive normal count is cleared to zero; when the drinking water completion rate is greater than or equal to the preset low threshold and less than or equal to the preset normal upper threshold, it is marked as a normal state, and both the consecutive low count and the consecutive high count are cleared to zero. When the water completion rate is greater than the preset normal upper limit threshold and less than or equal to the preset excessive warning threshold, it is marked as a high observation state, and the consecutive high count is incremented by one; when the water completion rate is greater than the preset excessive warning threshold, it is marked as an excessive concern state, the consecutive high count is incremented by one and the consecutive normal count is reset to zero. The age-based early warning module triggers a low warning signal when the continuous low count reaches a preset consecutive days threshold, and triggers an excessive warning signal when the continuous high count reaches a preset consecutive days threshold.

8. The pet intelligent drinking health monitoring pad according to claim 7, characterized in that: The age-based early warning module calculates the degree of deviation in the drinking completion rate after the early warning signal is triggered. It uses the target pet's pet type, life stage, and early warning signal type as core matching tags, and combines them with a preset deviation degree grading interval to select matching samples from a preset intervention strategy case library. The age-based early warning module sorts each intervention plan from highest to lowest based on its historical effectiveness, selects the intervention plan ranked first as the recommended strategy, and determines the recommended strategy according to the preset intervention priority rules when multiple intervention plans have the same historical effectiveness, and pushes it to the user terminal via the wireless communication module.

9. A smart pet drinking health monitoring pad according to claim 8, characterized in that: The user terminal is equipped with an intervention effect feedback entry point, which is used to receive valid or invalid feedback flags submitted by users; The age-based early warning module calculates the number of effective feedback samples and the total number of feedback samples for each intervention plan according to a preset iteration cycle, and updates the ratio of the number of effective feedback samples to the total number of feedback samples as the historical effectiveness rate.

10. A method for monitoring the health of a pet's intelligent drinking water monitoring pad, applied to the pet intelligent drinking water monitoring pad described in any one of claims 1-9, characterized in that, include: Step 1: Collect raw weight data sequence through weighing sensor, divide the raw weight data sequence into continuous local analysis windows in time order, distinguish abnormal disturbance intervals, trend change intervals and steady-state baseline intervals based on each local analysis window, calculate the actual net weight data of a single pet drinking process by combining environmental drift compensation, and accumulate the data through a daily drinking accumulator to obtain the total actual net weight data of the day. Step 2: Calculate the recommended daily water intake based on the target pet's individual characteristics and feeding information; Step 3: Generate an early warning signal based on the actual total net weight of drinking water on a single day and the daily recommended drinking water baseline value, and push the early warning signal to the user terminal.

Images