An artificial intelligence-based multi-sensor pet home management system and method

Abstract

The application discloses a kind of multi-sensor pet home management system and method based on artificial intelligence, it is related to pet home management technical field, the method is in view of the problem that prior art cannot distinguish device itself abnormality and external temporary interference leads to response abnormality, realizes intelligent management by the following steps: first, execute interactive action and collect initial data, if below threshold value then enter delay verification process, collect secondary data after pet attention dispersion period to carry out preliminary abnormality attribution;Only when initially judged that device interactive function is abnormal, start cross-validation process based on historical interaction data and environmental state record to review;Only after review confirms that device itself is abnormal, calculate and update device interaction parameter;Finally, according to the feedback of updated interaction effect, the interactive expected threshold value and attention dispersion period parameter of the system are adaptively optimized.

Description

Technical Field

[0001] This invention relates to the field of pet home management technology, specifically a multi-sensor pet home management system and method based on artificial intelligence. Background Technology

[0002] In existing pet home equipment management systems, a judgment mechanism based on a single instant response is commonly used. The common technical solution is that after the device performs a preset action, such as dispensing food or activating a toy, preset response indicators are monitored in a preset window, such as food consumption or pet approach signals. If the monitored response data is lower than a certain preset threshold, the system determines that the device action is invalid or unwelcome, and automatically triggers the adjustment of the device's operating parameters.

[0003] The drawback of this technical solution is that its judgment logic is single-dimensional and instantaneous, and it cannot effectively distinguish between "poor actual performance caused by improper device function or parameter settings" and "abnormal response caused by accidental factors such as temporary interference from the external environment or short-term distraction of the pet's attention". Since pets are living beings, their behavior is easily affected by a variety of instantaneous and uncontrollable external factors, and the existing system lacks the ability to identify such uncertainties.

[0004] Therefore, how to enable pet home equipment systems to reliably distinguish the root cause of response abnormalities and avoid unnecessary parameter adjustments triggered by temporary external interference has become a technical problem that urgently needs to be solved in this field.

[0005] To address the above issues, this invention proposes a multi-sensor pet home management system and method based on artificial intelligence. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-sensor pet home management system and method based on artificial intelligence to solve the problems raised in the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A multi-sensor pet home management method based on artificial intelligence includes the following steps: S1. Control the target pet home appliance to perform preset pet interaction actions and collect initial interaction data; S2. When the initial interaction data is lower than the preset interaction expectation threshold, the delayed verification process is entered, and secondary interaction data is collected after the pet's attention is distracted. S3. Based on the comparison between the initial interaction data and the secondary interaction data relative to the expected interaction threshold, a preliminary anomaly attribution judgment is made. S4. Only when the preliminary anomaly attribution judgment result indicates that the pet home device has an abnormal interactive function, a cross-validation process based on historical interaction data and environmental status records will be initiated for verification. S5. Only when the cross-validation process confirms that the pet home device has an abnormal interactive function, calculate the adjustment amount of the interactive parameters and update the interactive parameter values ​​of the pet home device. S6. Based on the feedback of the interaction effect after the interaction parameter value is updated, the interaction expectation threshold and the pet attention distraction period parameter are adaptively optimized.

[0008] S1 further includes the following: Control the target pet home device to perform a complete pet interaction action; After the action is completed, the first monitoring cycle is started, and the interaction-related physical quantity data generated during the cycle are collected by the interactive sensor built into the pet home device. The interaction-related physical quantity data refers to physical signal data generated by the pet's active actions on the device that can be directly measured and quantified by the device's built-in sensors; The collected interaction-related physical quantity data are weighted and summed, and the resulting value is used as the initial interaction evaluation value.

[0009] S2 further includes the following: The initial interaction evaluation value is compared with a preset interaction expectation threshold; If the initial interaction evaluation value is lower than the expected interaction threshold, then a preset period of pet attention distraction will begin. During the period when the pet's attention is distracted, the sending of commands to the pet home device to perform similar pet interaction actions is suspended; At the end of the pet's distraction period, a second monitoring cycle is initiated, and interaction-related physical quantity data generated during this cycle are collected in a non-contact monitoring manner using the interactive sensor. The interaction-related physical quantity data collected during the second monitoring period are weighted and fused, and the calculated value is used as the secondary interaction evaluation value.

[0010] S3 further includes the following: The initial interaction evaluation value is compared with the interaction expectation threshold to obtain a first comparison result; The secondary interaction evaluation value is compared with the interaction expectation threshold to obtain a second comparison result; Based on the combination of the first comparison result and the second comparison result, a preliminary anomaly attribution judgment is performed; If the first comparison result is lower than the expected interaction threshold and the second comparison result is not lower than the expected interaction threshold, the preliminary attribution judgment result is that the pet's temporary lack of interaction is caused by external interference. If both the first comparison result and the second comparison result are lower than the expected interaction threshold, the preliminary attribution judgment result is that the pet home device has an abnormal interactive function.

[0011] S4 further includes the following: When the preliminary anomaly attribution judgment result indicates that the pet home device has an abnormal interactive function, the cross-validation process is initiated. The cross-validation process includes comparison of historical interaction data and verification of environmental status records; The historical interaction data comparison includes the following steps: Query all historical interaction evaluation values ​​of the pet home device within a preset number of historical days that are in the same time period as the current execution time; Calculate the arithmetic mean of all historical interaction evaluation values, and use it as the historical normal level H; Calculate the absolute value of the difference between the initial interaction evaluation value V1 and the historical normal level H to obtain the deviation degree D, D=|V1-H|; The environmental status record verification is achieved by retrieving environmental sensor records for a preset duration before and after the pet interaction action, and analyzing whether there are any events marked as abnormal. The judgment is verified based on whether the deviation D is greater than the preset fluctuation threshold L and whether there are any abnormal events recorded by the environmental sensors. If the deviation D>L and there are no abnormal events recorded by the environmental sensors, the verification result is that the pet home device has an abnormal interactive function. Otherwise, the review result will be that the non-interactive function is abnormal.

[0012] S5 further includes the following: When the review result confirms that the pet home device has an abnormal interactive function, the interaction parameter adjustment amount ΔA is calculated. The formula for calculating the interaction parameter adjustment amount ΔA is as follows: ΔA=(θ-V1)×[1+(θ-V2) / θ+D / H]; Where θ is the expected interaction threshold, V1 is the initial interaction evaluation value, V2 is the secondary interaction evaluation value, D is the deviation, and H is the historical normal level; Obtain the current interaction parameter value C of the pet home device, and calculate the updated interaction parameter value C', where C' = C + ΔA.

[0013] S6 further includes the following: After the interaction parameter value is updated, control the pet home device to perform the same pet interaction action; Collect and calculate the updated interaction evaluation value V f ; V f Compare with the expected interaction threshold θ, if V f If V ≥ θ, then the parameter update is considered valid. f If the value is less than θ, the parameter update is deemed invalid. Perform adaptive optimization based on the determination result: If the result is deemed valid, the pet attention distraction period parameter T is updated to Tnew, where Tnew = α * T, and α is a preset shortening coefficient, 0 < α < 1. If the interaction expectation threshold θ is determined to be invalid, the interaction expectation threshold θ is updated to θnew, θnew=β*θ, where β is a preset reduction coefficient, 0<β<1.

[0014] An AI-based multi-sensor pet home management system includes an interactive control and data acquisition module, a delay verification and secondary monitoring module, an anomaly attribution analysis module, a cross-validation review module, a parameter calculation and update module, and a feedback learning and strategy optimization module. The interactive control and data acquisition module is used to control the pet home device to perform pet interactive actions and collect interactive data; The delay verification and secondary monitoring module is used to manage the pet's attention lapse period and collect secondary interaction data when the initial interaction data is lower than the expected interaction threshold. The anomaly attribution analysis module is used to make a preliminary anomaly attribution judgment based on the comparison relationship between the two interaction data and the interaction expectation threshold. The cross-validation review module is used to review the device interaction function based on historical interaction data and environmental status records when it is initially determined that the device interaction function is abnormal. The parameter calculation and update module is used to calculate the adjustment amount of the interactive parameters and update the device interactive parameters when the device itself is confirmed to be abnormal. The feedback learning and strategy optimization module is used to adaptively optimize the system threshold and delay parameters based on the feedback of the interaction effect after parameter updates.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. By introducing a dual judgment mechanism of "delayed verification" and "cross-verification," the system effectively distinguishes between abnormal pet responses caused by temporary external interference and abnormalities in the device's own interactive functions. Parameter adjustments are only triggered when a problem with the device itself is confirmed through comprehensive verification. This fundamentally avoids invalid or harmful changes to device parameters caused by accidental factors such as environmental noise or brief pet distraction, significantly reducing the number of system erroneous operations. 2. By reducing unnecessary parameter adjustments, the fluctuations in core operating parameters of the equipment, such as feeding amount, toy interaction mode, and water dispensing time, are significantly reduced, resulting in more stable operation. This not only improves the reliability of single-task execution but also reduces wear and tear on hardware such as motors and controllers caused by frequent parameter resets, helping to extend the overall lifespan of pet home equipment. 3. Through the "execution monitoring feedback optimization" process, the system can dynamically and in reverse optimize its core judgment thresholds and behavior waiting parameters based on the actual interaction effects after each adjustment. This allows the system to gradually adapt to the unique behavioral habits and specific environmental backgrounds of pets in users' homes, continuously improving its intelligence level and providing more personalized and reliable services. Attached Figure Description

[0016] Figure 1 This is a flowchart of a multi-sensor pet home management method based on artificial intelligence according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example: Figure 1 As shown, the present invention provides a technical solution. A multi-sensor pet home management method based on artificial intelligence includes the following steps: S1. Control the target pet home appliance to perform preset pet interaction actions and collect initial interaction data; S2. When the initial interaction data is lower than the preset interaction expectation threshold, the delayed verification process is entered, and secondary interaction data is collected after the pet's attention is distracted. S3. Based on the comparison between the initial interaction data and the secondary interaction data relative to the expected interaction threshold, a preliminary anomaly attribution judgment is made. S4. Only when the preliminary anomaly attribution judgment result indicates that the pet home device has an abnormal interactive function, a cross-validation process based on historical interaction data and environmental status records will be initiated for verification. S5. Only when the cross-validation process confirms that the pet home device has an abnormal interactive function, calculate the adjustment amount of the interactive parameters and update the interactive parameter values ​​of the pet home device. S6. Based on the feedback of the interaction effect after the interaction parameter value is updated, the interaction expectation threshold and the pet attention distraction period parameter are adaptively optimized.

[0019] S1 further includes the following: Control the target pet home device to perform a complete pet interaction action; After the action is completed, the first monitoring cycle is started, and the interaction-related physical quantity data generated during the cycle are collected by the interactive sensor built into the pet home device. The interaction-related physical quantity data refers to physical signal data generated by the pet's active actions on the device that can be directly measured and quantified by the device's built-in sensors; The collected interaction-related physical quantity data are weighted and summed, and the resulting value is used as the initial interaction evaluation value.

[0020] S2 further includes the following: The initial interaction evaluation value is compared with a preset interaction expectation threshold; If the initial interaction evaluation value is lower than the expected interaction threshold, then a preset period of pet attention distraction will begin. During the period when the pet's attention is distracted, the sending of commands to the pet home device to perform similar pet interaction actions is suspended; At the end of the pet's distraction period, a second monitoring cycle is initiated, and interaction-related physical quantity data generated during this cycle are collected in a non-contact monitoring manner using the interactive sensor. The interaction-related physical quantity data collected during the second monitoring period are weighted and fused, and the calculated value is used as the secondary interaction evaluation value.

[0021] S3 further includes the following: The initial interaction evaluation value is compared with the interaction expectation threshold to obtain a first comparison result; The secondary interaction evaluation value is compared with the interaction expectation threshold to obtain a second comparison result; Based on the combination of the first comparison result and the second comparison result, a preliminary anomaly attribution judgment is performed; If the first comparison result is lower than the expected interaction threshold and the second comparison result is not lower than the expected interaction threshold, the preliminary attribution judgment result is that the pet's temporary lack of interaction is caused by external interference. If both the first comparison result and the second comparison result are lower than the expected interaction threshold, the preliminary attribution judgment result is that the pet home device has an abnormal interactive function.

[0022] S4 further includes the following: When the preliminary anomaly attribution judgment result indicates that the pet home device has an abnormal interactive function, the cross-validation process is initiated. The cross-validation process includes comparison of historical interaction data and verification of environmental status records; The historical interaction data comparison includes the following steps: Query all historical interaction evaluation values ​​of the pet home device within a preset number of historical days that are in the same time period as the current execution time; Calculate the arithmetic mean of all historical interaction evaluation values, and use it as the historical normal level H; Calculate the absolute value of the difference between the initial interaction evaluation value V1 and the historical normal level H to obtain the deviation degree D, D=|V1-H|; The environmental status record verification is achieved by retrieving environmental sensor records for a preset duration before and after the pet interaction action, and analyzing whether there are any events marked as abnormal. The judgment is verified based on whether the deviation D is greater than the preset fluctuation threshold L and whether there are any abnormal events recorded by the environmental sensors. If the deviation D>L and there are no abnormal events recorded by the environmental sensors, the verification result is that the pet home device has an abnormal interactive function. Otherwise, the review result will be that the non-interactive function is abnormal.

[0023] The above steps aim to provide a secondary confirmation of the initial anomaly attribution judgment, distinguishing between equipment malfunctions and interaction anomalies caused by external, accidental factors. This step constructs a multi-dimensional judgment basis by combining historical interaction data comparison with environmental status record verification. Historical interaction data comparison calculates the deviation between the current interaction evaluation value and historical normal levels to determine whether the current anomaly represents a significant deviation from the equipment's normal performance. Environmental status record verification analyzes environmental sensor data before and after the interaction action to identify whether external interference events exist.

[0024] The purpose of this design is to improve the accuracy and reliability of the system's judgment, avoid unnecessary parameter adjustments triggered by non-equipment factors such as pets being temporarily distracted or sudden environmental noise, thereby reducing the system's misjudgment rate and invalid operations, and ensuring that equipment parameter adjustments are only initiated when actual functions are abnormal.

[0025] S5 further includes the following: When the review result confirms that the pet home device has an abnormal interactive function, the interaction parameter adjustment amount ΔA is calculated. The formula for calculating the interaction parameter adjustment amount ΔA is as follows: ΔA=(θ-V1)×[1+(θ-V2) / θ+D / H]; Where θ is the expected interaction threshold, V1 is the initial interaction evaluation value, V2 is the secondary interaction evaluation value, D is the deviation, and H is the historical normal level; Obtain the current interaction parameter value C of the pet home device, and calculate the updated interaction parameter value C', where C' = C + ΔA.

[0026] The main purpose of the above is to calculate and implement targeted adjustments to the interaction parameters after confirming that the device itself has an abnormal interactive function. This step constructs a formula for calculating the parameter adjustment amount by comprehensively considering multiple factors such as the initial interaction evaluation value, the secondary interaction evaluation value, the expected interaction threshold, historical normal levels, and deviations, thereby achieving dynamic and accurate parameter correction. Its design aims to avoid fixed or simple linear adjustment methods, and instead uses multi-dimensional data fusion calculations to make parameter adjustments more adaptive and scientific. Through this step, the system can optimize its working parameters, such as feeding amount, interaction intensity, and response time, in a timely and reasonable manner when the device malfunctions, thereby restoring or improving the interaction between the device and the pet, and enhancing the device's intelligent adaptability and long-term operational stability.

[0027] S6 further includes the following: After the interaction parameter value is updated, control the pet home device to perform the same pet interaction action; Collect and calculate the updated interaction evaluation value V f ; V f Compare with the expected interaction threshold θ, if V f If V ≥ θ, then the parameter update is considered valid. f If the value is less than θ, the parameter update is deemed invalid. Perform adaptive optimization based on the determination result: If the result is deemed valid, the pet attention distraction period parameter T is updated to Tnew, where Tnew = α * T, and α is a preset shortening coefficient, 0 < α < 1. If the interaction expectation threshold θ is determined to be invalid, the interaction expectation threshold θ is updated to θnew, θnew=β*θ, where β is a preset reduction coefficient, 0<β<1.

[0028] The main purpose of the above content is to enable the system to learn itself and optimize its strategies, and to dynamically adjust the system's key judgment interaction expectation thresholds and the pet's attention distraction period based on the actual interaction effect feedback after parameter updates. This step assesses the effectiveness of parameter updates by shortening the attention span or lowering the interaction expectation threshold, thereby gradually adapting to the current pet's behavior and environmental characteristics. Its design aims to enable the system to continuously evolve, constantly optimizing its judgment logic and behavioral strategies based on long-term operational data. Through this mechanism, the system can not only improve the success rate of single interactions, but also gradually personalize itself over a long period of time, enhancing the overall intelligence and reliability of the service and achieving truly adaptive pet home management.

[0029] An AI-based multi-sensor pet home management system includes an interactive control and data acquisition module, a delay verification and secondary monitoring module, an anomaly attribution analysis module, a cross-validation review module, a parameter calculation and update module, and a feedback learning and strategy optimization module. The interactive control and data acquisition module is used to control the pet home device to perform pet interactive actions and collect interactive data; The delay verification and secondary monitoring module is used to manage the pet's attention lapse period and collect secondary interaction data when the initial interaction data is lower than the expected interaction threshold. The anomaly attribution analysis module is used to make a preliminary anomaly attribution judgment based on the comparison relationship between the two interaction data and the interaction expectation threshold. The cross-validation review module is used to review the device interaction function based on historical interaction data and environmental status records when it is initially determined that the device interaction function is abnormal. The parameter calculation and update module is used to calculate the adjustment amount of the interactive parameters and update the device interactive parameters when the device itself is confirmed to be abnormal. The feedback learning and strategy optimization module is used to adaptively optimize the system threshold and delay parameters based on the feedback of the interaction effect after parameter updates.

[0030] This embodiment uses a smart pet feeder integrating a weight sensor and an infrared proximity sensor as the core management device, aiming to specifically and completely illustrate the operation flow and data processing logic of the system and method in a complete work cycle. At a preset noon (12:00), the system sends a command to the smart feeder to perform a preset pet interaction action: dispensing a fixed amount of pet food. After the device dispenses 30 grams of pet food, the system immediately starts the first monitoring cycle. During this cycle, the built-in weight sensor and infrared proximity sensor continuously operate, quantifying the pet's eating behavior and attention behavior, respectively. At the end of the cycle, the system standardizes the collected raw physical quantity data and performs a weighted sum according to preset weights, ultimately calculating a comprehensive initial interaction evaluation value V1. Assuming the calculated V1 is 40, this value is a normalized score used to quantify the intensity of the pet's proactive interaction triggered by this interaction action.

[0031] The system then enters the judgment process, comparing the initial interaction evaluation value V1 (40) with the preset expected interaction threshold θ (80) in real time. Since V1 is lower than the threshold θ, the system determines that the initial interaction effect did not meet expectations. To avoid misjudgment due to momentary interference such as the pet's brief distraction, the system does not immediately attribute it to equipment failure, but instead initiates a delayed verification process. The system enters a "pet attention distraction period" with a preset duration T of 5 minutes. During this period, the system suspends sending any new similar interaction commands to the feeder, providing the pet with a calm time window that may allow it to regain attention.

[0032] After the initial distraction period ends, the system initiates a second monitoring cycle, passively monitoring whether the pet actively returns to the device area to interact without new stimuli. The system collects sensor data during this period and calculates a secondary interaction evaluation value V2 using the same algorithm, assuming a value of 45. Based on this, the system performs a preliminary anomaly attribution judgment: comparing the two evaluation values ​​with thresholds. Since both V1 and V2 are below the threshold θ, according to preset rules, the system generates a preliminary judgment that the feeder may have an abnormal interactive function. This step initially rules out the possibility that temporary interference caused poor initial interaction, but the pet resumed interaction after the distraction period.

[0033] Given the initial assessment pointing to an inherent malfunction in the device itself, to enhance the rigor of the conclusion, the system initiated a cross-validation process for verification. This process performed historical interaction data comparison and environmental status record verification in parallel. In the historical data comparison phase, the system queried the historical interaction evaluation values ​​of the device for the same period over the past 7 days, assuming the dataset is [85,82,88,80,78,83,85], and calculated its arithmetic mean to obtain the historical normal level H as 83.0. Next, the system calculated the absolute difference between the initial evaluation value V1 (40) and H, obtaining a deviation D of 43.0. In the environmental verification phase, the system retrieved environmental sensor records within 2 minutes before and after the execution of this action, analyzing whether there were any abnormal events, assuming no abnormalities were recorded during this period. The system made a verification judgment based on the two results: the deviation D (43.0) was greater than the preset fluctuation threshold L (20), and there were no abnormalities in the environmental records. Therefore, the verification result ultimately confirmed that the feeder had an inherent malfunction in its interaction function, effectively eliminating the possibility of misjudgment due to large historical fluctuations or the presence of clear external interference.

[0034] After confirming the equipment malfunction, the system enters a precise parameter adjustment phase. The goal is to correct the equipment's core operating parameters, in this example, the basic feeding amount C (currently 30 grams). The system calculates the interactive parameter adjustment ΔA according to a predetermined formula: ΔA = (θ - V1) × [1 + (θ - V2) / θ + D / H]. Substituting the specific values: θ = 80, V1 = 40, V2 = 45, D = 43.0, H = 83.0. The calculation process is: (θ - V1) = 40, (θ - V2) / θ = 35 / 80 = 0.4375, D / H ≈ 0.5181, therefore ΔA = 40 × [1 + 0.4375 + 0.5181] = 40 × 1.9556 ≈ 78.22. Subsequently, the system calculates the updated interactive parameter value C'=C+ΔA=30+78.22=108.22 grams and immediately sends the new parameter to the smart feeder to complete the device parameter update.

[0035] To scientifically evaluate the effectiveness of parameter adjustments and drive the continuous evolution of the system strategy, the system immediately controls the feeder to perform the same interaction action again after updating the parameters, this time using the new parameter C'. After the action is completed, the system collects data and calculates the updated interaction evaluation value V. f Let's assume its value is 90. The system will use V. f Compared with the current expected interaction threshold θ (80), since V fIf the value is greater than or equal to θ, the parameter update is deemed valid. Based on this "valid" determination, the system performs adaptive optimization. The adjustment target is not the device parameters, but the core strategy parameters of the system: Since the update is valid, it indicates that the pet responds positively to the adjusted device, and its attention may recover faster. Therefore, the system optimizes the pet's attention lapse period parameter T to Tnew = α × T = 4.5 minutes according to the preset shortening coefficient α = 0.9. At the same time, since the current threshold is still applicable, the expected interaction threshold θ remains unchanged at 80 in this loop.

[0036] Through the step-by-step progression of the above process and data-driven decision-making, this embodiment fully demonstrates the complete intelligent closed loop of the management system in response to an interaction anomaly. The system introduces delayed verification to distinguish instantaneous interference, uses cross-validation to verify the root cause of the anomaly, and after confirming that the anomaly is due to the device itself, it uses multi-dimensional data fusion to calculate precise parameter adjustments, and finally optimizes its core decision parameters based on the adjustment effect. This "perception-judgment-execution-learning" closed loop significantly improves the reliability, adaptability, and personalization of management, effectively overcoming the misadjustment and frequency adjustment problems caused by single-response judgments in existing technologies.

[0037] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A multi-sensor pet home management method based on artificial intelligence, characterized in that: Includes the following steps: S1. Control the target pet home appliance to perform preset pet interaction actions and collect initial interaction data; S2. When the initial interaction data is lower than the preset interaction expectation threshold, the delayed verification process is entered, and secondary interaction data is collected after the pet's attention is distracted. S3. Based on the comparison between the initial interaction data and the secondary interaction data relative to the expected interaction threshold, a preliminary anomaly attribution judgment is made. S4. Only when the preliminary anomaly attribution judgment result indicates that the pet home device has an abnormal interactive function, a cross-validation process based on historical interaction data and environmental status records will be initiated for verification. S5. Only when the cross-validation process confirms that the pet home device has an abnormal interactive function, calculate the adjustment amount of the interactive parameters and update the interactive parameter values ​​of the pet home device. S6. Based on the feedback of the interaction effect after the interaction parameter value is updated, the interaction expectation threshold and the pet attention distraction period parameter are adaptively optimized.

2. The multi-sensor pet home management method based on artificial intelligence according to claim 1, characterized in that: S1 further includes the following: Control the target pet home device to perform a complete pet interaction action; After the action is completed, the first monitoring cycle is started, and the interaction-related physical quantity data generated during the cycle are collected by the interactive sensor built into the pet home device. The interaction-related physical quantity data refers to physical signal data generated by the pet's active actions on the device that can be directly measured and quantified by the device's built-in sensors; The collected interaction-related physical quantity data are weighted and summed, and the resulting value is used as the initial interaction evaluation value.

3. The multi-sensor pet home management method based on artificial intelligence according to claim 1, characterized in that: S2 further includes the following: The initial interaction evaluation value is compared with a preset interaction expectation threshold; If the initial interaction evaluation value is lower than the expected interaction threshold, then a preset period of pet attention distraction will begin. During the period when the pet's attention is distracted, the sending of commands to the pet home device to perform similar pet interaction actions is suspended; At the end of the pet's distraction period, a second monitoring cycle is initiated, and interaction-related physical quantity data generated during this cycle are collected in a non-contact monitoring manner using the interactive sensor. The interaction-related physical quantity data collected during the second monitoring period are weighted and fused, and the calculated value is used as the secondary interaction evaluation value.

4. The multi-sensor pet home management method based on artificial intelligence according to claim 1, characterized in that: S3 further includes the following: The initial interaction evaluation value is compared with the interaction expectation threshold to obtain a first comparison result; The secondary interaction evaluation value is compared with the interaction expectation threshold to obtain a second comparison result; Based on the combination of the first comparison result and the second comparison result, a preliminary anomaly attribution judgment is performed; If the first comparison result is lower than the expected interaction threshold and the second comparison result is not lower than the expected interaction threshold, the preliminary attribution judgment result is that the pet's temporary lack of interaction is caused by external interference. If both the first comparison result and the second comparison result are lower than the expected interaction threshold, the preliminary attribution judgment result is that the pet home device has an abnormal interactive function.

5. The multi-sensor pet home management method based on artificial intelligence according to claim 1, characterized in that: S4 further includes the following: When the preliminary anomaly attribution judgment result indicates that the pet home device has an abnormal interactive function, the cross-validation process is initiated. The cross-validation process includes comparison of historical interaction data and verification of environmental status records; The historical interaction data comparison includes the following steps: Query all historical interaction evaluation values ​​of the pet home device within a preset number of historical days that are in the same time period as the current execution time; Calculate the arithmetic mean of all historical interaction evaluation values, and use it as the historical normal level H; Calculate the absolute value of the difference between the initial interaction evaluation value V1 and the historical normal level H to obtain the deviation degree D, D=|V1-H|; The environmental status record verification is achieved by retrieving environmental sensor records for a preset duration before and after the pet interaction action, and analyzing whether there are any events marked as abnormal. The judgment is verified based on whether the deviation D is greater than the preset fluctuation threshold L and whether there are any abnormal events recorded by the environmental sensors. If the deviation D>L and there are no abnormal events recorded by the environmental sensors, the verification result is that the pet home device has an abnormal interactive function. Otherwise, the review result will be that the non-interactive function is abnormal.

6. The multi-sensor pet home management method based on artificial intelligence according to claim 1, characterized in that: S5 further includes the following: When the review result confirms that the pet home device has an abnormal interactive function, the interaction parameter adjustment amount ΔA is calculated. The formula for calculating the interaction parameter adjustment amount ΔA is as follows: ΔA=(θ-V1)×[1+(θ-V2) / θ+D / H]; Where θ is the expected interaction threshold, V1 is the initial interaction evaluation value, V2 is the secondary interaction evaluation value, D is the deviation, and H is the historical normal level; Obtain the current interaction parameter value C of the pet home device, and calculate the updated interaction parameter value C', where C' = C + ΔA.

7. The multi-sensor pet home management method based on artificial intelligence according to claim 1, characterized in that: S6 further includes the following: After the interaction parameter value is updated, control the pet home device to perform the same pet interaction action; Collect and calculate the updated interaction evaluation value V f ; V f Compare with the expected interaction threshold θ, if V f If V ≥ θ, then the parameter update is considered valid. f If the value is less than θ, the parameter update is deemed invalid. Perform adaptive optimization based on the determination result: If the result is deemed valid, the pet attention distraction period parameter T is updated to Tnew, where Tnew = α * T, and α is a preset shortening coefficient, 0 < α < 1. If the interaction expectation threshold θ is determined to be invalid, the interaction expectation threshold θ is updated to θnew, θnew=β*θ, where β is a preset reduction coefficient, 0<β<1.

8. A multi-sensor pet home management system based on artificial intelligence, applied to the multi-sensor pet home management method based on artificial intelligence as described in any one of claims 1-7, characterized in that: It includes interactive control and data acquisition module, delay verification and secondary monitoring module, anomaly attribution analysis module, cross-validation verification module, parameter calculation and update module, and feedback learning and strategy optimization module; The interactive control and data acquisition module is used to control the pet home device to perform pet interactive actions and collect interactive data; The delay verification and secondary monitoring module is used to manage the pet's attention lapse period and collect secondary interaction data when the initial interaction data is lower than the expected interaction threshold. The anomaly attribution analysis module is used to make a preliminary anomaly attribution judgment based on the comparison relationship between the two interaction data and the interaction expectation threshold. The cross-validation review module is used to review the device interaction function based on historical interaction data and environmental status records when it is initially determined that the device interaction function is abnormal. The parameter calculation and update module is used to calculate the adjustment amount of the interaction parameters and update the device interaction parameters when the device itself is confirmed to be abnormal. The feedback learning and strategy optimization module is used to adaptively optimize the system threshold and delay parameters based on the feedback of the interaction effect after parameter updates.

Images